TL;DR

• Mod42 and Duos Technologies Group joined Oracle and Radiant at DCD Connect | Virginia 2026 to examine where modular construction can accelerate AI infrastructure delivery—and where operational and design challenges remain.
• Mod42’s Valerie Crafton emphasized factory-built, integrated infrastructure and phased capacity delivery, while Duos’ Kristen Sanderson shared a practical deployment example from Columbus, Georgia.
• The discussion positioned modular as a construction methodology rather than a temporary container solution, with success dependent on customer requirements, maintainability, supply-chain planning, and repeatability.

# # #

The question was whether modular construction represents a transformative shift or temporary hype, but the discussion at DCD Connect | Virginia 2026 quickly moved beyond the label: where can off-site construction shorten the path to usable capacity, and what must operators get right to realize those benefits?

During the November 4 session, “Debate: Modular construction – Transformative shift or temporary hype?,” Valerie Crafton, Chief Revenue Officer at Mod42, and Kristen Sanderson, Senior Vice President at Duos Technologies Group, brought complementary perspectives on factory-built infrastructure and customer-driven deployment.

Moderated by DCD Editor in Chief Dan Swinhoe, the panel also featured Meloni Boatswain, Director of Data Center Infrastructure Delivery at Oracle, and Manesh Patel, Head of Data Center Development at Radiant. Together, they explored how modular approaches can address construction constraints without overlooking the operational, technical, and economic realities of running a data center.

A Methodology, Not a Container

Both Crafton and Sanderson challenged the assumption that modular data centers are simply shipping containers repurposed for compute, describing purpose-built buildings and integrated infrastructure designed around customer and site requirements.

For Mod42, Crafton described an approach that extends beyond prefabricated electrical or cooling equipment to the broader AI facility, integrating systems in a factory environment and adapting configurations to the technology being deployed. Sanderson similarly emphasized that modular buildings can be designed for regional conditions and incorporated within conventional building shells when customers require additional enclosure or security.

The panel also examined the scheduling advantage of parallel execution: while site construction progresses, electrical rooms and mechanical systems can be assembled off-site rather than waiting for the building to be completed. The distinction matters because modular delivery can change the sequence of construction, not simply the location where individual components are built.

Turning Speed Into Revenue

Crafton connected faster delivery to earlier revenue generation, explaining that customers do not necessarily need to wait for an entire campus to be completed before bringing initial capacity online. Phased deployment, she argued, allows infrastructure to begin serving demand while subsequent capacity is delivered.

Sanderson offered a practical example from Duos’ work in Columbus, Georgia, where the company repurposed a vacant building using modular electrical houses and chillers. She described an approximately 7 MW critical-load deployment at a site capable of scaling to 20 MW, with capacity brought online in phases as the customer prepared its computing equipment.

That example illustrated the different scales and applications represented on the panel: Duos’ discussion focused on customer-specific deployments of roughly 1–20 MW, while Crafton addressed how factory-built infrastructure can support phased delivery at larger campus scale.

Flexibility Requires Deliberate Design

The debate was not an unqualified endorsement of modular construction, with Boatswain questioning how prebuilt infrastructure can accommodate changing customer requirements, including a shift from air-cooled workloads to denser, liquid-cooled systems.

Patel raised related concerns about maintenance access, equipment replacement, and land constraints, emphasizing that a faster construction process must also produce a facility that works operationally. Standardization, in other words, cannot substitute for planning how equipment will be serviced, replaced, or upgraded.

Sanderson explained that Duos addresses these considerations through customer-specific design, redundancy, and replaceable units, while future expansion can incorporate different modules or buildings as requirements evolve. Crafton likewise acknowledged that modular is not the right economic fit for every project, even as she urged developers to evaluate it seriously given the industry’s delivery challenges.

“I think it’s a greater risk not to consider it at this point,” Crafton said.

Workforce and Community Opportunities

Factory-based construction also prompted a discussion about workforce development, with Crafton describing how repeatable tasks and controlled working conditions can support training while reducing exposure to weather-related disruption. Boatswain connected that opportunity to a broader need to introduce students and emerging tradespeople to careers in digital infrastructure.

For Sanderson, the Columbus project demonstrated a community dimension as well: she described using local labor, working with community leadership, and hiring interns as Duos develops its infrastructure footprint.

Crafton highlighted the potential to reuse industrial buildings, while Sanderson cautioned that available power alone is not enough—appropriate network connectivity must also be part of the site equation.

Repeatability Without One-Size-Fits-All

Panelists were equally clear that modular construction does not eliminate supply-chain constraints, with long-lead equipment and smaller components still capable of delaying delivery. Their discussion emphasized forecasting, supplier coordination, and repeatable designs as ways to manage—not erase—those risks.

Patel stressed the value of repeating familiar delivery and commissioning processes across projects, while Crafton encouraged engineering choices that respond to equipment availability and customer needs rather than defaulting to conventional designs.

Sanderson’s closing perspective captured the customer-first discipline behind those decisions: “It’s not what we want. It’s what we’re building for them.”

Learn more about Mod42 and Duos Technologies Group.

The post Beyond the Hype: Mod42 and Duos Explore Modular Data Center Delivery at DCD Connect Virginia appeared first on Data Center POST.

TL;DR

• Aureon has introduced Unified Protection, a three-tier offering that combines managed IT and cybersecurity services.
• The offering is designed to help businesses strengthen security and simplify the management of their IT environments.
• Unified Protection brings together technology management and cybersecurity support to address evolving business security needs.

# # #

Cybersecurity remains a critical priority for organizations managing connected systems, sensitive information, and essential business operations. With threats continuing to evolve, businesses are evaluating how their IT management and security strategies work together to protect their environments.

Aureon has introduced Unified Protection, a new three-tier offering that combines managed IT and cybersecurity services to help businesses strengthen their security posture and simplify technology management. The offering provides a structured approach designed to address different organizational needs.

Managing cybersecurity alongside everyday IT operations can present challenges, particularly when responsibilities are spread across multiple systems, service providers, and internal teams. Bringing these services together can help organizations improve coordination, maintain visibility into their technology environments, and address potential security concerns.

“IT and cybersecurity can’t be treated as separate parts of the business anymore,” said Joe Johnson, Vice President of Product Development at Aureon. “Unified Protection brings them together in a way that gives organizations stronger visibility across their environment and the ability to actively respond when threats emerge. With three levels of protection, businesses can choose the coverage they need today and strengthen that protection as their needs evolve.”

Cloud applications, interconnected networks, and digital services are central to daily business operations, making cybersecurity and business continuity important considerations across the digital infrastructure ecosystem.

The introduction of Unified Protection expands Aureon’s managed IT and cybersecurity services, providing organizations with a more coordinated approach to technology management and security.

To learn more about Aureon’s Unified Protection offering, read the full announcement here.

The post Aureon Introduces Unified Protection to Strengthen Business Cybersecurity appeared first on Data Center POST.

TL:DR

• Customized egress solutions address unpredictable cloud data transfer fees by aligning outbound traffic, IP requirements, and hosting capacity with application needs.
• MOD Mission Critical combines IP Transit, IPv4 procurement, managed colocation, and carrier multi-homing to deliver wholesale pricing on a unified bill.
• The MOD PaaSPort platform simplifies global sourcing for colocation and connectivity, allowing egress deployments to adapt seamlessly as enterprise workloads grow.

# # #

Written by iMiller Public Relations on behalf of MOD Mission Critical.

Managing outbound cloud traffic and egress costs has become a critical focus for enterprises operating across public clouds, private infrastructure, and multiple locations. Outbound data transfer charges and inter-region fees frequently lead to unpredictable monthly expenses as application workloads expand. To resolve these operational and financial complexities, tailored egress planning must account for specific application destinations, source-IP requirements, traffic locations, and scalable network capacity.

MOD Mission Critical provides customized egress infrastructure designed to complement public cloud setups by combining specialized IP options with global hosting and connectivity. Their portfolio incorporates residential, business-class, and mobile IPs alongside managed colocation and IPv4 address procurement. Supported by over 80 Points of Presence worldwide in leading data center markets, this footprint allows organizations to route outbound traffic through specific geographic points while adapting to evolving infrastructure demands.

Through services like IP Transit offering Gig-E and 10G ports across dozens of carriers, businesses can access wholesale pricing and multi-homing options on a single bill. Additionally, the MOD PaaSPort platform streamlines the exploration and sourcing of global connectivity, colocation, and managed network capabilities. By integrating these services into a cohesive egress strategy, enterprises can simplify infrastructure sourcing, maintain reliable connectivity, and better control outbound traffic costs

To read the original article, please click here.

The post Customized Egress for Hybrid Cloud: Built Around Your Business appeared first on Data Center POST.

TL;DR

• The telecommunications industry generates massive volumes of operational and commercial transaction data that remain underutilized for financing connectivity services.
• By using verifiable transaction records and debtor creditworthiness, receivable financing can rely on actual payment obligations rather than a supplier’s balance sheet constraints.
• Connecting transaction validation, legal enforceability, continuous monitoring, and automated payment controls creates institutional-grade financeable assets.
• Securely linking operational network data to institutional capital improves working capital liquidity for suppliers and unlocks new revenue models for digital infrastructure platforms.

# # #

By Lawrence Gilioli, Founder & CEO, Lenderwize 

The Overlooked Financial Value Of Connectivity Data

The telecommunications industry has spent decades developing sophisticated systems to measure, authenticate, bill and reconcile the exchange of information. Every day, telecommunications carriers and connectivity providers generate enormous volumes of operational and commercial data supporting the world’s digital infrastructure.

Yet there is a striking contradiction.

The industry has become extraordinarily efficient at measuring the delivery of connectivity services, but the financing of those same services still frequently relies on conventional invoices, periodic reporting and manual verification. What if one of the telecommunications industry’s most valuable untapped assets is not its network infrastructure, but the commercial transaction data that infrastructure generates?

The Information Already Exists

Consider a wholesale connectivity provider supplying services to an international telecommunications operator.  The underlying commercial relationship generates multiple sources of information: service records, traffic measurements, contractual rates, billing data, invoices, reconciliation results and payment histories.  These records serve operational and commercial purposes. They help establish which services were delivered, what amounts were billed and whether discrepancies exist.

However, that same information can potentially serve another purpose. It can provide evidence supporting the existence, value and performance of receivables that financial institutions may be willing to finance.

This distinction matters.

A lender financing a traditional corporate borrower primarily assesses the company’s overall capacity to repay.

A financier purchasing a receivable also needs to establish whether a specific payment obligation exists, whether it is enforceable, whether the debtor is creditworthy and whether the resulting cash can be collected.

Reliable transaction information can help answer some of these questions. It cannot eliminate credit risk, but it can reduce certain forms of informational uncertainty.

Why The Distinction Matters Economically

Imagine two connectivity providers delivering comparable services to the same investment-grade telecommunications operator. One provider is a large, established corporation with extensive banking relationships. The other is a smaller supplier with limited access to conventional working-capital facilities. Although both may hold receivables against the same customer, their financing costs and access to liquidity can differ substantially.

Some of that difference is justified by their respective financial positions. But an important question remains: If the payment obligation is owed by the same creditworthy debtor, and the underlying receivable can be independently verified, how much of the financing decision should depend on the supplier’s own balance sheet?

This is not a new concept in receivables finance. What is changing is the technology available to implement it at scale.

The opportunity is to connect transaction verification, debtor credit assessment, payment control and financing decisions within a coordinated digital process.

From Operational Data To Financeable Assets

Transforming commercial transaction data into an asset that institutional investors can finance requires more than an attractive dashboard. It requires a chain of independently supportable evidence and enforceable rights.

• Transaction evidence: Information supporting the delivery and commercial value of services.
• Receivable validation: Confirmation that the amount is owed, subject to contractual terms and potential disputes.
• Credit assessment: Evaluation of the debtor’s capacity and willingness to pay.
• Legal enforceability: Appropriate contractual rights and, where required, assignment or acknowledgment arrangements.
• Payment control: Mechanisms ensuring that collections follow agreed settlement arrangements.
• Continuous monitoring: Visibility over payments, exceptions, disputes and outstanding exposures.

The combination is potentially more valuable than any individual component.

A verified invoice without reliable collection controls may still present significant risks. A controlled payment account without reliable transaction evidence may offer limited protection. And excellent transaction data cannot compensate for an insolvent debtor.

The opportunity comes from integrating the different components into a coherent financial infrastructure.

The Implications For Data Centers And Digital Infrastructure

Why should the data center industry care?

Because the growth of digital infrastructure depends on an interconnected ecosystem of telecommunications operators, connectivity providers, cloud platforms, equipment suppliers and specialized service companies.

These businesses do not all share the same financial characteristics. A major infrastructure operator may have considerable access to institutional capital, while smaller connectivity and technology suppliers may experience tighter liquidity constraints.

As digital infrastructure becomes more interconnected, commercial relationships can become more complex. Financial processes need to evolve accordingly. There may be opportunities to finance eligible receivables arising from connectivity and infrastructure services through platforms that integrate operational records, commercial validation and payment controls.

Not every infrastructure contract will be suitable. Long-term construction commitments, contingent obligations and disputed payments can have very different risk characteristics from established, undisputed trade receivables.

But where commercial obligations are sufficiently clear and verifiable, financing may become more efficient.

Who Will Capture The Value?

This leads to a second question that may be even more consequential.

If commercial transaction data becomes a gateway to financing, who will control that gateway?

Will it be the banks providing capital? The infrastructure operators generating the data? The technology platforms managing commercial transactions? Or the independent financial infrastructure providers connecting them?

The answer may reshape commercial relationships.

Platforms that can combine verified transactional information with financing and payment services may be able to generate new revenues without lending directly from their own balance sheets.

Infrastructure businesses may obtain improved access to working capital. Financial institutions may gain access to assets accompanied by better verification and monitoring. And customers may benefit from financial services embedded within the platforms they already use.

However, these benefits are not automatic. Data access, confidentiality, interoperability, regulatory obligations and contractual enforceability all need to be addressed.

The Real Opportunity Is Not Simply Faster Financing

The underlying proposition is broader than accelerating payments. It is about recognizing that the operational systems supporting digital infrastructure can also support a more transparent and potentially more efficient financial ecosystem. The telecommunications industry already possesses extensive information about the commercial activity flowing through its networks.

The opportunity is to connect that information securely and reliably to institutional capital. The next generation of digital infrastructure may therefore create value in two ways: by transporting information and by enabling the financing of the commercial activity that information represents.

The question is no longer simply how efficiently we can move data. It is how much economic value we can unlock from the transactions that data makes possible.

# # #

About the Author

Lawrence Gilioli is Founder and CEO of Lenderwize, a financial technology company developing infrastructure for embedded receivables financing, programmable payment controls and transaction-level financial services.

With an extensive background in telecommunications and technology entrepreneurship, he focuses on connecting verified commercial transactions with institutional capital to improve liquidity, transparency and efficiency across business-to-business ecosystems.

The post The Telecom Industry Is Sitting on a Financing Asset It Doesn’t Know How to Monetize appeared first on Data Center POST.

TL;DR

• Global data center electricity use is projected to grow 26% in 2026 to 565 terawatt-hours, driven by AI servers that encounter memory supply and power grid limits.
• AI inference servers average roughly 20% utilization, leaving four-fifths of paid hardware capacity and floor space idle while continuing to draw power.
• Decentralized compute networks pool idle GPUs across facilities into an open inference market, directing contributed cycles straight to active AI workloads with minimal overhead.
• Selling unused accelerator hours maximizes existing infrastructure efficiency, serving more AI tokens without needing to build, cool, or plug in additional hardware.

# # #

Global data center electricity use will grow 26% in 2026, to 565 terawatt-hours, according to Gartner. Most of that growth will come from AI-optimized servers. New clusters also run into limits on memory supply and grid access.

Hyperscalers want more control over hardware costs, and several are building their own chips to get it. Microsoft plans to begin deploying its Maia 300 accelerator in 2027 and claims it will deliver 40% more AI tokens per megawatt than leading GPU-based systems.

Having built a decentralized AI compute network, I think the industry underuses a cheaper lever, which is getting more productive work out of the GPUs it has already racked, powered, and paid for.

Installed Capacity and Delivered Compute Are Drifting Apart

Training clusters run close to capacity, and Lawrence Berkeley National Laboratory’s latest U.S. modeling assumes they average about 80% utilization. Inference fleets are sized for peak traffic and spend long stretches of the day underused.

The same report assumes AI inference servers will reach about 20% average utilization by 2030, half the figure its previous edition used, and that idle machines still draw a fifth of their rated power. It also estimates that grid interconnections run at around half their rated capacity because of redundancy requirements and maintenance.

At 20% average utilization, four-fifths of what an inference server costs in depreciation and floor space pays for capacity that does no work.

Hardware Design Raises the Price of Every Idle Hour

In large-model inference, the amount and speed of high-bandwidth memory (HBM) on each card often caps throughput before raw compute does; the IEA expects the current HBM shortage to last through at least the end of 2027. Models split across many GPUs slow down sharply once the interconnects between them saturate.

Rack-scale systems respond with denser packaging. The IEA estimates that AI server power density rose elevenfold between 2020 and 2025 and will quadruple again by 2027.

At these densities, liquid cooling has become a requirement, because each AI rack has to shed as much heat as 30 natural gas boilers put out. Liquid-cooled AI halls helped bring the U.S. average power usage effectiveness (PUE) down from 1.55 to 1.45 between 2018 and 2024, according to the Berkeley Lab report. All of this adds to the capital cost a GPU has to earn back during the hours it actually works.

Spare GPU Cycles Can Find Buyers Across a Wider Network

A decentralized compute network pools GPUs from data centers, cloud providers, and independent operators into one inference market. Requests from anywhere can go to whichever accelerators have spare capacity, so a GPU that one facility can’t keep busy still earns.

When we designed our network’s protocol, the goal was to send nearly all contributed compute straight to AI workloads and keep consensus overhead to a minimum. Hosts prove their capacity in a short, transformer-based compute sprint, and outside that window their hardware serves inference. Other participants re-check only a small random sample of the results.

Data center operators can earn on capacity they built for peak demand. Owners of older A100 and H100 fleets can keep those cards earning next to newer ones, because a node can mix NVIDIA GPUs anywhere from the A100 to the B300 as long as it pools at least 320 GB of memory.

Inference is the workload that suits open networks today. A shared pool can absorb demand spikes and narrow the gap between peak provisioning and average load. Frontier-scale training is another matter, since it still needs ultra-fast fabrics inside a single facility.

Useful Work Belongs in Every Capacity Plan

Power is going to stay tight for the rest of the decade, and custom chips will keep squeezing a bit more work out of every watt. Operators don’t have to wait on either one, though, because idle accelerator hours are inventory they can start selling today. Every token served on a GPU that would otherwise have sat dark is compute nobody had to build, cool, or plug into an already stretched grid. My bet is that utilization ends up right next to megawatts and PUE on every AI infrastructure scorecard, and open networks will be one of the things nudging that number up.

# # #

About the Author

David Liberman is a serial founder and operator who had spent two decades building across AI, media, fintech, and augmented reality. He is a co-creator of Gonka, a decentralized network for high-efficiency AI compute that directs nearly 100% of GPU cycles toward productive AI inference and training rather than the wasted computation of conventional blockchains.

The post Decentralized AI Compute: How to Get More Out of Every GPU Hour appeared first on Data Center POST.

TL;DR

• 365 Data Centers team members joined a Surfrider Foundation beach cleanup in New Jersey and logged the types and amounts of trash they found. 
• That data helps identify pollution trends and supports efforts to cut single-use plastics.
• The same measure-first approach applies to infrastructure sustainability.

# # #

Recently, 365 Data Centers team members spent a sunny day cleaning a New Jersey beach with the Surfrider Foundation. As part of the cleanup efforts, the team also recorded the type and quantity of each item collected. 

This tally is a key part of Surfrider’s cleanup program. It helps identify common sources of pollution and supports advocacy for reducing single-use plastics and keeping trash from reaching beaches in the first place.

The data adds up. Surfrider’s 2026 cleanup reports, based on 630 report forms so far, show about 99,400 pounds of trash and recycling removed. In 2025, the total was more than 338,000 pounds of trash collected.

“The event went really well, and several team members enjoyed spending the day outside at the beach, which beats sitting at your desk all day,” said Fiona Rada, Strategic Director of Sales Operations at 365 Data Centers, following the event. “We were fortunate to have great weather, and we were the only company participating in the Surfrider Foundation that day.”

The Surfrider Foundation is a nonprofit organization dedicated to protecting the world’s ocean, waves, and beaches and ensuring everyone can enjoy them. Through a grassroots network of volunteers and experts, it tackles plastic pollution, monitors water quality, organizes beach cleanups, and advocates for coastal protection and equitable beach access.

Volunteerism is part of 365’s culture, with team members across the company’s markets seeking ways to get involved in supporting their local communities. For more information on 365 Data Centers’ locations, visit 365datacenters.com.

The post 365 Data Centers Joins Surfrider Foundation Beach Cleanup appeared first on Data Center POST.

TL;DR

• Organization leaders face steep cloud licensing increases following Broadcom’s acquisition of VMware.
• Accepting price hikes or delaying decisions leaves companies without control over their budgets or cloud platforms.
• Small to mid-sized organizations with modest workloads can execute a platform migration in months rather than years with proper planning.

# # #

Cloud licensing costs have come up in almost every customer conversation I’ve had this year, often starting with concerns about a high renewal price, and moving quickly to a discussion about “what are our options?”

It’s a challenge many organisations are facing since Broadcom’s acquisition of VMware, and frankly, it’s been hard to avoid the headlines of AT&T’s 1,050% renewal increase, and Tesco’s decision to transition 40,000 workloads from VMware and sue Broadcom.

While there’s no doubt that large enterprises are struggling with increased licensing fees, most of the organisations I speak with don’t have the scale to negotiate or litigate. For them, they are forced to simply accept unwanted changes to per-core subscriptions, bundled products, and an annual renewal that comes with an increasingly hefty price tag. To this end, it’s often easier to do nothing, and accept slightly worse terms.

What I tell IT leaders is that a decision to stay put is a valid choice to make, but it’s not one that will put them in control of their budget or how they use the platform. Instead, it’s kicking the can down the road and hoping that next year’s quote isn’t so high.

Now, I appreciate leaving a platform like VMware isn’t easy, in fact, it’s difficult and takes time. The good news for IT leaders at small or mid-size organisatiofns, is that unlike Tesco who has to migrate 40,0000 workloads, their estate might only cover 400, meaning that with the right planning a migration can happen in months rather than years.

It’s planning well that I believe is fundamental to success. I always advise customers to start with a detailed discovery phase, to develop a clear understanding of what’s running where and what each workload depends on.

Once this is done, the migration should start small, for example, with a greenfield project on a second platform. This way, you can keep your most sensitive workloads where they are until the team feels confident in the new environment, avoiding unnecessary downtime or risk.

Cloud platform migration can be challenging, but it can also be the difference between being in control of your budget and data in a way that is not possible if you choose to stay put.

# # #

About the Author

​​Juan Aguirre is Ilkari’s Chief Commercial Officer, responsible for driving commercial strategy, partner ecosystems and customer success. With more than 25 years’ experience across Internet Protocol Television (IPTV), enterprise networking and SaaS, Juan has a strong track record of aligning technology offerings with evolving customer needs.

Before joining Ilkari, he served as EMEA Director for Partner Sales at CommScope, overseeing EMEA distribution and vertical markets such as hospitality and multi-dwelling unit (MDU). At Ilkari, Juan shapes the company’s go-to-market strategy, ensuring clients gain competitive advantage through tailored, forward-thinking solutions. His commercial leadership is central to Ilkari’s growth and its commitment to customer-centric innovation.

The post The Cost Of Kicking Your Cloud Renewal Down The Road appeared first on Data Center POST.

TL;DR

• Enterprise AI procurement should move beyond headline model capabilities to evaluate how the underlying architecture aligns with specific operational workloads.
• Security teams need to trace prompt and document data paths, inspect subcontractor access, and evaluate on-premises or private cloud placement to ensure regulatory compliance.
• Organizations must confirm how vendors manage model updates, pricing adjustments, and behavioral shifts, ensuring teams can retain or retest validated model versions.
• Buyers should secure data export capabilities, prompt portability, and model substitution terms prior to purchase to avoid vendor lock-in as technology evolves.

# # #

By Robert Johnston, Chief Innovation Officer, N-able

Enterprise AI buying decisions often begin with a familiar set of questions. How capable is the model? How fast is it? What will it cost? Those details matter, but they do not tell an IT or security leader how the product will fit into the organization’s environment.

Before comparing products, define the workload in plain terms. What recurring problem will the tool solve, which data will it need, and what volume should the team expect in production? An AI feature may send data to a third-party provider, rely on a model the customer cannot replace or become far more expensive at real world volume. Leaders can uncover those risks early by asking vendors five practical questions before onboarding a new tool.

What model is actually running underneath the product?

“AI-powered” is a product description, not an architecture. Buyers should know which model or models perform the work, whether they are proprietary or open weight, and why the vendor selected them. They should also ask whether the product uses the same model for every task.

Many enterprise workflows do not require frontier-level reasoning from beginning to end. Understanding which models are used for which tasks helps buyers determine whether the product’s architecture actually matches the workload. Classification, extraction and routing may work well with smaller models, while rules-based software can handle other steps without a model at all. A vendor that has broken the workflow into parts should be able to explain where high end capability is necessary and where it is not. That answer gives buyers a clearer view of cost, performance, resilience, and dependency risk.

Where does our data go?

Ask the vendor to trace the data path during inference and after the response is returned. Where is information processed? Is it stored, logged or used to improve a model? Which subcontractors or model providers can access it, and in what jurisdictions? Existing security controls do not automatically extend to an AI feature simply because it sits inside software the organization already uses.

This review should cover prompts, uploaded documents, retrieved records and model outputs. Source code, service tickets, financial documents and customer records may all contain sensitive information. Security teams need enough detail to decide whether the proposed data flow fits their regulatory obligations, customer agreements and internal policies.

Can sensitive workloads stay inside our environment?

Some workloads are well suited to a vendor-hosted service. Others may need to run in the customer’s cloud tenant, on premises or close to the application that uses them. The right placement depends on the sensitivity of the data, the performance required and the controls the organization must enforce.

Self-hosting doesn’t always mean building a large GPU cluster. Smaller open weight models can run on existing hardware for some tasks, and larger deployments can operate in private cloud or data center environments. Buyers should ask what deployment options the vendor supports, what hardware each option requires and which product features change outside the hosted version. This makes it possible to weigh added control against the cost and operational work of running the model internally.

What happens if the underlying model changes?

AI products can change even when the application interface looks the same. A provider may update a model, retire it or adjust its pricing and terms. Those changes can affect output quality, latency, cost, compliance, and even application behavior. Ask how the vendor evaluates model updates, communicates material changes and tests whether existing workflows still perform as expected.

It is also worth asking whether customers can delay an update or keep a validated model version for a defined period. Organizations in regulated or high-risk settings may need time to retest a workflow before a new model reaches production. The vendor’s answer will show how much control the customer retains after purchase.

What is our exit path?

No organization can predict which model or provider will be the best fit two years from now. New models will emerge, smaller ones will improve and economics will shift. A tool that tightly couples the workflow to one provider can make an otherwise routine change expensive and disruptive.

Before signing, ask whether the model can be substituted without rebuilding the application. Confirm how the organization can export its data, prompts, configurations and evaluation results. Review any license terms that restrict commercial use, redistribution or deployment at scale, especially when a product relies on a model described as open.

A credible vendor should be able to explain the exit path as clearly as the onboarding plan so the buyer understands any dependencies and can accept them deliberately.

Match the architecture to the workload

These questions give procurement, IT and security teams a common way to evaluate AI products. They shift the conversation from a model’s headline capabilities to the practical details that determine whether a tool will remain secure, affordable and useful in production. Successful AI adoption depends as much on understanding dependencies, deployment requirements, and operational ownership as it does on model performance.

The best answer will always vary by workload. What matters is that leaders understand the choice they are making, including where the work runs, how data moves and how easily the organization can change course later if needed.

# # #

About the Author

Robert Johnston is Chief Innovation Officer at N-able, where he leads strategic innovation initiatives across the company’s security services and product portfolio. A seasoned cybersecurity leader and entrepreneur, Robert previously founded and served as CEO of Adlumin, a security operations platform acquired by N-able in 2024. Prior to Adlumin, he served as a principal consultant at CrowdStrike and held cybersecurity leadership roles in the U.S. Marine Corps. He brings extensive expertise in cybersecurity, threat intelligence and emerging technologies, with a focus on driving practical innovation to address evolving security challenges.

The post Five Questions to Ask Before Onboarding an AI Tool appeared first on Data Center POST.

TL;DR

• Confirm exact liquid cooling specifications, hardware compatibility, and responsibility boundaries before signing a lease, rather than relying on general building claims.
• Require a written handover package containing verified commissioning and safe testing results before connecting production hardware.
• Establish clear operating arrangements for routine server replacements, fluid additions, and maintenance checks to prevent ownership disputes.
• Agree on shared monitoring data, joint incident escalation procedures, and engineering review steps before expanding deployment density

# # #

By Brenda Pak, co-founder and CEO, Reliability Engine

The lease is almost agreed, the servers are on order, and the site is described as AI-ready. Before anyone signs, there is one more conversation worth having: what, exactly, will be ready for these racks on move-in day?

A tenant buying liquid-cooled capacity needs more than a statement that the building supports liquid cooling. The facility and the proposed hardware have to work together, and both sides need to know where their responsibilities begin. An answer that sounds clear in a sales meeting can become surprisingly vague when a technician asks who is allowed to connect a hose.

Resolve those details before committing to a move-in date. Five questions can help the tenant and operator agree on what the space must deliver.

What is available to our racks on day one?

Ask the operator to describe the proposed deployment, not the building’s eventual capacity. Which cooling equipment and connections will serve the contracted space at handover? What depends on a later expansion, additional equipment or a future commissioning test?

Then bring the hardware requirements into the conversation. A rack power figure alone does not describe its cooling needs. The design team should reconcile the required liquid flow, temperature and pressure conditions, approved coolant and connections. It should also account for heat that the chosen servers still reject to the room.

Tenants do not need to perform the engineering themselves. They do need a written record of who has checked compatibility and what remains unresolved. Otherwise, two competent teams can arrive at handover with different assumptions about the same installation.

What evidence will we receive before connecting?

Request a handover package that relates to the equipment and operating arrangement being delivered. It should identify the tests completed, the conditions used and any limitations or open items. A test elsewhere in the building may be informative without proving readiness at the tenant’s connection.

Agree in advance how the teams will demonstrate normal operation and the relevant maintenance or failure conditions. What happens while a pump is unavailable? Which party observes the response, and what result counts as acceptable? Testing should follow an agreed, safe commissioning plan rather than being improvised with production hardware.

I have written about the physical checks needed before connecting the first GPU. For a tenant, the commercial implication is simple: decide which evidence must exist before accepting the service, and who can accept any outstanding work.

Who checks the cooling after maintenance?

The initial installation is only the starting configuration. Servers will be replaced, filters serviced and coolant added. Some of that work may be carried out by the operator, some by the tenant and some by a hardware service provider.

Walk through one ordinary job together: a server replacement that requires opening a liquid connection. Who authorizes it? Who confirms the replacement is compatible? Who records the work and checks the affected branch before it returns to service?

Put those answers in the operating arrangements. The person who owns the server may not own the cooling equipment, and the person doing the work may own neither. A shared maintenance calendar and an agreed return-to-service check can prevent a routine task from becoming a dispute about who was supposed to verify the result.

This discussion should also cover fluid additions and substitutions. Approval to purchase a container is not automatically approval to put its contents into a particular loop.

What will we be able to see during a problem?

An incident is a poor time to discover that useful cooling data are unavailable to the tenant or retained for too short a period.

Ask which measurements the operator can share for the affected equipment, how often they are recorded and how they will be made available during an investigation. Depending on the design, useful evidence may include rack or branch flow, supply and return temperatures, pressure measurements, control states and maintenance events. The data need clear locations and timestamps to be interpreted together.

Where a measurement is unavailable, agree how the team would investigate without it. A building-level summary is not necessarily enough to explain a problem affecting one rack. Equally, unrestricted access to the operator’s control system is not a prerequisite for getting useful evidence. The teams can agree a secure reporting and escalation process.

Name the people who will coordinate a joint investigation. A service desk ticket should have somewhere to go when the hardware team needs help from facilities.

What happens when the deployment grows?

Ask what must be reviewed before adding the next generation of servers or increasing the contracted density. More electrical capacity does not automatically establish that the installed liquid path and room cooling can support a different hardware configuration.

The agreement should distinguish capacity available now from capacity that requires an engineering review or further work. It should also make the approval process understandable: what the tenant supplies, who assesses it and how any testing affects the schedule.

Procurement can then build that review into the expansion schedule, before new hardware arrives with cooling requirements the installed system has not been assessed against.

Before signing, ask the operator and your own team to explain the same deployment back to you. They should agree on what will be delivered, the evidence for accepting it and who takes charge when it changes. If their answers differ, that conversation has already earned its place in the schedule.

# # #

About the author

Brenda Pak is co-founder and CEO of Reliability Engine. She writes about the operational and commercial decisions behind liquid-cooled AI deployments, including commissioning, ownership and ongoing maintenance.

The post The Cooling Questions AI Tenants Should Ask Before Signing a Colocation Lease appeared first on Data Center POST.

TL;DR

• Alexander Marcus, CEO and president of ESI Total Fuel Management, will speak at the free online Kingdom Impact Summit, Nov. 16-20.
• IBAM, led by founder and president Steve Adams, hosts the summit. It brings together more than 30 leaders from business, ministry, missions, generosity, and sports.
• Marcus’ Nov. 20 session is “From Success to Significance: Trusting God in Business and Mission.”
• He draws on more than 40 years in engineering and critical power. His message is that business works best when it is stewardship: caring for people, resources, and communities.
• Registration is free, and anyone can attend from anywhere.

# # #

Ask anyone who has spent a career keeping critical systems running, and you will hear the same lesson. The work that matters most is rarely noticed, and it depends on people who take responsibility for it. For Alexander Marcus, that lesson reaches well beyond the generator yard.

After more than 40 years in engineering and critical power, Marcus will share it from a different stage. As CEO and president of ESI Total Fuel Management, he will be a featured speaker at the Kingdom Impact Summit 2026, a free online event running Nov. 16-20.

A Summit Built Around a Village

IBAM hosts the summit, and its founder and president, Steve Adams, leads it. The theme is “One Village Can Spark a Global Kingdom Movement.” More than 30 leaders from business, ministry, missions, generosity, and sports will take part. Each day has its own focus, from seeing everyday work as part of a calling to using success to serve others. Speakers include Jordan Raynor, Mark Batterson, Peter Greer, and Dr. Ché Ahn.

IBAM is a faith-driven organization that equips entrepreneurs in underserved communities. Its Give, Teach, Equip model combines startup support, practical business training, and discipleship. Excellence in Giving recognizes IBAM as a Certified Transparent organization for 2026.

From Success to Significance

Marcus will speak on Nov. 20, the final day. His session is titled “From Success to Significance: Trusting God in Business and Mission.” It will explore how leaders can lean on faith, practice stewardship, and shift from relying on themselves to building lasting Kingdom impact.

“After more than 40 years in engineering and critical power,” said Marcus, “I’ve seen that business works best when it is a form of stewardship, caring for people, resources, and communities. I hope attendees leave with one faithful next step they can take into the workplace they’re already doing.”

That perspective fits the mission behind ESI’s work. ESI serves the global data center community by increasing generator uptime, maintaining environmental compliance, and responding to fuel-related threats. Its scope covers the design, development, manufacture, installation, and service of fuel quality management systems. It also covers products across the fuel lifecycle, including consumption monitoring and emissions reduction systems for mission-critical facilities. Dependable work like this serves the people and communities who rely on those facilities.

How to Join

New sessions are released each day, and each day’s sessions are free to watch for 24 hours after release. Closed captions are included. Attendees who want to keep the sessions can add an optional All-Access Pass after registering. Registration is free at summit2.ibam.org/sessions.

To learn more about ESI Total Fuel Management, visit fuelmanagement.com.

The post Where Success Meets Significance: ESI’s Alexander Marcus Joins the Kingdom Impact Summit appeared first on Data Center POST.

TL;DR

• Increasing electrification and growing demand for advanced computing across the Northeast are creating severe electrical infrastructure limits and long utility interconnection queues.
• Combining modular facilities with hybrid microgrids, solar generation, and battery storage reduces initial reliance on single large utility interconnections while supporting regional sustainability goals.
• Adopting scalable modular infrastructure helps retain research, life sciences, and artificial intelligence workloads within the region while managing utility and community impacts.

# # #

Power constraints are becoming a significant planning factor for advanced computing across the Northeast. This article examines whether modular data centers paired with clean-energy microgrids could support phased deployment, reduce grid pressure, and align new capacity with community and climate priorities.

Across the Northeastern United States, demand for advanced computing capacity is rising as the region’s electrical system adapts to electrification and emerging large loads. ISO New England’s 2024-2033 forecast projected net annual energy use of 140,001 GWh in 2033 and identified more than 23,000 GWh of annual consumption from electrification.¹

This growth is occurring alongside a large and changing interconnection queue. ISO New England cautions that queue totals require careful filtering because projects can withdraw, become operational, or connect outside New England; the public queue is therefore best treated as a current project pipeline rather than a fixed measure of future supply.² For organizations requiring high-performance computing, such as biotech, AI, and research institutions, interconnection uncertainty can complicate infrastructure planning.

Traditional data center development depends on large utility interconnections, extensive permitting cycles, and community acceptance of sizable permanent facilities. In constrained regions, those requirements can extend development timelines as compute-intensive workloads grow. Recent public evidence shows the scale can be substantial: a 2026 Meta-authored technical paper describes a 150 MW data center hosting an 83,000-GPU cluster.³

A modular, clean-energy-integrated approach is one alternative to conventional large-campus development. In this model, capacity is deployed in smaller blocks (for example, 5-25 MW) rather than as a single large build-out; the appropriate block size depends on workload, site, interconnection, cooling, and economic requirements. Hybrid microgrids may incorporate clean-energy generation, battery storage, and energy-management systems. Phasing capacity and combining on-site resources with grid supply may reduce initial dependence on a single large interconnection, although timing and performance remain project-specific.

Why Modularity Matters

Modular data center blocks allow infrastructure to scale in phases rather than through a single large build-out. This approach may reduce initial demands on utilities and let developers align capacity additions with actual demand. It can also give communities a clearer view of each phase’s footprint and impacts, although acceptance will still depend on siting, design, and local priorities.

From an engineering perspective, modularity can support modern compute requirements. AI workloads may require high-density racks, liquid cooling, and adaptable power distribution. Uptime Institute’s 2024 survey found average rack densities remained below 8 kW and that most facilities had no racks above 30 kW; its subsequent analysis described AI systems ranging from under 50 kW per rack to 132 kW for an NVL72 configuration.⁴ Modular facilities can be designed around the selected density and cooling architecture while allowing capacity to be added in phases.

The Role of Clean‑Energy Integration

Hybrid microgrids are one option within this development model. Clean-energy generation, such as solar, fuel cells, or other distributed resources, can supply part of a facility’s demand, while battery storage and controls can shift loads, manage peaks, and support continuity during grid disturbances. The U.S. Department of Energy identifies autonomous operation during grid outages as the primary resilience benefit of microgrids and notes that additional environmental and grid benefits depend on system components and configuration.⁵

Clean-energy integration also aligns with regional sustainability goals. Massachusetts’ 2050 Decarbonization Roadmap outlines aggressive emissions-reduction targets, and communities are increasingly attentive to the environmental impact of large infrastructure projects. Depending on the technologies selected, facility design, and operating profile, clean-energy resources can help reduce emissions, noise, and land-use impacts compared with traditional campuses.

Implications for the Northeast’s Innovation Economy

The Northeast’s economic competitiveness depends heavily on access to reliable, scalable compute. Research institutions, life-science organizations, and emerging AI companies rely on high-performance infrastructure to conduct modeling, run simulations, and process large datasets. If regional capacity cannot keep pace with demand, these organizations may face slower expansion or stronger incentives to place new workloads in states with greater power availability.

Modular, clean-energy-integrated data centers could provide one pathway for expanding digital infrastructure while managing utility constraints and project timing. Their usefulness will depend on local grid conditions, technology choices, economics, permitting, and community priorities.

A Framework for Policymakers and Industry Leaders

As power constraints shape infrastructure planning, policymakers, utilities, and industry leaders are evaluating alternative development models. Clean-energy-integrated modular data centers warrant consideration alongside conventional campuses, grid upgrades, demand flexibility, and other options for supporting advanced computing growth.

Key considerations for stakeholders include:

• Grid planning: Hybrid microgrids may help manage peak demand and forecasting when their operating profiles are coordinated with utility needs.
• Sustainability: Outcomes depend on the generation mix, storage strategy, facility efficiency, and emissions accounting.
• Community impact: Smaller phases may clarify local impacts, but siting, noise, water, land use, and public engagement remain material considerations.
• Economic competitiveness: Scalable compute capacity is one factor in retaining and attracting innovation-driven industries.

Conclusion

The Northeast faces a widening tension between demand for advanced computing and limits in electrical infrastructure. Modular, clean-energy-integrated facilities may complement traditional data center development by allowing phased capacity additions and more flexible use of grid and on-site resources, but they do not eliminate the need for interconnection studies, permitting, economic analysis, or community engagement.

As AI adoption and research computing expand, power availability will increasingly influence where digital infrastructure can grow. Policymakers, utilities, and industry leaders should evaluate modular development against other options using project-specific evidence on reliability, cost, emissions, land use, deployment timing, and community impact.

# # #

About the Author

Boaz Petrus is the CEO and Founder of NovaDynamic Technology Services.

References

¹ ISO New England, Load Growth & Demand Drivers in New England, March 6, 2025, citing the 2024–2033 CELT forecast. ² ISO New England, Interconnection Request Queue, including guidance on filtering and interpreting queue data. ³ Ehsan K. Ardestani et al., Provisioning to Runtime Optimization of a 100 MW-Scale AI Cluster, arXiv:2605.24461, revised October 2, 2026. ⁴ Uptime Institute, Global Data Center Survey 2024, July 2024; Daniel Bizo, Density choices for AI training are increasingly complex, June 25, 2025. ⁵ U.S. Department of Energy, Grid Deployment Office, Microgrid Overview, January 2024. Massachusetts policy context: Executive Office of Energy and Environmental Affairs, Massachusetts 2050 Decarbonization Roadmap, December 2020.

The post A Scalable Strategy for Power‑Constrained Regions appeared first on Data Center POST.

TL;DR

• Sun outages occur when solar radiation overwhelms satellite signals passing directly in front of the sun, causing temporary broadcast disruptions.
• Fiber TV users experience outages because national networks broadcast content via satellite before distributing it across local fiber systems.
• Outages stem strictly from solar positioning, meaning no technical solutions exist to prevent them.
• Empire Fiber Internet provides bulk Wi-Fi and streamlined service agreements for multi-dwelling units to ease property management operations.

# # #

Originally published on Empire Fiber Internet.

Solar Satellite Interference, also known as a sun outage or sunspot, occurs when solar activity disrupts geostationary satellite signals carrying television broadcasts. This natural phenomenon takes place when the sun passes directly behind a broadcast satellite, causing intense solar radiation to temporarily overwhelm the satellite’s transmission back to Earth. Consequently, affected television channels experience brief signal interruptions until the alignment shifts.

Viewers subscribed to fiber optic networks are also impacted because major television channels rely on satellite distribution across the country before entering local fiber systems. Because solar interference is entirely dependent on celestial positioning and solar radiation, there is no engineering solution available to prevent these temporary outages.

In addition to explaining broadcast disruptions, Empire Fiber Internet offers bulk service options and managed Wi-Fi solutions specifically designed for multi-dwelling units. These solutions streamline property administration through a single agreement, simplified installation, and unified resident support.

To read the original article, please click here.

The post What Are Solar Satellite Interferences? appeared first on Data Center POST.

TL;DR

• Mission-critical facilities require fuel systems designed around the specific operating environment, rather than a standardized approach.
• Data centers often prioritize scale, system consistency, fuel distribution, and visibility across complex or multi-site operations.
• Hospitals must account for continuity of care, existing infrastructure constraints, regulatory requirements, and emergency fuel access.
• ESI tailors fuel systems by combining engineering expertise, lifecycle support, operational intelligence, and emergency readiness to align each system with the mission it serves.

# # #

“Mission critical” describes many types of facilities, but it does not describe a single operating environment. A data center protecting digital infrastructure and a hospital supporting patient care may both depend on standby generators during a power interruption, yet their facilities, operating priorities, emergency procedures, and infrastructure constraints can be very different.

Those differences should influence how the fuel system is engineered and managed. Mission-critical fuel infrastructure cannot be treated as a standardized equipment package that is applied the same way from one facility to the next. The system has to reflect the environment it supports, the demands placed on the generators, and the consequences if standby power does not perform as expected.

For ESI Total Fuel Management, this is a central part of mission-critical specialization. ESI is purpose-built for environments where fuel-related generator failure is not an acceptable outcome, including data centers, hospitals, utilities, financial institutions, and government infrastructure. Its approach is centered on understanding the application first, then engineering and supporting the fuel system around the operational needs of the facility.

Data Centers: Designing for Scale, Growth, and Consistency

Data center fuel systems often have to support infrastructure at considerable scale. As facilities grow, the challenge is not simply storing enough fuel. Fuel must be received, maintained, monitored, distributed, and delivered to generators in a way that supports the facility’s critical power strategy.

That requires the individual parts of the system to function together. Bulk tanks, day tanks, fuel transfer equipment, receiving infrastructure, monitoring technology, and generator supply systems all contribute to the same operational objective. ESI’s approach reflects this interconnected relationship by managing fuel quality, storage, distribution, and generator supply as part of a unified system rather than as isolated components.

Scale also introduces the need for consistency. Operators managing several facilities may want common engineering standards and a repeatable approach to fuel system performance across different locations. At the same time, repeatability should not mean designing every site identically. Tank configuration, generator requirements, site access, local conditions, and facility design can vary significantly.

Hospitals: Engineering Around the Demands of Continuity of Care

Health care facilities place a different set of demands on standby power infrastructure. A hospital must be able to maintain essential operations through a power disruption while continuing to function as an active care environment.

That changes the context in which a fuel system is designed. Existing tankage, facility layout, available space, delivery routes, and emergency access may all place constraints on what can be installed and how fuel can move through the site. A system may need to integrate with infrastructure that has been operating for years rather than being designed as part of a new campus.

Regulatory considerations are also part of the operating environment. ESI’s fuel assurance model incorporates fuel handling, storage, and usage into an approach intended to support operational readiness, regulatory requirements, and audit confidence.

Emergency fuel access deserves similar consideration. Extended outages, severe weather, and grid instability can create conditions in which normal fuel deliveries become more difficult. ESI identifies disaster preparedness and emergency fuel access as important elements of mission-critical fuel management, helping facilities account for conditions that may extend beyond the capacity of their on-site storage.

Specialization Continues Throughout the Life of the System

A tailored fuel system should not become static once it is commissioned. Mission-critical facilities change over time as infrastructure expands, equipment ages, loads evolve, and operating requirements change.

Those changes can alter how the fuel system performs and what the facility requires from it. A data center may add capacity and generators. A hospital may renovate or expand existing infrastructure. Both may face changes in compliance requirements or emergency planning expectations.

ESI’s model addresses fuel-related performance across the lifecycle rather than treating installation as the end of the process. Systems can be evaluated, maintained, monitored, and optimized as operating conditions change. This reflects ESI’s broader position that mission-critical reliability depends on ongoing accountability rather than equipment performance at commissioning alone.

Operational intelligence can also help facilities compare original assumptions with actual system behavior. Insight into consumption, fuel condition, and system performance gives operators a clearer basis for evaluating readiness as the facility evolves.

Mission-Critical Specialization Is About Understanding What the Facility Protects

ESI approaches fuel as a performance-critical part of mission-critical infrastructure. By combining engineering expertise with lifecycle accountability and operational insight, ESI can tailor fuel systems around the specific requirements of the environments they are designed to support.

For data centers and hospitals alike, the goal is not to make every fuel system the same. It is to make each system appropriate for the mission it serves.

The post Mission-Critical Specialization: Tailoring Fuel Systems for Data Centers and Hospitals appeared first on Data Center POST.

TL;DR

• Fixed calendar maintenance is insufficient for modern high-density environments, necessitating condition-based monitoring and predictive strategies to catch early equipment failures.
• Advanced AI workloads and liquid-to-chip cooling introduce complex primary and secondary cooling loops, split responsibilities between stakeholders, and unique maintenance demands.
• Addressing the industry skills gap and integrating installation expertise into full lifecycle facilities management are essential for sustaining reliable operations across remote and evolving sites.

# # #

Written by iMiller Public Relations on behalf of Datalec Precision Installations (DPI).

Data center resilience requires effective maintenance, proactive monitoring, and skilled operational teams to maintain reliable performance as power, cooling, and operational requirements evolve. While redundant power and cooling systems provide a necessary foundation, 80% of serious outages stem from management, process, or configuration failures that could have been prevented. Facilities Management plays a key role in aligning technical capacity with ongoing operational performance through documented procedures, skilled personnel, and proactive monitoring.

Higher rack densities and the adoption to  liquid-to-chip cooling are also changing maintenance requirements. Liquid cooling creates shared monitoring responsibilities between operators and customers, making condition-based maintenance and predictive monitoring vital for identifying issues early. At the same time, a growing industry skills gap complicates these operations, with nearly two-thirds of operators reporting difficulties in recruiting or retaining qualified candidates to manage advanced mechanical and electrical systems.

Achieving resilience requires a lifecycle approach that considers infrastructure from installation and commissioning through ongoing operations. Decisions around system design, thorough commissioning, and clear documentation directly affect long-term maintainability, particularly as facilities expand into geographically remote areas. Integrating installation expertise with ongoing operational support enables operators to manage evolving risk, maintain equipment health, and maintain continuous delivery across the entire facility lifecycle.

To read the original article, please click here.

The post The Role of Facilities Management in Data Centre Resilience appeared first on Data Center POST.

TL;DR:

• One design isn’t enough to deliver data centres across Europe, because local planning, permitting, grid, water and heat rules shape how each project is approved and built.
• Hyceon’s baseline design draws on Europe’s toughest requirements, in Germany and Ireland.
• This helps projects avoid redesigns as rules tighten and protects long-term asset value.
• Hyceon’s local teams in each market handle permitting.
• Clients work with one team from feasibility through commissioning.

# # #

Consistent design standards, supported by local planning, permitting and regulatory expertise across Europe. 

By Brian Murphy, Managing Director, Hyceon

If you have delivered a data centre project in more than one European country, you know that successful delivery takes more than a good design. It also requires a clear understanding of the local planning, permitting, power and regulatory requirements that can affect project timelines and costs.

Even when the underlying design remains consistent, local requirements can significantly influence how a project is approved and built. Understanding these differences early helps avoid redesigns, delays and unexpected costs.

One design, different planning requirements

The challenges are familiar to anyone delivering data centre projects across European markets:

• Backup generators: In Germany, emissions-control permitting under the Federal Immission Control Act (BImSchG) can require generator flue stacks to meet calculated minimum heights. This can affect building height, massing and planning applications. In other markets, comparable requirements may not apply.
• Planning and architecture: Local plans can determine building heights, setbacks, façade treatments and landscaping. Some local authorities also analyse a building’s appearance as part of the planning process. A design accepted in one industrial zone may need a different building envelope a few hundred kilometres away.
• Grid connection and power supply: Connection requirements, capacity constraints and delivery timelines vary considerably across Europe. From Ireland’s requirements for new data centre grid connections to grid access milestones in Spain and Portugal, these factors influence site selection as much as design.
• Water and heat: Water availability, discharge requirements and opportunities to connect to district heating networks vary by region, as do the expectations of the relevant authorities.

These differences do not necessarily change the core design, but they can change how a project is approved and delivered. Understanding which elements can remain standardised and which need to respond to local requirements, from the earliest site assessment, helps prevent redesigns, delays and cost overruns.

Experience across Europe’s most demanding markets

Hyceon brings together experience from two of Europe’s most demanding data centre markets. Our roots go back to Dublin in 1988 and Frankfurt in 1990, and our experience in both markets continues to drive and strengthen our approach in data centre design and delivery.

In Germany, under the Energy Efficiency Act (EnEfG), new data centres beginning operation on or after 1 July 2026 must reach a power usage effectiveness (PUE) of 1.2 or lower. They must also reuse a minimum share of their energy: According to London-based Bird & Bird, the law sets 10% from 1 July 2026, 15% from 1 July 2027 and 20% from 1 July 2028 for facilities going into operation. Renewable electricity requirements, assessed on a balance-sheet basis, are tightening too, requiring 50% from 1 January 2024 and 100% from 1 January 2027, according to Stein law firm.

In Ireland, grid capacity decides whether a project goes ahead at all. Under the CRU’s December 2025 Connection Policy, new data centres at 10 MVA or above must bring their own dispatchable onsite or nearby generation and/or storage capacity, and the de-rated capacity must match the site’s maximum import capacity, reports law firm Mason Hayes & Curran. Within six years of energisation, 80% of their annual demand must come from additional renewable generation in Ireland, according to the CRU.

Together, these experiences inform our approach to projects across the UK, France, Italy, Spain, Portugal, Poland, Austria and the Nordic countries.

Why designing for demanding standards pays off

Our baseline design approach draws on Germany’s energy efficiency and energy reuse obligations, Ireland’s grid connection and security of supply requirements, and the common reporting framework established by the EU Energy Efficiency Directive.

In markets where requirements are less demanding today, this approach may appear to go beyond what is immediately necessary. In practice, designing with demanding regulatory standards in mind can help reduce the need for future modifications as requirements evolve across Europe.

A design built around robust standards can offer three clear advantages:

• Greater resilience to regulatory change: Reduces the likelihood of significant redesign as national requirements evolve.
• Long term asset value: Helps maintain the appeal of assets to investors and tenants as expectations change.
• Consistency across portfolios: Gives operators a common design standard rather than multiple country specific versions.

Heat reuse is a good example. In Germany, energy reuse obligations make it important to consider the infrastructure needed to capture and transfer waste heat. This can include reserving space for a heat transfer station, even where a suitable network is not yet available. In other markets, heat reuse may not yet be mandatory. Designing for it from the outset, with the necessary space, connection points and operating temperatures, preserves the option to connect to a district heating network if a suitable partner becomes available later.

Successful permitting depends on local expertise

A strong design does not secure planning permission on its own. Every country, and often every local authority, has its own procedures, timelines, decision makers and expectations.

That is why we coordinate permitting across markets while working with local teams who understand the relevant planning systems, regulatory requirements and approval processes. The people involved in pre-application meetings, public consultations and technical reviews understand how the process works, who needs to be consulted and what questions are likely to arise. They also understand the issues receiving increasing examination. Across Europe, authorities are paying close attention to noise and vibration, air quality, water and wastewater, ecology, traffic and visual impact.

Applications are better positioned when these issues are addressed before submission, rather than raised for the first time during the assessment process. We commission the necessary assessments early and incorporate appropriate mitigation measures into the master plan. This helps align the design submitted for approval with the project ultimately delivered.

Designed for demanding standards. Delivered locally.

With Hyceon, operators benefit from consistent design standards across markets, developers gain access to local planning and regulatory expertise, and investors benefit from assets designed with evolving requirements in mind.

Clients work with one Hyceon team, from feasibility through to commissioning, bringing together:

• One consistent design standard, informed by the most demanding applicable requirements.
• One technical assurance and quality process across every project.
• Shared sustainability benchmarks for energy, water, carbon and materials.
• Local teams and partners in each market, equipped to work in the language of the relevant authorities.

The result is a consistent European approach, supported by the local expertise needed to turn designs into deliverable projects.

Talk to our team about your next European data centre project.

Regulatory information checked as of October 2026.

The post Hyceon: A New Pan-European Force in Data Centre Design and Engineering appeared first on Data Center POST.

TL;DR

• The Nomad Futurist Foundation and iMiller Public Relations (iMPR) have extended their longstanding collaboration through a new three-year agreement focused on workforce awareness and educational outreach.
• The partnership coincides with iMPR’s 15th anniversary and aims to introduce more students to career opportunities across the digital infrastructure industry.
• iMPR will support the Foundation’s educational initiatives through strategic communications, industry storytelling, media relations, and platforms including Data Center POST and NEDAS Live!
• As demand for skilled professionals increases, partnerships between industry organizations and educational programs are becoming increasingly important to workforce development.

# # #

The rapid expansion of AI, cloud computing, and data center infrastructure is creating new career opportunities across the digital economy, but attracting qualified talent remains a significant industry challenge. As organizations invest in new facilities and technologies, workforce development and industry education are becoming increasingly important to supporting long-term growth.

Against this backdrop, the Nomad Futurist Foundation and iMiller Public Relations (iMPR) have extended their longstanding collaboration with a new three-year agreement focused on workforce awareness and educational outreach. Announced November 2, 2026, the partnership also marks a milestone for iMPR as the communications firm celebrates 15 years serving the digital infrastructure industry.

Addressing the Digital Infrastructure Talent Gap

Finding qualified professionals to design, build, operate, and maintain digital infrastructure is becoming increasingly difficult as the industry expands. The challenge extends across technical disciplines, highlighting the need to introduce students to career opportunities long before they enter the workforce.

According to Deloitte’s 2026 research, 63% of surveyed data center executives identified skilled labor shortages as their leading talent challenge, while job postings for core technical roles increased 64% between 2023 and 2025. These findings reinforce the importance of developing a stronger talent pipeline through education, professional exposure, and industry collaboration.

The Nomad Futurist Foundation addresses this need through programs designed to make digital infrastructure more accessible and understandable to students. Initiatives such as the Nomad Futurist Academy, scholarships, university partnerships, and global ambassador programs introduce emerging talent to the technologies, professionals, and career pathways supporting the digital economy.

Under the renewed agreement, iMPR will help expand awareness of these opportunities through strategic communications, thought leadership, media relations, event promotion, and educational outreach. The collaboration will also leverage industry platforms including Data Center POST and NEDAS Live! to connect the Foundation’s programs with a broader audience.

Expanding Industry Engagement and Educational Opportunities

The partnership extends beyond career awareness to address the importance of stronger relationships among students, educators, industry professionals, and the communities where digital infrastructure is developed. iMPR Founder and CEO, Ilissa Miller, who also serves as an advisor to the Nomad Futurist Foundation, supports these efforts through her involvement with the OIX Association‘s Digital Infrastructure Framework Committee and iMPR’s Groundswell™ community engagement program.

Miller’s participation in initiatives such as the CxC Hackathon 2025, where she spoke to more than 350 students about the industry’s talent gap and career opportunities, reflects the value of direct engagement with future professionals. Such interactions can help students better understand the variety of roles available across data centers, telecommunications, connectivity, and emerging technologies.

The renewed collaboration will focus on making digital infrastructure careers more visible to students and educators while encouraging greater participation from industry professionals. Through educational outreach, industry storytelling, and engagement initiatives, the organizations aim to strengthen connections between emerging talent and companies seeking the next generation of skilled workers.

These efforts align with the Foundation’s three strategic pillars of Brand & Engagement, Programs & Impact, and Partnerships & Growth. By increasing the visibility of educational initiatives and encouraging broader industry participation, the partnership will support the Foundation’s efforts to reach more students and expand its workforce development programs.

As demand for digital infrastructure continues to accelerate, workforce readiness will remain a critical consideration for companies developing and operating the next generation of facilities and networks. Expanding awareness of career opportunities and strengthening connections between education and industry will be essential to building the talent pipeline needed to sustain that growth.

Read the full announcement and explore how the Nomad Futurist Foundation and iMPR are working to expand awareness of digital infrastructure careers.

The post Building Tomorrow’s Digital Infrastructure Workforce Through Education and Industry Collaboration appeared first on Data Center POST.

TL;DR

• Grid connection timelines do not match AI data center deployment schedules, creating a multi-year gap where facilities are ready in two to three years while utility interconnection can take up to seven years.
• Data center electricity demand is rapidly climbing, projected by Lawrence Berkeley National Laboratory to increase from 4.4% of U.S. electricity in 2023 to between 6.7% and 12% by 2028.
• Developers are actively leveraging on-site primary generation, phased energization, demand flexibility, and repurposed sites with existing headroom to bridge the gap until grid power arrives.
• Power strategy must be prioritized from day one as a foundational design input alongside land and fiber, rather than treated as a late-stage procurement task

# # #

By Sean Hughes, Business Development Manager, Wärtsilä Energy

Everyone wants to talk about AI. Very few people want to talk about the one thing that decides whether any of it actually gets built: power. Not cooling. Not chips. Not even land. Power.

More specifically: when can I get it, and how much? A lot of projects start with a conversation that goes something like this: “We’ve got the land. We’ve got the tenant. The utility says 2031. Can you help?”

Developers can line up the site, the capital and the tenant. What they increasingly can’t line up is a grid connection that matches their schedule. That’s the conversation dominating the U.S. market right now, and it’s changing how data centers get planned, financed and delivered.

Two clocks, running at different speeds

Here’s the core problem. A hyperscale or AI campus can go from site selection to ready-for-load in two to three years. The grid infrastructure to serve it runs on a completely different clock.

New transmission takes the better part of a decade to plan, permit and build. Large transformers and switchgear have long lead times. Across the United States, developers are being quoted multi-year waits for large new loads. In response, grid operators in PJM, ERCOT, MISO and SPP are rewriting the rules for how those loads are studied and connected.

Meanwhile, demand keeps climbing. Lawrence Berkeley National Laboratory estimates data centers used about 4.4% of U.S. electricity in 2023 and could reach somewhere between 6.7% and 12% by 2028. Utilities haven’t planned for load growth like this in a generation.

I want to be fair to the utilities here. They have to protect every other ratepayer from paying for capacity that never gets used, and their queues are full of speculative requests. Nobody’s doing anything wrong. It’s a system built for slow, steady growth running into a step change.

But a developer with a tenant waiting doesn’t have seven years. And the tenant definitely doesn’t.

What developers are actually doing about it

My view is simple: waiting on the grid is not a strategy. When the grid date and the build date don’t line up, there are really four levers to pull. The best projects I see pull more than one.

Phase it. A partial connection now is often available sooner than a full one later. Build the campus to energize in stages and get revenue flowing while the rest catches up.

Be flexible. A data center that can reduce its demand on the grid for a limited number of hours each year can be much easier to connect.

Utilities and grid operators are starting to reward that.

Bring power on site. Dispatchable generation on site can carry part or all of the load until the grid arrives, then shift into a different role once it does.

Rethink the map. Sites with existing headroom, retired power plant sites with interconnection rights still attached, and locations near new generation are worth a lot more than they were a few years ago.

Primary power isn’t backup power

This is where I see the most confusion.

Backup generation is the spare tire. It sits in the trunk and you hope you never need it. Primary power is the engine. It runs every day, carrying the load as the main source of supply. Those are completely different problems, with different equipment, different economics and different operations. Plenty of developers are facing the primary power question for the first time, and it pays to treat it that way.

And going with primary power doesn’t mean walking away from the grid. Fully islanded campuses have to carry all of their own redundancy and reserve margin, and they give up the resilience a big interconnected system provides. I think of the better approach as a macro-grid: dependable power on site that can run the campus on its own, then connect to the grid when the timing makes sense. Once connected, explore opportunities to export power, benefiting both communities and data centre operators. The added capacity lowers electricity prices while giving operators an additional revenue stream.

The smarter play is building enough power for what the data center needs today, and designing in the flexibility to be part of the grid tomorrow. That’s not a sales line. It’s an engineering reality. On-site capacity that can flex down at peak, or support the system when it’s tight, changes the whole relationship. The result? A data center that’s a grid asset, not a grid problem. And a large load that can flex, is one that utilities are far more willing to connect.

Will it be on when you need it?

Developers don’t want a spec sheet, they want answers to two questions: will it be on when we need it, and can you fix it if it’s not?

That second question is going to matter more than people think. A lot of new players are winning U.S. data center work right now. Not all of them have the field technicians, parts and service coverage in place to back it up. A primary power plant is only as reliable as the people and parts behind it. Ask about that before you sign, not after your first outage.

Fuel deserves the same scrutiny. Whatever you build will likely run for decades. Fuel flexibility, predictable operating costs and getting the most out of every unit of fuel all add up to cost certainty over the life of the asset. That’s good business, and it’s good stewardship of resources.

Put power first

If I could give developers one piece of advice, it would be this: stop treating power as a procurement task that starts after the site and design are locked in. By then, your options have narrowed and your schedule is already at risk.

Power belongs at the very start of the plan, right alongside land and fiber. Talk to the utility before you commit to a site, and treat its realistic timeline as a design input. If on-site generation will carry load for a while, plan its permitting, fuel and service from day one. Bring in your EPC and power specialists early, so generation and grid connection get engineered as one system.

The grid will catch up. Utilities and grid operators are investing heavily, and new transmission and generation are on the way. But that will take years, and the AI build-out isn’t waiting.

Speed to power is everything right now. The developers who win will be the ones who treat power as a core part of the development strategy from day one, not a constraint to solve later. If you’re thinking through what that could look like for your next project, give Wärtsilä a shout.

# # #

About the Author

Sean Hughes is a Business Development Manager at Wärtsilä Energy, focused on primary power for data centers in the U.S. market. He has 17 years of business development experience across the LNG and power generation industries and works with data center developers, operators and EPCs on power strategy, grid connection and on-site generation.

The post Speed to Power Is Everything Right Now. Here’s Why the Grid Can’t Keep Up. appeared first on Data Center POST.

TL;DR

• Securing power grid connections is the primary bottleneck halting new facilities, with U.S. data center electricity usage projected to increase from 4.4% in 2023 to between 6.7% and 12% by 2028.
• Rather than permanent site headcount, the core economic impact stems from heavy construction spending and tax revenues, supporting 5.5 million total U.S. jobs and $204 billion in tax contributions in 2024.
• Water consumption is a deliberate engineering choice rather than an inevitability, as closed-loop sealed systems recycle water continuously while open evaporative builds consume water for heat rejection.

# # #

What a data center actually costs

Data centers are a ginormous investment, but the environmental impact is higher. It relies not only on the operation, but also on the colossal construction that we see.

A typical data center invests in construction, raw materials, labour, land, electricity, water. But still basic building components

What separates it from a regular construction is the tech hardware. GPUs, running to thousands of dollars in a single square foot of space. Network connection, time, energy, memory, rack, vSAN, rack firmware and many more.

And only after building the entire building of the data center that we can actually operate it.

What the scale actually looks like

It is the usage of this labour force. And it’s the colossal structures. Based on county permitting records and facility reporting, xAI’s Colossus cluster spans roughly 207 acres and about 2.6 million square feet of building across its three sites in Memphis and Southaven, holding around 770,000 GPUs and delivering about 1 GW of compute as of March 2026. That is roughly 45 football fields of building. That is what you’re thinking of when you compare the DC.

The labour nobody counts

In all of this we forget the time and cost of engineers and skilled labour transported from all parts of the world. Maintenance and construction workers for the building, access points and portals, generators, electrical engineers, mechanical engineers, plumbers, civil engineers, the big architecture firm employed to complete the work, and the construction companies building each one of the units in the most cost-effective method.

Once the DC is set up you have the network operators, the servers, the actual IT teams, and many more employed for the maintenance and career-long employment that this development would otherwise not have created.

What the buildings give back

There are benefits of data centers, in society as a whole. The introduction of employment opportunities and the economic value unlocked for society as a whole.

By the Data Center Coalition’s own accounting, the US data center industry directly employed more than 1 million workers in 2024 and supported 5.5 million jobs in total, with a tax contribution of $204 billion to local, state and federal government. The American Edge Project projects a further 4.7 million temporary construction jobs, against buildout spending McKinsey estimates at up to $7 trillion by 2030. Those are industry figures and should be read as such, but the construction side of them matches what I have seen on site.

Power, not code, is the bottleneck

Energy utilization in operation has never been the whole burden. Getting power to the site is. Data centers used about 4.4% of total US electricity in 2023, and the Lawrence Berkeley National Laboratory projects 6.7% to 12% by 2028, which is exactly why interconnection, not code, is the thing that stops a build.

Water, and the loop everyone describes backwards

Only after which the operation itself is a conversation point. Only after the completion of development itself could it be reduced to an understanding of climate impact. The DC does not by itself operate in a closed or open loop system.

It’s a choice. In a closed loop system the water circulates in a sealed circuit and is reused, so almost none of it is consumed. It is the same technology as the air conditioning you’re sitting in, the same systems used in offices, malls and your car. An open loop or evaporative system is the one that consumes water, because the water is evaporated to reject heat and has to be continuously replaced. So if your standalone office building on a sealed loop doesn’t drain the local water supply, a data center on the same design won’t either. The ones that do are the evaporative builds, and that is a design decision, not an inevitability.

The comparison problem

You also have to see the impact of DCs not on an individual level but from the humanitarian impact of DCs as a whole. Industry trackers count roughly 2,100 operational data centers in the US, or as many as 4,400 depending on whether enterprise sites are included, with another 800 under construction, catering to hundreds of millions of users globally, including all AI and non-AI users. A simple Google search today is also powered by AI, and Google says it engages over a billion people in AI-assisted interactions each day. Can any person interacting with digital tech go without seeing the effects of AI in their day-to-day lives?

It’s like comparing a group activity to the individual impact. And creating a massive misalignment in the expectation of how much power will be used, measured against a sub-Saharan comparison. Let me tell you, it is also high power to use your phone and device to read this article, given that the same water could be provided to people in sub-Saharan Africa.

The US Geological Survey puts the water consumption of a quarter-pound beef patty at about 460 gallons, and a full cheeseburger at around 660. Against the minimum annual domestic water need used in the UN University’s own reporting, roughly 7,400 litres per person, four beef patties is one person’s entire year of water in sub-Saharan Africa. Every industry looks monstrous under this metric. So if saving water and environmental impact is the goal of your life, starting by quitting your cheeseburgers is a good strategy.

The comparison, by far, makes an individual-to-individual consumption and impact. You would at least therefore actually be able to contribute and balance the compensation of using AI, by quitting a cheeseburger once a month.

So next time you think a DC is bad for the environment, it will be the same as questioning the field of agriculture, transportation, steel, or anything consumer.

What this means if you are building one

Everything above is an argument about context. It is not a defence you get to use. If you are financing, siting, or operating a build, it is a set of conditions you have to actually meet. Six things I would put in front of a board.

1. Pick the loop, and publish the choice. Closed versus evaporative is the single decision that most changes your water story, and almost nobody outside the industry knows which one you picked. If you are on a sealed loop, say so in plain language with a number attached. If you are evaporative, say that too, say why, and say where the blowdown goes. Operators who stay quiet get assigned the worst case by default, and they deserve to be.
2. Lead the community pitch with construction and tax base, not permanent headcount. A 250,000 square foot facility runs on roughly 50 full-time staff, about half of them contract. If you promise a county a jobs engine, someone will check that number and it will embarrass you. The construction trades, the supply chain, and the tax contribution are the real case, and they survive scrutiny.
3. Treat interconnection as the long pole, not the land. Ireland paused new Dublin-area grid connections until 2028. New York imposed a one-year ban on new hyperscale facilities. The grid queue and the permitting calendar are what kill schedules now. Land, steel, and chips are procurable. Megawatts on a date certain are not.
4. Report serving energy, not just training. Most of the lifecycle energy of a model is inference, not the training run. A sustainability report that covers only training is reporting the smaller half and inviting the correction.
5. Do not budget for efficiency to rescue you. Per-query energy is falling, but computations per joule now double roughly every 2.6 years, not every 18 months, and cheaper inference reliably produces more inference. Plan capacity against demand growth, not against a hoped-for decline in the per-query number.

The industry’s problem is not that the criticism is unfair. It is that we have handed people almost nothing to check it against. Every operator that publishes a real number makes the next build easier to approve, including yours.

Why I am writing this

The headlines focus on multibillion-dollar investments and chip shortages, but the immediate bottleneck is not code. It is concrete, high-voltage power, specialized global supply chains, and the workforce needed to scale data centers. Even YC’s call for startup addresses this business prospect.

I have been the lead project manager on data-center builds for companies including Tesla, Cisco, and major data companies. I started by managing physical builds in India and later brought that infrastructure experience into the U.S. startup ecosystem. The article aims to give a real perspective on Datacenter investment.

# # #

About the Author

Karishma is an expert in AI infrastructure and system design. She writes about the newest AI technology and about the changing landscape of tech affecting every other field.

Founder of Aidni, her company focuses on power analytics for data centers. She has previously held AI positions at Salesforce and Cisco

The post The Math and Engineering Behind Data Centers appeared first on Data Center POST.

TL;DR

• Europe’s data center investment market continues to attract institutional capital but power availability, development costs, and execution risks are influencing investment decisions.
• Growing constraints in established FLAP-D markets are encouraging investors to explore opportunities in Spain, the Nordics, and Southeastern Europe.
• Energy storage, alternative power solutions, and advanced cooling technologies are becoming increasingly important considerations in data center valuations and development strategies.
• Evolving computing requirements, operational expertise, and data sovereignty are creating new opportunities and challenges for investors across the European market.

###

Europe’s data center investment market is entering a period of significant transformation. While demand for digital infrastructure continues to grow, investors are navigating increasingly complex challenges involving power availability, construction costs, regulatory requirements, and evolving computing technologies.

These issues were central to IMN’s Data centers Private Equity Europe, held October 1, 2026, at the Hilton London Tower Bridge in London. The event brought together investors, developers, operators, and financial institutions to examine the opportunities and risks shaping Europe’s data center market with discussions spanning investment strategies, emerging markets, energy infrastructure, and long-term asset performance.

Investment Opportunities Expand Beyond Established Markets

Europe remains an important destination for institutional investment in digital infrastructure supported by demand for cloud computing, artificial intelligence (AI), automation, and digital services. However, investors are increasingly evaluating how infrastructure constraints and regional market conditions influence risk-adjusted returns.

The European investment landscape was a central focus of discussions featuring Helena Fagraeus Lundstrom, CEO of T. Loop, who participated in an examination of the continent’s evolving data center investment thesis. The session addressed the factors influencing capital allocation across hyperscale, colocation, and edge infrastructure, along with the economic and geopolitical considerations affecting future investment.

As established markets such as Frankfurt, London, Amsterdam, Paris, and Dublin face mounting development constraints, attention is expanding toward emerging European locations. Regional discussions examined opportunities in Madrid, Barcelona, the Nordics, and Southeastern Europe where power availability, land costs, permitting requirements, and local partnerships are becoming critical factors in investment decisions.

The growing interest in secondary markets reflects a broader shift in how investors evaluate development opportunities. Rather than focusing exclusively on established data center hubs, investment strategies increasingly require a detailed understanding of regional infrastructure readiness and long-term market fundamentals.

Power Availability Reshapes Investment Decisions

Access to reliable, affordable power has become one of the most significant factors determining where new data center developments can proceed. Across Europe, grid constraints and lengthy connection timelines are complicating site selection, project financing, and development schedules.

The importance of long-term power access was examined in a discussion that included Artur Brandao, Head of Infrastructure Investments at euNetworks. The session challenged participants to consider whether investors should avoid markets without guaranteed power availability, highlighting the growing relationship between energy infrastructure and investment risk.

Beyond traditional grid connections, developers are exploring battery energy storage systems, microgrids, on-site generation, and other alternative energy solutions. These technologies could help bridge infrastructure gaps while creating opportunities for data centers to participate in demand response and grid-balancing initiatives.

For investors, energy strategies are becoming an increasingly important component of asset evaluation. Understanding power procurement, infrastructure development timelines, and the economics of alternative energy solutions is essential to assessing project feasibility and long-term returns.

Development Risks Demand Greater Financial Discipline

Strong market demand does not eliminate the operational challenges associated with delivering large-scale infrastructure. Equipment shortages, contractor availability, supply chain disruptions, and rising construction costs continue to introduce uncertainty into development schedules and investment projections.

Addressing these concerns, Jonathan Fielding, Investment Director at Tikehau Capital, participated in a discussion examining how developers and investors can manage construction risks and project delays. Topics included procurement strategies, supplier diversification, contingency planning, and the expectations of lenders and equity partners before committing capital.

These considerations extend beyond individual projects. Delays in securing transformers, specialized equipment, and other essential components can affect financing arrangements, customer commitments, and anticipated investment returns.

As European data center developments become larger and more technically complex, disciplined risk management is increasingly important. Early procurement planning, realistic construction schedules, and closer coordination among developers, contractors, and financial partners can help reduce uncertainty throughout the development process.

Changing Computing Requirements Influence Asset Valuations

The rapid evolution of computing technology is also changing how investors assess existing facilities and future developments. Higher rack densities, advanced cooling requirements, and new tenant expectations are influencing both infrastructure design and the economics of data center ownership.

These considerations were explored in a discussion featuring Ekaterina Almasque, Co-Founder and Managing Partner at BlankPage Capital, focused on redesigning data centers for a new computing era. The session examined the financial implications of retrofitting existing facilities compared with developing purpose-built infrastructure capable of supporting AI-intensive workloads.

Cooling technologies were another important component of the investment discussion. Direct-to-chip cooling, liquid immersion, and hybrid systems offer potential approaches to managing higher-density environments, but their adoption also introduces considerations involving capital expenditure, operational complexity, and environmental performance.

The implications extend to portfolio strategy. Investors must consider how changing customer requirements could affect existing assets, whether infrastructure upgrades can deliver competitive returns, and how technological developments may influence future demand for data center space.

Data Sovereignty Adds Another Dimension to European Investment

Alongside energy and technology, regulatory priorities are influencing the structure and ownership of Europe’s digital infrastructure. Data sovereignty, domestic control of critical infrastructure, and international partnerships are becoming increasingly relevant to investment decisions.

The balance between European infrastructure independence and global investment was explored in a discussion that included Mark Daly, Head of Data Centers & Emerging Technologies at BloombergNEF. The session examined whether sovereignty should depend on European ownership, effective control of critical assets, or risk-based approaches that accommodate international partnerships.

These questions highlight the complexity of investing in infrastructure that supports both national economic priorities and global digital services. As governments place greater emphasis on infrastructure resilience and digital sovereignty, investors must account for regulatory expectations alongside financial and operational considerations.

Balancing Growth Opportunities With Investment Realities

IMN’s Data centers Private Equity Europe highlighted the interconnected factors shaping investment decisions across the continent. Power availability, emerging market opportunities, construction risks, technological change, and regulatory requirements are influencing how capital is deployed and how projects are evaluated.

While demand for digital infrastructure continues to create investment opportunities, successful development increasingly depends on understanding the constraints that can affect project execution and long-term asset performance. Investors and developers must balance growth ambitions with practical assessments of infrastructure readiness, operational requirements and financial risk.

Additional insights into the investment strategies and market developments shaping the sector at next year’s IMN’s Data centers Private Equity Europe that will take place in October 2027.

The post New Markets, New Risks, and the Future of European Data Centre Investment appeared first on Data Center POST.

TL;DR

• Operating efficiency and financial outcomes depend heavily on local climate conditions, Power Usage Effectiveness (PUE) metrics, and sustainable water access for cooling systems.
• High-performance data center locations require redundant fiber pathways, low latency measured in milliseconds, and proximity to internet exchanges to support time-sensitive digital applications.
• Long-term project viability relies on assessing grid readiness, resource resilience, expandable land, permitting regulations, and access to a qualified technical workforce simultaneously.

# # #

Data centers tend to sprout up in locations where land and energy costs are low. However, these factors are just part of why data centers are established in a certain location. New developments must also ensure a source of dependable power, a sufficient amount of water, a strong fiber system, suitable weather, and a pool of labor available.

The advent of advancements like AI, cloud computing, and digital services changes the way data center developers assess sites. Thus, a low-cost site may be complicated to operate at some point due to rising electricity and expansion issues.

As a result, the upcoming generation of data centers will rely on a variety of technical, environmental, economic, and regulatory affected factors. By understanding these lesser-known factors, one can better understand the shifts in investments to certain locations and the struggles of other locations that may seem appealing.

Electricity Infrastructure Is More Important Than Low Rates

Electricity is one of the most important factors determining the location of data centers, especially when it comes to AI-based workloads. In the year 2024, worldwide electricity consumption by data centers is expected to be approximately 415 TWh, or about 1.5% of the total electricity demand. By 2030, the consumption figure will reach about 945 TWh because of the development of AI and increase in demand for digital services.

However, the concern here is not merely to find an electricity provider. A major site may need many hundreds of MW of stable capacity, as well as substations, transmission lines, and backup supplies. A 200 MW data center is supposed to consume about 1.49 million MWh of electricity per year at an average usage rate of 85%.

Considering that locations with sufficient capacity can be advantageous over low rate locations, it may be crucial to keep in mind that necessary infrastructure development may take years. Some aspects, such as transformer availability, waiting lists at power lines, and conditions of long-term contracts should be considered in relation to construction timing.

Impact of the Climate on Long-Term Operating Costs

The impact of temperature is such that it dictates how much energy is required to remove heat from the servers. Conventional cooling system performance diminishes considerably in the event of high ambient temperatures, while the use of liquid cooling systems proves beneficial for moderating concentrated heat produced by modern AI accelerators.

Power Usage Effectiveness (PUE) is determined by taking the energy consumed by the center divided by the amount of energy used by the IT equipment. For instance, if PUE is equal to 1.2, it indicates that the facility, in addition to heating, utilizes 1.2 units of power to receive one unit of power for the operation of computers. For example, if the IT load is 100MW, PUE is 1.2 that indicates that the power necessary for the center is 120 MW. The same load with PUE of 1.5 implies that it requires 150 MW of power.

This discrepancy in the numbers will prove to be significant particularly with regard to the financial outcomes of the business for the year ahead. Therefore, we can conclude that climate has an influence, but that there are also many elements that can affect operating efficiency like humidity, cooling technology, server density, and operational practices.

As a result, the most desirable places are those that have climate advantages supported by reliable infrastructure and acceptable environmental risks.

Water Availability Might Be Important for Site Selection

Water is a criterion that is often neglected. Many data centers operate water-based cooling systems, and water consumption can increase during times of extreme heat and heavy computing usage. Areas facing drought or competing for water due to agricultural and residential uses may impose regulations that will have consequences for operations for years.

The extent of the issue depends on cooling technology. A data center with a continuous IT load of 100 MW will consume 2,400 megawatt-hours of IT electricity per day. The cooling requirement can turn into high water needs; however, water usage depends significantly on the environment, equipment, and design used.

The water stress status is now being evaluated by most corporations across watersheds instead of only rainfall assessments. Several technologies are now available to reduce freshwater usage, including recycled wastewater, closed-loop cooling, and air cooling. The major disadvantage of these technologies is their reliance on power usage, capital investment, and maintenance costs. Water availability must not merely be considered along with energy usage.

Geographical and Connectivity Factors Influence Location Choices

Data centers utilize a large amount of fiber to send data to their clients, businesses, and other data centers. Latency, which is calculated in milliseconds, is particularly relevant in industries dealing with finance, industrial automation, and transportation. A difference of only 10 milliseconds may be important for particular time-sensitive services.

Any place near any of the existing Internet exchanges, cable landings, and cloud hubs can be considered a good location since it has a smaller distance to the point of connection, however, the geographical placement alone cannot guarantee the effectiveness of communication. Fiber diversity, traffic, routing, or availability of alternative pathways are crucial factors, as well.

As per Dataintelo, the worldwide next generation data center market was valued at $163.2 billion in 2025 and is anticipated to grow to $510.6 billion by 2034, indicating a compound annual growth rate of 13.5% over the forecasting period.

The forecasted growth demonstrates the demand for computing capabilities, stable network infrastructures, and premises able to support the most advanced artificial intelligence and cloud computing workloads. In connection with the increasing investments, the selection of sites will depend on the ability of local utility grids, water resources, and connection links to meet the needs of the far-sighted future.

Regional clustering has advantages as well. An established data center hub can provide multiple independent fiber routes, which may lead to a reduction of the risks associated with the investment. Excessive centralization can lead to pressure on local energy grids and land values, thereby creating more favourable conditions for secondary markets in the future.

The Effect of Regulation, Labor, and Climate Risk Factors

Even in the best of locations, problems regarding permits, environmental assessments, and limitations on energy and water resources may result in construction delays. Construction timelines are mainly dependent on local procedures for obtaining permits and the presence of energy resources on the grid, which may take longer to get ready than the data center itself.

Availability of workforce is also vital. Facilities need to hire electrical engineers, cooling specialists, security experts, and technicians who can ensure the proper functioning around the clock. Regions with strong IT branches can allow for larger talent pool selection, but the cost of the land in remote locations may be lower, while more money will have to be invested in training and retaining the staff.

Climate risks are another significant factor to consider. There are several damages that can be caused by flooding, wildfires, heat waves, and storms. Thus, it becomes necessary to take into account historical hazards, climate forecasts, insurance costs, as well as roads and transmission lines.

The key aspects involved in selecting a site are:

• Grid Readiness: The future ability to provide the required power capacity on time and according to schedule.
• Resource Resilience: The sustainability of the power supply and water in view of seasonal deficiencies and severe weather.
• Expansion Potential<span”>: The future capability of the site to cover new construction in the years to come.
• Operational Readiness: The existence of skilled personnel, support services, and network links.

These aspects are interdependently connected. For example, a cheap land site may have connections taking too long to be established. In some area with plenty of renewable energy, transmission may be a problem. These factors require simultaneous consideration thus helping developers understand the opportunities of the site in the future.

Conclusion

The geography of future data centers will depend mainly on the convergence of infrastructure, environmental limitations, network efficiency, and long-term operational risks. By 2030, global electricity consumption by data centers is predicted to rise to around 945 TWh which will put additional pressure on various regions to generate reliable energy amid other resources that are equally necessary.

Success in choosing a site will no longer rely so much on peeking for the cheapest land or even land best suited for data centers. Rather, it would be pre-eminently essential to find locations where factors, such as electricity, water provision, connectivity, workforce availability, and regulation would help find opportunities for data centers of the future.

# # #

About the Author

Ashish Kolte is a Marketing Manager at DataIntelo with expertise in marketing, market intelligence, and business strategy. He combines marketing insights with industry research to analyze market trends, identify growth opportunities, and provide data-driven perspectives on emerging industries and global business developments.

The post The Hidden Factors That Determine Where the Next Generation of Data Centers Gets Built appeared first on Data Center POST.

TL;DR

• AI is changing the economics of global connectivity, increasing demand for high-capacity fibre networks, distributed infrastructure, and new service models.
• Greater collaboration among carriers, data centre operators, and cloud providers is becoming essential to supporting AI workloads.
• Power availability, permitting delays and community opposition remain significant obstacles to Europe’s digital infrastructure expansion.
• Network automation, infrastructure investment, and resilience are increasingly important to long-term growth.

# # #

The telecommunications industry is facing a fundamental shift in how networks are built, operated, and monetized. Artificial intelligence (AI) is creating new infrastructure demands, while power limitations, investment pressures, and evolving customer expectations are challenging traditional business models.

These issues shaped discussions at Capacity Europe 2026, held October 12–15 in London. Bringing together leaders from telecommunications, cloud computing, data centers, and digital infrastructure investment, the event explored how the industry can translate growing demand for connectivity into sustainable business opportunities.

AI Is Changing What Networks Need to Deliver

As AI workloads become increasingly distributed, network infrastructure must support more than traditional data transport. High-capacity fiber corridors, low-latency connectivity, and reliable interconnection between computing facilities are becoming essential components of AI deployment.

The implications of AI-generated internet traffic were among the topics explored during a keynote by Avijit Singh, Chief Strategy Officer at Colt Technology Services. His presentation examined the growing importance of autonomous networks and the capacity, security, and governance challenges associated with increasing machine-generated activity.

Beyond the technical requirements, discussions addressed the commercial opportunities associated with AI infrastructure. Service providers are evaluating how managed services, automation, and strategic partnerships can create additional value for customers rather than relying exclusively on connectivity capacity.

Fiber Investment Faces New Economic Realities

Europe’s fiber market is entering a period in which network utilization, financial performance, and long-term asset value are receiving greater attention. While demand for connectivity continues to increase, operators and investors must balance expansion with the practical economics of deploying and maintaining infrastructure.

The development of fiber corridors capable of supporting AI training, inference, and edge computing was examined by industry leaders including Marisa Trisolino, CEO of euNetworks. The discussion addressed network design, dark fiber strategies, and investment decisions as providers prepare for increasingly demanding workloads.

Financial considerations were equally important as Europe’s fiber market matures. A panel examining the changing valuation of fiber assets included Thibault Contat Desfontaines, Co-Head of Infrastructure Equity, Europe at UBS Asset Management alongside other infrastructure and investment executives. The focus on long-term value reflected an environment in which network reach alone is no longer sufficient to establish a competitive advantage.

Infrastructure Convergence Creates New Opportunities

The traditional boundaries separating telecommunications networks, data centers, and cloud platforms are becoming less distinct. Supporting AI applications requires greater coordination across these infrastructure segments, particularly as workloads move between centralized computing environments, and distributed locations.

Opportunities for greater collaboration were explored in discussions involving carriers, colocation providers, and technology companies. Among the participants was Paula Cogan, Managing Director, EMEA at Digital Realty who joined a panel examining shared infrastructure models, strategic partnerships, and the convergence of Europe’s digital infrastructure ecosystem.

This convergence also raises questions about ownership, investment responsibility, and revenue generation. Partnerships between carriers, colocation providers, internet exchanges, and cloud companies could help address capacity constraints while improving service delivery.

Power, Permitting and Resilience Remain Critical Challenges

Despite strong demand for digital infrastructure, Europe’s ability to expand depends on factors extending well beyond network technology. Electricity availability, grid access, permitting requirements, and community acceptance continue to influence where projects can be developed and how quickly they can proceed.

The relationship between power availability and data centre expansion was explored during a panel that included David Adkins, Head of Network Architecture and Innovation at National Grid. The discussion examined grid access, on-site generation, and renewable energy strategies as potential approaches to supporting infrastructure growth.

Infrastructure resilience also received attention as geopolitical uncertainty, physical threats, and cybersecurity risks increase the importance of protecting critical connectivity assets. Discussions emphasized route diversity, subsea cable security, and stronger coordination between governments and network operators.

These challenges reinforce the need to consider energy, security, regulatory requirements, and community engagement early in infrastructure planning rather than treating them as separate issues after investment decisions have been made.

Building a More Connected and Sustainable Future

Capacity Europe 2026 highlighted how the industry’s next phase of growth will depend on more than increasing network capacity. Commercial innovation, strategic investment, infrastructure collaboration, and operational resilience will all influence how successfully providers respond to AI-driven demand.

Continue exploring the technologies, investment strategies, and partnerships shaping global connectivity at Capacity Europe 2027, taking place October 19–21, 2027 in London.

The post Capacity Europe 2026 Examines the Changing Economics of Global Connectivity appeared first on Data Center POST.

TL;DR

• MaiaEdge uses SD-Ethernet architecture to provision private Ethernet instantly over fiber, internet, 5G, or satellite without expensive fiber builds or manual configurations.
• Combining Path Border Controller (PBC) hardware with a cloud-based Path Computation Engine (PCE), the platform delivers automated provisioning and end-to-end network visibility.
• Serving service providers, data center operators, and AI infrastructure platforms, the company enables rapid interconnectivity, distributed GPU site networking, and sovereign AI deployment

# # #

Originally posted on SuperbCrew.

MaiaEdge is a networking technology company co-founded in 2024 by CEO Abilash Menon and CRO Tim Ziemer to make private connectivity instantly accessible across any provider’s network. Founded by former leaders behind Acme Packet and 128 Technology, the company provides cloud-managed hardware and cloud-based software that allow service providers, data center operators, and AI infrastructure companies to expand their private network footprint. Its core architecture, SD-Ethernet (Software-Defined Ethernet), automates private Ethernet provisioning over existing transport options such as fiber, internet, 5G, and satellite without requiring custom routing, complex integrations, or expensive fiber construction.

The solution pairs Path Border Controller (PBC) hardware with cloud-based Path Computation Engine (PCE) software to manage automated provisioning while maintaining end-to-end performance visibility across network boundaries. By using hardware like the PBC 100 Series at customer premises, network operators can connect with partner networks in minutes, bypass lengthy construction delays, and offer brand-owned services such as cloud connectivity and AI inference. Unlike Network-as-a-Service providers that own the underlying network and customer relationships, MaiaEdge equips operators with tools to run on their own networks, ensuring they maintain full control over their revenue, service offerings, and clients.

Key use cases span service providers seeking off-net customer expansion, data center operators transforming facilities into interconnection hubs without capital investments, and AI infrastructure operators connecting distributed GPU sites with secure, encrypted connectivity. Moving forward, the organization plans to focus heavily on sovereign AI requirements and expand its provider ecosystem, ensuring GPU clusters can be safely linked with high performance, secure data paths, and rapid deployment capabilities

To read the full interview with Abilash Menon, please click here.

The post How MaiaEdge is Automating Private Connectivity Beyond Traditional Footprints appeared first on Data Center POST.

TL;DR

• Real Estate vs. Compute Trade-Off: India offers substantial savings on land, shell construction, and labor, but imported compute hardware like GPUs receives no local discount and incurs additional import duties, logistics, and currency exposure.
• Cooling and Power Penalties: Hot and humid climate conditions increase mechanical cooling demands and necessitate early liquid cooling adoption, while local grid instability requires heavier investments in power resilience and backup redundancy.
• Strategic Capital Reallocation: Successful operators must reallocate budget away from physical buildings and toward modular design, advanced cooling, and power infrastructure to prevent underfunding critical operational systems.

# # #

Every AI infrastructure budget starts with the same global numbers: roughly $11.3 million per MW for a standard 2026 build, north of $20 million per MW once liquid cooling and high density enter the design, and $45-55 billion per gigawatt at hyperscale. Those figures are accurate, and applying them directly to an Indian project will misbudget it by 30-40%, in both directions. India doesn’t just cost less; it costs differently, because the drivers behind each line item behave very differently than they do in Northern Virginia or Frankfurt.

Getting this right matters more every quarter. India’s operational capacity reached about 1,123 MW of IT load by mid-2025, and roughly 387 MW was added in 2025 alone, more than double the prior year. The build wave is here. The benchmark it’s being budgeted against is the wrong one.

The India numbers, and where they diverge

To answer what it costs to build an AI data center for large-scale model training, across compute, power, cooling, and land, in India specifically, you have to adjust every line of the global benchmark rather than convert it.

Start with the headline. A Tier III colocation build in India costs roughly ₹50-65 crore per MW of IT load, about $6-8 million per MW. Compared with the global shell-and-core average of $11.3 million, that’s the 30-40% discount the market talks about. AI-ready high-density facilities (20-50+ kW per rack with liquid cooling) costs ₹70-95 crore per MW. A 50 MW hyperscale build including land lands around ₹3,000-4,500 crore ($360-540 million), with 100 MW campuses reaching ₹6,000-9,000 crore ($720 million-$1.1 billion).

But the discount is not uniform across the four cost components, and that’s the whole point.

Land and construction: India’s genuine advantage

This is where the 30-40% saving actually materialises. Land is significantly cheaper than in FLAP-D Europe or Silicon Valley, construction labour costs are a fraction of those in Western markets, and the shell-and-core scope has lower input costs across the board. For the physical building, India is one of the most cost-efficient large markets in the world. If your project were nothing but a shell, the global benchmark would overstate your cost by a third.

Compute: where the discount disappears

It isn’t anything but a shell. GPUs and accelerators account for roughly 60% of an AI facility’s capital, yet they receive none of India’s local advantage. Silicon is priced globally in dollars, and India then adds import duties, logistics, and rupee-dollar exposure on top. So the single largest cost bucket in an AI build either matches global pricing or exceeds it, and because it’s the biggest bucket, it drags the whole project’s effective discount well below the 30-40% the shell suggests. Budget compute at global-plus, not India-minus.

Cooling: an India penalty, not a saving

Here the direction reverses outright. India’s hot, humid climate is among the least favourable for data centre cooling. Temperate markets harvest hundreds of free-cooling hours a year from cool ambient air; across most of India, that opportunity is limited, so mechanical cooling runs harder, longer, and at higher energy cost. Water scarcity compounds the problem, limiting evaporative approaches and pushing designs toward liquid- and air-cooled chillers. Globally, mechanical scope has already climbed from about 22% to 33% of AI facility cost; in India’s climate, cooling carries even more weight. What global benchmarks treat as a moderate line item is, in India, one of the largest sources of both capital and operating cost.

Power: the wild card

Power cuts both ways. Electricity and land can be cheaper in the right state, and behind-the-meter and captive generation are increasingly viable. But grid reliability is the offsetting reality: variable supply and voltage instability demand heavier backup, larger UPS energy stores, more generator capacity, and higher redundancy than a stable Western grid requires. Grid-connection timelines and substation availability add schedule costs. Net, an India power train is often engineered for more resilience than the global benchmark assumes, and that resilience isn’t free.

The real insight: a different cost structure, not just a lower price

Put the four together, and the lesson is not “India is 30-40% cheaper.” It’s that India has a different cost structure. Relative to the gln obal template, an Indian AI build spends less on land, shell, and labour and more, proportionally, on imported compute, climate-driven cooling, and power resilience. A budget built by taking a global $/MW figure and shaving 30% off it will be wrong in a specific and dangerous way: it will over-fund the building and under-fund the cooling and power systems that actually determine whether the facility runs. The money doesn’t just shrink; it moves.

What this means for anyone building in India

You cannot copy-paste a global AI data centre design onto Indian soil and expect the economics to hold. Projects that come in on budget are re-engineered for local reality on three fronts. Liquid cooling stops being optional far sooner here, because the climate makes air cooling both costlier and less effective; the cooling investment that looks like a premium is actually the efficiency play. Power resilience has to be designed for the Indian grid from the first line, sized for a synchronous training peak on a supply that is less predictable than the benchmark assumes. And prefabricated, modular construction addresses both the speed the market demands and the shortage of specialised data centre and liquid-cooling skills by moving complexity into the factory.

Operators who treat India’s 30-40% divergence as a discount to pocket will misallocate their budgets. Those who treat it as a signal that the cost structure itself is different, and that the design must follow suit, will build facilities that are genuinely cheaper to own, not just cheaper to quote. In one of the fastest-growing data centre markets on earth, that distinction is the difference between a project that pencils and one that stalls. The global benchmark tells you what a data centre costs elsewhere. India requires you to do the harder work of costing it here.

# # #

About the Author

Sanjay Sahu is Director of Data Center Sales at Schneider Electric, Greater India, where he leads the company’s data center business across the region. With more than two decades in India’s data center and secure-power industry, spanning three-phase UPS, critical power, and end-to-end infrastructure, he advises IT/ITES, BFSI, and government organizations on designing data center solutions that are commercially and technically sound. He writes on data center power strategy and the evolving Indian market.

The post Why Hyperscale AI Data Center Build Costs in India Are 30-40% Different From Global Benchmarks? appeared first on Data Center POST.

TL;DR

• Standard elastomer seal qualifications cover material compatibility, but they ignore surface contamination like deflashing grit, oils, and bioburden.
• Tight thermal resistance budgets in narrowing microchannels make loops vulnerable to partial blockages and thermal excursions.
• Initial commissioning flushes cannot remove debris hidden within seal glands, and outdated filtration standards let fine particles circulate.
• Direct-to-chip liquid cooling systems require standardized incoming seal cleanliness specifications based on absolute particle counts.

# # #

By Bill Sharratt, President, Pacific Rubber & Packing

Take an inventory of the wetted materials in a modern direct-to-chip cooling loop and note which ones carry a published specification.

The cold plate base is specified: typically C11000 electrolytic copper, with a thermal conductivity floor. The surface treatment is specified, commonly electroless nickel at five to eight microns to prevent galvanic attack. The joining method is specified, with vacuum brazing producing a documented burst pressure. The fluid is specified: glycol fraction, pH band, inhibitor package. Contact flatness is specified to hundredths of a millimeter. The leak testing is extraordinary: cold plates in this class are helium tested to better than 10⁻⁸ mbar·L/s on 100 percent of units, then pressure tested at twice operating pressure.

 The elastomer seals are specified too, and that is exactly where the industry has convinced itself the job is finished. Compound compatibility with PG25, with dielectric immersion fluids and with two-phase refrigerants has been characterized exhaustively. Extractables, swell, hardness drift, thermal aging: that work has been done, and it is good work. Material qualification is not the gap.

The gap is that none of it describes the part that actually arrives on the dock.

A seal clears three checkpoints before it enters a loop: material qualification for coolant compatibility, an incoming inspection for gross defects such as excess flash and dimensional non-conformance, and the loop’s filtration rating. Not one of the three looks at what is sitting on the surface of the part.

We looked. Across materials and across manufacturers, what we found sorts into three classes, each needing a different removal method: insoluble particulate such as dirt, dust and carbon black; oils and release films deposited during molding and by human handling; and living bioburden, including bacteria and fungal spores. None of that is the polymer. It is what the polymer picked up on the way to you. Deflashing tumbles molded parts against silica and abrasive media, which leaves fine grit embedded in the surface. Carbon black accumulates throughout processing. The load climbs at every processing step, and then a person picks the part up.

What makes this hard is the variance, not the average. Cleaner and dirtier materials genuinely exist, but the spread across lots, across bags and across manufacturers is wide enough that choosing a material or auditing a supplier does not control it. Within a single bag, the cleanest seal and the dirtiest seal differ by multiple factors. Qualification describes a compound; contamination is a distribution, and it is the tail of that distribution that ends up in a channel.

That channel has less room than it used to. Published design guidance for current-generation cold plates puts the junction-to-coolant thermal resistance budget near 0.030 °C/W, with production units landing between 0.026 and 0.029. That is 10 percent margin or less against GPU thermal design power now measured in thousands of watts per device. Any deposit on a channel wall is series resistance added directly to that budget, and channel geometry is moving from roughly 200 microns toward 100, half the gap for the same particle.

The second-order effect is the one that catches operators out. These cold plates distribute flow across dozens of parallel sub-millimeter channels under a tightly capped pressure drop. Flow follows the path of least resistance, so partial obstruction does not simply add local resistance; it diverts flow to neighboring channels. The affected channel starves, its wall temperature climbs, and deposition accelerates in the channel that can least afford it. What operators see is an intermittent thermal excursion rather than a clean, diagnosable blockage, and it surfaces months before anyone can trace it to a cause.

The usual answer is that commissioning flush handles it. It does not, and the reason is geometric. A seal is designed not to fill its gland, because an overfilled seal extrudes and fails. Every pressurization cycle translates and rotates the seal inside that gland, presenting surface that was never wetted during the initial flush. We call it cyclic contamination displacement. It means the loop keeps receiving new contamination long after handover.

Filtration does not close it either. A meaningful share of seal-borne particulate falls below typical loop filter ratings, small enough to pass through and either foul a microchannel or coagulate over time; on the facility side, filtration is often no more than a 35 to 40 mesh strainer, which is 400 to 500 microns. The larger particles a filter does catch shorten filter life and add maintenance. The guidance we lean on predates the problem: ASHRAE’s Liquid Cooling Guidelines for Datacom Equipment Centers, second edition, calls for filtering particles of 10 microns and larger “or at a minimum of the smallest passage in the system,” and it was published in 2014, before 120 kW racks and current channel geometry existed. The Open Compute Project’s cold plate documents cover fluids, leak detection and serviceability, but not the cleanliness of wetted non-metallics.

None of this is novel engineering. In semiconductor fabs, wetted elastomers are qualified for particle shed as a matter of routine, because a single particle scraps a wafer. The protocols and the measurement methods exist. This industry has never asked for them. What it needs is an incoming cleanliness standard for seals, structured the way ISO cleanroom classifications are: absolute particle counts at defined sizes, applying across materials, manufacturers and geometries. An engineer would specify the class the application requires, then work backwards to material, supplier and cleaning process.

# # #

About the Author

Bill Sharratt is President of Pacific Rubber & Packing, which has supplied precision elastomer sealing to semiconductor manufacturing for three decades.

The post What Elastomer Seals Bring to Liquid Cooling Loops appeared first on Data Center POST.

TL;DR

• Structured Workload and Power Planning: Sustainable growth requires a disciplined seven-step sequence that aligns specific computing needs directly with grid operators and utility interconnection plans.
• Maximizing Facilities and Hardware: Operators can optimize existing capacity by repurposing suitable industrial buildings, adopting direct-to-chip liquid cooling for dense racks, and extending server lifespans to curb electronic waste.
• Transparent Community Impact: Long-term viability relies on publishing measurable environmental data regarding power, water, and noise usage, alongside formal community benefit agreements

# # #

By Jeffrey Gregor, General Manager, OVHcloud US

AI is increasing the need for computing capacity faster than many power systems, supply chains and local permitting processes can accommodate. The 2024 United States Data Center Energy Usage Report, prepared by Lawrence Berkeley National Laboratory for the Department of Energy, estimated that data centers consumed 176 terawatt-hours of electricity in 2023, or 4.4% of all U.S. electricity. Its scenarios put the 2028 total between 325 and 580 terawatt-hours, equal to 6.7% to 12% of U.S. consumption. A June 2026 update from Berkeley Lab extends the outlook to 2030, with a central estimate of 649 terawatt-hours, or 11.8% of U.S. electricity use, and a scenario range of 521 to 843 terawatt-hours (9.5% to 15.3%).

Meeting that demand calls for a disciplined seven-step sequence: define the workload, build a credible power plan, recover capacity in existing facilities, consider adaptable buildings, address cooling and water early, extend hardware life, and make local impacts measurable and public.

The first step is to define the demand precisely. Training a frontier model, running inference and supporting enterprise applications create different requirements for power density, latency, resiliency and location. The 2024 Berkeley Lab report estimated that servers allocated to inference accounted for nearly 60% of AI server energy use in 2023, while training server energy is projected to surpass inference by 2028 as additional higher-power GPUs are allocated to training. Because inference accounts for a large share of AI server energy use and many inference workloads are latency-sensitive, much of that work can run closer to users rather than only in giant centralized campuses. A distributed footprint can also spread demand across markets.

The second step is to make power a siting and operating plan. The Federal Energy Regulatory Commission staff’s 2025 State of the Markets report identifies size, speed of interconnection and load flexibility as defining features of data center demand. Before committing to a site, developers should work with utilities and grid operators to test interconnection queue timing, transmission and substation upgrades, rate design and cost allocation. Where on-site generation or storage is proposed, it should be evaluated with the utility for reliability, emissions, fuel supply, permitting and consumer-cost impacts. Flexible workloads can support phased energization or demand-response agreements, but service commitments and operating controls must be defined up front.

The third step is to inventory what is already available. The Uptime Institute’s 2026 Global Data Center Survey says concerns about capacity forecasting and power availability are growing, while legacy infrastructure and cooling constraints continue to slow efficiency gains. Operators should measure actual IT load against UPS and cooling capacity, identify stranded electrical and thermal headroom, and validate redundancy and network limits before assuming that available floor space can support denser systems.

The fourth step is to consider whether an existing building can do the job. A building with suitable power access, fiber, structural strength and industrial zoning may offer a sound alternative to ground-up construction. Former factories, warehouses and government facilities can sometimes be adapted for data center use, reducing demand for land and building materials. At OVHcloud, our Beauharnois campus in Quebec occupies a converted former aluminum smelter and houses one of our server-production plants. Each candidate still requires a rigorous assessment of floor loads, cooling, backup generation, security and expansion options. Reuse should be treated as a site-selection strategy, with economics and engineering determining where it is feasible.

The fifth step is to design cooling and water use with the site. Uptime Institute’s 2026 survey notes average rack densities continue to rise, and more operators now report peak rack densities of 30 kW or above. Direct-to-chip water cooling carries heat away close to high-density CPUs and GPUs and can support denser racks with less cooling energy. Water efficiency needs a specific design target: a 2026 paper published by the National Academy of Engineering, citing Berkeley Lab’s 2024 report, puts U.S. data center water consumption at 140 to 280 billion liters in 2028 (about 37 to 74 billion gallons). Closed-loop systems can recirculate water and use dry coolers for much of their heat rejection, limiting ongoing water consumption. Cooling architecture also affects energy use, noise, building design and the computing capacity that can fit within an existing footprint, so these choices belong in the first site assessment.

The sixth step is to extend hardware life. The Global E-waste Monitor 2024 found that the world generated 62 million metric tons of electronic waste in 2022; only 22.3% was documented as formally collected and recycled. Extending server life, reconfiguring machines for lighter workloads and harvesting usable components reduce pressure on supply chains and waste systems. Servers designed for disassembly, repair and redeployment can serve several roles over their useful lives. Procurement teams should therefore evaluate serviceability and reuse alongside performance, price and energy efficiency.

The seventh step is to make local impacts measurable and public. Residents and local officials reasonably want to know how a facility will affect electricity, water, traffic, noise, views and public finances. Before approval, developers should publish projected electricity and water use, noise, backup-generator testing and traffic impacts, then report actual site performance after opening. Community benefit agreements can fund local priorities such as workforce training or infrastructure, while heat-reuse projects can serve nearby buildings or industry where a stable off-taker exists. Site selection should account for nearby homes, zoning and visual and acoustic mitigation. Engagement should continue after construction.

Meeting AI demand will take more than new data centers. We need to get more from the power, buildings and equipment already in place, make thoughtful choices about cooling and water, and be candid with communities about each project’s impact. If we do that well, we can add capacity without pushing the costs onto the grid, the environment or the people who live nearby.

# # #

About the Author

Jeffrey Gregor is General Manager of OVHcloud US and President of the Board of Managers of OVH US LLC. He joined the company in 2017 after serving for 20 years as a commissioned officer in the U.S. Navy. He holds a master’s degree in naval architecture and marine engineering from MIT and an MBA from the UCLA Anderson School of Management.

The post The New Playbook for Responsible AI Infrastructure Growth appeared first on Data Center POST.

TL;DR

• Yotta 2026 highlighted the challenges of turning digital infrastructure investment into operational capacity as power availability, construction timelines, and community opposition complicate development.
• Project delays and cancellations underscored the importance of realistic planning, financial discipline, and stronger coordination among developers, investors, and utilities.
• Modular construction and new deployment models emerged as potential strategies for accelerating infrastructure delivery while maintaining scalability and reliability.
• Community engagement and emerging technologies demonstrated how the industry is reconsidering where, how, and at what scale future capacity will be built.

# # #

The digital infrastructure industry has no shortage of ambitious development plans. Turning those plans into operational capacity, however, requires navigating increasingly complex challenges involving energy, financing, construction, and community acceptance.

These priorities shaped discussions at Yotta 2026, held September 28–30 at Caesars Forum in Las Vegas. Bringing together leaders from across the digital infrastructure ecosystem, the event explored how the industry can overcome development constraints, introduce new technologies, and establish more sustainable approaches to supporting growing computing demand.

Community Support Becomes Essential to Project Success

Securing land, power, and financing may be fundamental to data center development, but community acceptance is increasingly influencing whether projects move forward. Concerns about electricity costs, water consumption, land use, and local economic benefits are creating new expectations for developers and infrastructure operators.

The changing relationship between developers and local stakeholders was explored in a discussion featuring Ilissa Miller, Founder and CEO of iMiller Public Relations. The session examined how community engagement strategies must evolve as data center development expands and public scrutiny intensifies.

Rather than treating community outreach as a final step in the approval process, developers face growing pressure to establish relationships earlier, communicate project benefits transparently, and address concerns before opposition escalates. These efforts can influence permitting timelines, investment confidence, and the long-term success of infrastructure projects.

Development Delays Create Industry-Wide Consequences

As infrastructure projects grow larger and more complex, delays and cancellations can create financial and operational consequences extending far beyond individual development sites. Grid limitations, permitting requirements, rising construction costs, and changing market conditions are placing additional pressure on project schedules.

Examining these challenges from an investment and development perspective, Dr. Atif Ansar, Executive Chairman and Co-Founder of Foresight, participated in a discussion addressing the ripple effects of project cancellations and delays. The topic highlighted how setbacks can affect contractors, suppliers, investors, and communities that anticipate economic benefits from infrastructure expansion.

These concerns are particularly relevant as developers pursue increasingly ambitious projects. Greater emphasis on feasibility assessments, risk management, and coordination among stakeholders can help the industry distinguish viable opportunities from projects vulnerable to costly disruptions.

New Construction Models Address the Need for Speed

Traditional construction approaches are being tested by the scale and urgency of today’s infrastructure requirements. Labor availability, equipment lead times, and supply chain constraints are encouraging developers to explore alternatives that can reduce complexity and accelerate deployment.

Moving beyond conventional prefabrication, the discussion of factory-based infrastructure development included Valerie Crafton, Chief Revenue Officer at Mod42. The session explored how standardized manufacturing, modular systems, and repeatable construction processes could provide greater predictability while supporting faster infrastructure delivery.

The need to balance speed with operational and financial discipline also emerged in conversations about infrastructure expansion. Doug Recker, CEO of Duos Technologies, participated in a CEO briefing examining the decisions required to scale through the current investment cycle, including the challenges of aligning market demand with available resources and execution capabilities.

Together, these discussions demonstrated why infrastructure delivery strategies must continue evolving. Flexible construction methods, distributed deployments, and more coordinated development practices offer potential pathways for bringing capacity online while managing costs and operational risks.

Emerging Technologies Challenge Conventional Infrastructure Thinking

Beyond immediate development requirements, Yotta 2026 provided an opportunity to examine technologies that could fundamentally change how and where computing infrastructure operates. Among the more unconventional concepts was orbital computing, with Scott Charter, Head of GTM Data Centers at Physical Superintelligence (PSI), participating in a discussion about the engineering and physical limitations of orbital data centers. The session explored challenges involving power generation, thermal management, and the feasibility of operating computing infrastructure beyond Earth.

Although these concepts remain at an early stage, they reflect the industry’s interest in exploring new approaches to meeting future computing requirements. They also reinforce the importance of evaluating emerging technologies against practical engineering, economic, and operational considerations.

Building Confidence in the Industry’s Future

Innovation and investment alone cannot guarantee successful infrastructure development. The industry’s ability to earn public confidence and demonstrate its economic and community value is becoming equally important.

That challenge was reflected in a discussion featuring Chris Crosby, Founder and CEO of Compass Datacenters, examining whether data centers can rebuild public trust amid increasing scrutiny of infrastructure projects. The conversation addressed the broader responsibilities of developers as they seek to expand capacity while responding to concerns about resource consumption and local impacts.

Across these discussions, a common priority emerged: infrastructure growth must be supported by credible development strategies, stronger stakeholder relationships, and a clearer understanding of the communities in which projects operate.

Turning Ambitious Plans Into Operational Infrastructure

Yotta 2026 demonstrated how closely connected the industry’s technical, financial, and community challenges have become. Power availability influences investment decisions, construction strategies affect deployment timelines, and community relationships can determine whether projects receive the support needed to proceed. As digital infrastructure requirements continue expanding, success will depend on more than announcing new projects or increasing planned capacity. The industry must demonstrate its ability to deliver infrastructure efficiently, responsibly, and in ways that create lasting value.

Continue exploring the technologies, investment strategies, and industry partnerships shaping the future of digital infrastructure at Yotta 2027, taking place September 20–22, 2027 at The Venetian in Las Vegas. Visit the Yotta 2027 website to learn more about next year’s event.

The post Beyond the Building Boom, Digital Infrastructure Faces a Reality Check appeared first on Data Center POST.

TL;DR

• Artificial intelligence (AI) is increasing demand for submarine cable capacity, accelerating interest in new transmission technologies and more efficient network architectures.
• Growing concerns about cable damage, sabotage, and geopolitical tensions are making subsea security and infrastructure resilience increasingly important.
• The relationship between submarine networks, cable landing stations, and data centers is evolving as operators seek more direct and efficient connectivity.
• Changing investment priorities, public-private partnerships and new cable projects are influencing the future of global subsea connectivity.

# # #

The world’s submarine cable systems carry the international data traffic that keeps businesses, cloud platforms, and digital economies connected. As artificial intelligence (AI) drives greater demand for bandwidth and governments place increased emphasis on protecting critical infrastructure, the submarine cable industry is navigating a period of technological change, investment opportunities, and growing security concerns.

These priorities were reflected at Submarine Networks World 2026, held September 23–24 in Singapore. The event brought together telecommunications providers, network operators, investors, technology developers, and policymakers to examine the future of global subsea connectivity with discussions spanning AI-driven capacity requirements, cable protection, infrastructure financing, and the evolving relationship between submarine networks and data centers.

AI Demand Accelerates the Need for Greater Network Capacity

The rapid expansion of AI applications is placing new demands on global networks, increasing pressure on submarine cable operators to deliver greater capacity and efficiency. As data-intensive workloads expand across geographic markets, the industry is evaluating technologies that can support higher transmission volumes while maximizing existing infrastructure investments.

The impact of AI on subsea networks was explored by Walid Wakim, Chief Technology Officer of Systems at Nokia, while discussions featuring Amit Malik, Vice President, APJI Sales at Ciena, examined the growing capacity requirements facing global networks. Topics included the implications of AI-driven traffic growth and the technologies needed to support future connectivity demands.

Additional discussions addressed petabit-scale connectivity, multicore and hollow-core fiber technologies, and the commercial viability of existing cable systems. These developments reflect the industry’s efforts to increase capacity while managing the costs and technical complexities of deploying and maintaining subsea infrastructure.

Subsea Security Becomes a Global Infrastructure Priority

Submarine cables are increasingly recognized as critical infrastructure requiring stronger protection against accidental damage, deliberate interference, and geopolitical threats. With international communications and economic activity dependent on these systems, cable resilience has become an important consideration for governments, operators, and infrastructure investors.

The importance of coordinated security strategies was examined by industry and government representatives, including Kent Bressie, Partner at HWG LLP and Legal Adviser to the International Cable Protection Committee (ICPC). Discussions explored how public and private organizations can align responsibilities, improve monitoring, and translate security policies into practical protection measures.

The agenda also addressed threats ranging from fishing activity and maritime incidents to potential sabotage, along with vulnerabilities at cable landing points. These considerations reinforce the need for improved situational awareness, international cooperation, and coordinated response strategies to strengthen the resilience of global communications networks.

Data Centers and Submarine Networks Become More Closely Connected

As hyperscale cloud providers and AI infrastructure operators expand globally, the relationship between submarine cables and terrestrial data center networks is changing. Traditional cable landing station models are being reconsidered as organizations seek to reduce latency, improve network efficiency, and connect international capacity more directly to computing infrastructure.

The evolving role of these facilities was explored through discussions featuring Mike Cunningham, CEO of Crosslake Fibre, who examined whether direct connections between submarine cables and data centers could become a more prominent deployment model. Other sessions addressed neutral cable landing stations, hybrid landing center architectures, and the growing connectivity requirements of neocloud providers.

These discussions highlighted the importance of coordinating submarine and terrestrial infrastructure development. As demand for distributed computing and high-capacity connectivity increases, greater integration between cable systems, landing facilities, and data centers could influence future network design and investment strategies.

Investment and Partnerships Shape the Next Generation of Cable Projects

Building and maintaining submarine cable systems requires substantial capital, long-term planning, and coordination among governments, telecommunications providers, and private investors. As demand for international connectivity grows, the industry is evaluating how investment models and partnerships can support new projects while managing financial and operational risks.

The changing investment landscape was examined by Matthew Goh, Director at KKR, during discussions exploring private-sector approaches to submarine cable investment. Topics included evolving investor priorities, supply-and-demand dynamics, and the factors influencing the commercial viability of new infrastructure projects.

The event also featured international cable initiatives and public-private partnership strategies, particularly across the Indo-Pacific region. As countries seek to strengthen digital sovereignty and network resilience, collaboration among governments, investors, and industry stakeholders will remain important to expanding secure and reliable global connectivity.

Securing the Future of Global Connectivity

Submarine Networks World 2026 highlighted how the industry’s responsibilities are expanding alongside the demand for international connectivity. Increasing capacity alone is no longer sufficient. Operators and infrastructure partners must also address security risks, evolving network architectures, and the financial requirements of building resilient subsea systems.

The discussions reinforced the importance of balancing technological innovation with long-term infrastructure protection and investment. As AI adoption accelerates and digital economies become more interconnected, submarine networks will remain essential to supporting reliable communications and the continued growth of global digital infrastructure.

Continue exploring the developments shaping global subsea connectivity at next year’s Submarine Networks World 2027, taking place September 29–30, 2027 in Singapore alongside Subsea Security World.

The post From Ocean Floors to Data Centers, Connectivity Strategies Are Evolving appeared first on Data Center POST.

• Award recognition: Duos Edge AI wins Best Data Centre / Edge Service Innovation at the 2026 Global Connectivity Awards.
• Local access: Modular deployments support low-latency computing for education, healthcare and AI applications in underserved communities.
• Industry momentum: The honor follows Duos’ Outstanding Innovation Award at PTC’26, adding to a year of recognition for its distributed infrastructure approach.

# # #

Duos Technologies Group (Nasdaq: DUOT), working to expand access to digital infrastructure through its subsidiary Duos Edge AI, has earned recognition at the 2026 Global Connectivity Awards. The Best Data Centre / Edge Service Innovation award acknowledges an approach centered on placing enterprise-grade computing closer to organizations that have historically lacked local access to it.

As an operating subsidiary of Duos Technologies Group, Duos Edge AI develops modular edge data centers that can grow alongside regional demand. These facilities give schools, healthcare providers and businesses access to low-latency computing for applications such as digital learning, telemedicine and real-time AI processing, without depending solely on distant hyperscale cloud facilities.

“Winning this award underscores the importance of bringing digital infrastructure closer to the communities and organizations that depend on it,” said Doug Recker, CEO of Duos Technologies Group, Inc.

The recognition highlights both the infrastructure and its intended impact. Duos Edge AI’s modular model is designed to make high-performance computing more accessible while allowing capacity to expand as local needs develop. Its deployment strategy emphasizes repeatability and capital efficiency, with a focus on rapid installation and resilient operations.

Earlier this year, Duos received the Outstanding Innovation Award at PTC’26 for advancing digital equity, efficiency and scalable connectivity through its modular Edge AI infrastructure. The latest honor builds on that acknowledgment as the company works to expand its platform and help local organizations adopt emerging technologies and strengthen their services.

To learn more about Duos Edge AI, visit duosedge.ai.

To learn more about Duos Technologies Group, visit duostechnologies.com.

The post Duos Edge AI Earns Global Recognition for Bringing Compute Closer to Communities appeared first on Data Center POST.

TL;DR:

• AI data center design choices directly influence construction costs, operating expenses and financing risks.
• Decisions involving chip density, power architecture and modular infrastructure affect how lenders and investors evaluate projects.
• Developers who prioritize cost efficiency and infrastructure flexibility can strengthen their financing prospects and support long-term profitability.

# # #

Artificial intelligence (AI) is changing how data centers are designed and built. As AI workloads grow more demanding, facilities need greater computing density and higher electricity capacity. These requirements directly influence construction costs, operating expenses and long-term financial performance.

Infrastructure design is also important to data center financing. Lenders and investors must evaluate how technical decisions affect project risks, borrowing capacity and financing terms. Their assessments can determine how much debt developers can secure and whether additional financial protections are necessary.

Why AI Infrastructure Design Matters to Data Center Financing

Data center financing depends on predictable construction costs, steady operating income and long-term asset value. AI infrastructure can complicate each of these factors because higher rack densities, specialized cooling systems and expanded power requirements often increase both initial development costs and long-term operating expenses.

Rapidly changing hardware requirements can also create risks of technological obsolescence. As a result, data center design affects how lenders evaluate projects. They assess whether a facility can generate enough stable cash flow to cover operating expenses and debt payments. Projects with higher design-related risks may support less debt, requiring developers to contribute more equity or provide additional financial guarantees.

How Chip Density Changes Construction Costs and Financial Risk

Powerful AI accelerators allow data centers to process more complex workloads within individual server racks. However, greater computing capacity also generates more heat and increases electricity demand. To support higher rack densities, operators may need upgraded electrical distribution systems and specialized cooling equipment.

These requirements can significantly increase construction costs. Future chip generations may also introduce different power and cooling needs, forcing operators to invest in expensive infrastructure upgrades. This uncertainty creates financial risks, particularly for facilities designed around specific hardware configurations. Projects with limited upgrade flexibility may face stricter loan terms or higher equity requirements to account for technological obsolescence.

Cooling Technology and Its Impact on Long-Term Operating Costs

Traditional air cooling uses fans and airflow systems to remove heat from servers. However, high-density AI computing often requires more efficient solutions. Direct-to-chip liquid cooling circulates coolant through cold plates attached to processors, while immersion cooling submerges equipment in specialized cooling fluids.

A 2025 S&P Global survey found that 21% of data center decision-makers planned to shift to liquid cooling, with another 25% expecting to transition within two to four years. Although these systems may require higher up-front investments, they can improve cooling efficiency and reduce certain operating expenses.

Cooling choices also affect data center financing. Future hardware may require different cooling configurations, creating expensive upgrade risks. Lenders therefore examine projected energy costs, maintenance expenses and infrastructure flexibility to determine whether a facility can maintain reliable cash flow and meet debt obligations.

What Lenders and Investors Look for in AI Data Center Projects

Lenders and investors assess construction budgets, projected operating expenses and long-term asset value when evaluating data center financing requests. As the market has grown, financing structures have also expanded. Since 2018, U.S. data center issuers have used asset-backed securities and commercial mortgage-backed securities structures across 88 transactions, representing over $48 billion in issuance and underscoring how large these capital needs have become.

They also examine whether expected revenue can cover operating costs and debt payments. Tenant creditworthiness matters because financially stable customers with long-term leases can provide more predictable cash flow. Facilities that can accommodate future AI hardware may face lower risks of technological obsolescence, while highly specialized systems could require expensive upgrades.

Projects with greater financial or technical uncertainty may qualify for less debt. To secure financing, developers may need to contribute additional equity, maintain larger financial reserves or obtain stronger tenant guarantees.

Power Architecture and the Financial Risks of Energy Availability

AI workloads require high-capacity power distribution and backup systems to support continuous computing operations. For example, a hyperscale AI data center with 100 megawatts of capacity can consume as much electricity annually as approximately 100,000 households, depending on utilization. This enormous demand makes power availability, utility infrastructure and electricity costs major considerations in data center financing.

Electrical infrastructure also influences construction budgets and project timelines. High-capacity transformers and backup generators require substantial investments, while grid connection delays can postpone operations and revenue generation. Insufficient power capacity and fluctuating electricity prices may further weaken profitability. Projects facing greater power-related uncertainty may require additional financial reserves or more developer equity.

How Modular Design Can Reduce Obsolescence and Financing Risk

Modular data center design allows developers to expand or upgrade electrical and computing infrastructure in stages. Rather than building an entire facility up front, operators can add capacity as customer demand grows. This phased approach helps reduce initial capital requirements and limits spending on unused infrastructure.

Modular systems also offer greater flexibility as AI hardware evolves. Operators can upgrade individual components to support new computing and power requirements without rebuilding entire facilities. This adaptability can lower future reconstruction costs and reduce the risk of technological obsolescence.

From a financing perspective, phased investments can help developers align construction spending with expected revenue. However, lenders still assess integration costs, expansion timelines and potential compatibility issues. These factors influence how lenders evaluate project risks and determine financing terms.

Designing AI Data Centers With Financing in Mind

AI infrastructure design decisions carry financial consequences throughout a data center’s operating life. Chip density, cooling technology and modular construction influence capital requirements and the risk of technological obsolescence. By aligning these decisions with lender expectations, developers can improve data center financing flexibility and support stronger long-term asset performance.

# # #

About the Author

April Miller is a Technology writer with more than 5 years of experience writing on AI applications that help SMBs compete and Enterprises strategize. You can connect with her on LinkedIn.

The post How AI Design Choices Shape Data Center Financing appeared first on Data Center POST.

• Long-term customer commitment: Duos signed a 10-year, 20 MW colocation agreement worth more than $400 million, with deployment expected in Q2 2027.
• Expanding the development pipeline: The company is pursuing approximately 100 MW of additional powered, data center-zoned land, including 25 MW under contract and 75 MW in active negotiations.
• Continued momentum: This announcement builds on Duos’ Columbus, Georgia expansion.

# # #

Duos Technologies Group (Nasdaq: DUOT) announced this week a 10-year colocation agreement covering 20 megawatts of facility load for an established, publicly traded U.S. cloud services provider. The agreement represents more than $400 million in total contract value over its initial term, with deployment expected in the second quarter of 2027 across Duos’ growing Edge AI campus portfolio.

This announcement combines contracted demand with a pipeline of potential sites, a pairing central to Duos’ contract-first development model. Rather than building speculative inventory and then seeking tenants, the company works to establish customer demand before committing significant capital to a location. The latest agreement provides a long-term commercial commitment as Duos pursues additional capacity for AI, cloud, content delivery and enterprise computing workloads.

Alongside the customer agreement, Duos is pursuing approximately 100 MW of additional powered, data center-zoned land across multiple U.S. sites. Approximately 25 MW is currently under contract, while another 75 MW remains in active negotiations. The company expects to finalize all site agreements by year-end, pending due diligence and other factors.

CEO Doug Recker framed the agreement and land expansion efforts as complementary steps in the company’s growth strategy. Its modular architecture is designed to support scalable, power-dense infrastructure, while the contract-first approach ties development decisions to customer requirements. Together, these elements underpin Duos’ efforts to expand its customer base and contracted revenue.

This expansion comes as power availability increasingly shapes data center development. As power constraints and lengthy transmission-development timelines have increasingly become obstacles in desirable markets, cloud providers are seeking to secure power ahead of future demand. Against that backdrop, Duos’ combination of powered, appropriately zoned sites, modular infrastructure, and contracted customers addresses closely linked industry priorities: securing suitable locations, aligning investment with demand and shortening the path to deployment.

To learn more about Duos Technologies Group, visit duostechnologies.com.

The post Duos Technologies Group Advances Edge AI Expansion With 20 MW Customer Agreement and Plans for Additional Sites appeared first on Data Center POST.

TL;DR

• Growing community rebellion: Mounting public backlash against AI data centers over power, water, and land use is turning sustainability into an urgent license-to-operate challenge.
• Educational transparency over PR: To protect proprietary business secrets while building local trust, developers must move beyond generic PR to clearly explain technology categories and tradeoffs.
• Distinguishing efficiency from total demand: Communicators must clarify that efficiency gains often support higher compute densities rather than reducing a facility’s overall energy or resource consumption.
• Proactive communications readiness: PR teams must partner early with engineering, legal, and executive leadership to prepare credible, context-rich answers before introducing projects publicly.

# # #

By Lisa Ann Pinkerton, CEO and Founder of Technica Communications

Earlier this year, The Wall Street Journal described the growing pushback against artificial intelligence as an “American rebellion against AI.” In the months since, that backlash has become even more tangible for the data center industry.

Communities across the country are scrutinizing proposed facilities over electricity costs, water use, pollution, land use, jobs and who ultimately benefits from the massive investments required to support AI. Some local governments have paused new data center development while they reconsider the rules governing these projects, and opposition is increasingly moving from public meetings into organized campaigns.

For data center developers, this is no longer simply a sustainability challenge. It is becoming a license-to-operate challenge.

But you don’t quell a rebellion with PR.

Sustainability Is Important, But It Is Not Self-Explanatory

Many data center developers are making serious investments in better infrastructure. The industry is advancing more efficient and sustainable systems and more sophisticated approaches to facility design. These innovations matter, and in many cases, they can materially reduce a project’s environmental or community impact.

And while sustainable infrastructure may be one of the most important ways they earn or regain their license to operate, simply telling people it’s better for water or energy use is not sufficient. Communities must easily understand what is being used, why it was chosen, how it changes the facility’s impact, and how it could benefit their region.

The challenge is that most of these technologies are not easily understood outside the industry, simply because the industry has never had to communicate them before. A cooling system that dramatically improves thermal performance may sound impressive to an operator. However, a nearby resident may still want to know whether the facility will draw from local water supplies.

This is why sustainability claims are no longer enough. The public is not simply asking whether a data center is efficient. Communities are asking how that efficiency translates into shared benefit. They want to know the real-world results. And if the industry treats local questions as uninformed or hostile, it risks deepening the mistrust it needs to overcome.

Data Centers Have a Real Confidentiality Challenge

The obvious recommendation is often to be more transparent, and in many ways that is correct. However, transparency in this sector is not as simple as publishing every detail.

During interviews for the Earthlings 2.0 Data Center series, I was struck by how many developers, hyperscalers, and operators viewed infrastructure decisions as closely guarded business assets. From cooling and backup power systems to vendor selection, energy strategies, facility design, and operating models, these choices are often seen as sources of competitive advantage that companies are understandably reluctant to disclose publicly.

These infrastructure decisions are not merely technical details. They are strategic assets that influence deployment speed, compute density, operational performance, and competitive positioning. That creates a legitimate communications tension. Communities deserve enough information to understand whether a project is responsible, beneficial, and accountable. Companies, at the same time, require enough confidentiality to protect commercially sensitive systems and maintain their competitive advantage. Pretending this tension does not exist hurts everyone.

A more balanced approach is educational transparency rather than total disclosure. Data center companies do not necessarily need to reveal every proprietary design decision to communicate responsibly. They get a chance to explain the categories of systems being used, the purpose those systems serve, and the tradeoffs involved. Such information will help communities be better informed when evaluating the safeguards in place and the outcomes the region can reasonably expect. This approach respects both the company’s need for confidentiality and the community’s need for trust.

Efficiency Requires Clear Explanation

One of the most important nuances in this conversation is that a more efficient data center does not necessarily use less total energy. In the context of AI, efficiency can often enable greater computing power. In other words, a facility may do more work with the same amount of energy. From an operational standpoint, that can be a meaningful improvement. From a community standpoint, however, the total demand on the power grid, land, and water may remain significant.

This is where data center sustainability messaging can become vulnerable. If a company implies that efficiency automatically means lower overall resource use, and the community later learns that the facility is simply using those gains to support more compute, the message can feel misleading even if the technology itself is valuable. Efficiency is not the problem. Overstating what efficiency means is the problem.

A more credible communications model would distinguish between resource intensity and total resource demand. A data center may reduce energy use per unit of compute while still increasing overall energy consumption as workloads grow. Similarly, a cooling system may use less water than alternatives while still creating a water demand profile that communities deserve to understand.

This level of explanation may feel uncomfortable for companies used to communicating in technical or investor-oriented language. But it is increasingly necessary. Communities do not need simplified claims that obscure complexity. They need clear explanations of how projects affect their communities.

Communications Readiness Matters More Than Speed

For public relations (PR) professionals, this moment requires more than starting communications early. It requires communications readiness before a project is ever introduced publicly.

By the time a company is seeking approvals for a new AI data center, every stakeholder will have questions about water use, electricity demand, emissions, jobs, land use, reliability, and community benefits. If leadership is still developing answers as those questions arise, communications become reactive rather than strategic.

That is why PR teams should work alongside engineering, operations, permitting, legal, and executive leadership well before a project enters the public eye. The goal is not simply to draft messaging. It is to understand the rationale behind key infrastructure decisions, identify where legitimate concerns are likely to arise, and build messaging that explains both the opportunities and the tradeoffs in language non-technical audiences can understand.

Communications cannot solve underlying business or infrastructure challenges. PR cannot make a water-intensive project water-light, turn a vague jobs promise into a workforce development plan, or eliminate legitimate concerns about power demand. What it can do is ensure companies are prepared to answer difficult questions with consistency, context, and credibility rather than improvising under public pressure.

The goal is not to persuade communities to accept every data center project. It is to ensure companies enter the conversation prepared—with messaging that is accurate, transparent, and grounded in the realities of the project rather than developed in response to controversy.

The New License to Operate

The Wall Street Journal’s coverage of an “AI rebellion” is useful because it signals how quickly public sentiment around AI is changing. But the more important story for the data center industry is that communities are beginning to connect AI’s lofty or apocalyptic promises with its real physical demands on their regions and their people. That means the license to operate data centers will increasingly depend on factors beyond technical performance, capital investment, and broad sustainability language. It will depend on whether companies can explain their infrastructure choices in a way that is credible, understandable, and connected to community benefit. For communicators, the challenge is no longer simply translating technical performance into marketing messages. Increasingly, it is translating infrastructure decisions into outcomes that communities can understand and evaluate.

# # #

About the Author

From the newsroom to the boardroom, Lisa Ann has used her keen analytical skills to share technology stories with the world since 2001. She is the founder and CEO of the award-winning Technica Communications, founder and CEO of the PR software platform PRVIEW, founder of the non-profit Women In Cleantech & Sustainability, host of Earthlings 2.0 and an international speaker, moderator and documentary filmmaker. She was named a 2026 PRNEWS Top Woman in PR, the PRNEWS 2021 Industry Innovator, PR Executive of the Year by the American Business Awards (2020), Female Entrepreneur of the Year for Advertising and Marketing by the Women in Business and the Professions World Awards (2020), and a Woman of Influence by the Silicon Valley Business Journal (2017). She got her start as a broadcast journalist covering environmental science stories for NPR and PBS for over a decade.

The post AI Data Centers Must Go Beyond Sustainability Claims to Win Back Their License to Operate appeared first on Data Center POST.

TL;DR

• Singapore’s EDB and IMDA provisionally allocated 200 MW of new data centre capacity equally among Digital Realty, Equinix, Keppel Data Centres, and ST Telemedia Global Data Centres under the DC-CFA2 application outcome.
• The selection marks a shift from capacity expansion to national value creation, requiring operators to commit to over 50% green energy usage and advanced liquid cooling deployments.
• Winning proposals were evaluated on their alignment with national priorities, such as strengthening Singapore as an AI hub, fostering local R&D, and building talent ecosystems.
• Physical layer infrastructure must be engineered for long-term scalability, accommodating higher fibre density, low optical loss, and technology refreshes reaching 1.6T network speeds

# # #

Why the four selected operators reveal a shift from capacity expansion to national value creation – and what it means for the physical infrastructure behind AI-ready data centres.

On 21 August 2026, Singapore’s Economic Development Board (EDB) and Infocomm Media Development Authority (IMDA) announced the outcome of the second Data Centre – Call for Application (DC-CFA2).¹ From more than 20 proposals, Digital Realty, Equinix, Keppel Data Centres and ST Telemedia Global Data Centres were selected. Each has been provisionally allocated 50 MW of new capacity, creating a combined total of 200 MW.¹

At first glance, this appears to be another allocation of data centre capacity. The evaluation priorities and the capabilities represented by the four selected operators, however, point to a broader change in Singapore’s policy: the competition is no longer only about who can build efficiently, but who can convert constrained resources into the greatest strategic and economic value.

For infrastructure providers, that shift has direct consequences. Higher rack densities, liquid-cooled environments and the move from 400G and 800G towards 1.6T demand a physical layer designed for higher fibre density, lower optical loss and continuous technology upgrades.

What DC-CFA2 Asked Operators to Deliver

DC-CFA2 was launched on 1 December 2025 to support continued growth in Singapore’s digital economy and AI infrastructure while keeping resource use under control. Successful proposals had to address three linked objectives:¹

• Strengthen Singapore’s position as a trusted hub for AI, data centres and international digital connectivity, while improving infrastructure resilience.
• Create economic contributions beyond the data centre investment itself, including R&D, technology innovation, talent development and collaboration with Singapore-based organisations.
• Present a credible sustainability plan that can be implemented in practice.

The four selected operators committed to exceeding the minimum sustainability requirements, including powering more than 50% of their data centre capacity with green energy, deploying advanced technologies such as liquid cooling, using highly energy-efficient IT equipment and meeting Green Mark Data Centre Platinum requirements.¹

Eligible green-energy pathways include biomethane, low-carbon ammonia, low-carbon hydrogen, fuel cells with carbon capture, and building-integrated or building-applied photovoltaics.^{1, 2} Together, these requirements show that DC-CFA2 is not simply comparing investment size, data hall area or delivery speed. It is evaluating the wider value each proposal can deliver to Singapore.

The Scarce Resource Is Every Megawatt

Singapore faces an unusually tight set of development constraints. Demand for AI, cloud computing and digital services is growing rapidly, yet land, electricity and water are limited, and expansion must remain aligned with national decarbonisation goals.² Singapore therefore cannot follow the model of much larger markets by continuously adding land and power supply.

“The key question is no longer how much data centre capacity Singapore can build, but how much value each constrained megawatt can create,” which is a policy implication derived from DC-CFA2’s published priorities.

Low PUE, financial strength, reliability and construction capability remain essential, but they increasingly look like entry requirements rather than final differentiators. A competitive proposal must also connect new capacity to advanced computing, international connectivity, local R&D, talent development and credible low-carbon energy outcomes.

What the Four Selected Operators Reveal

EDB and IMDA did not disclose individual applicant scores or the full content of each proposal. The comparison below is therefore an interpretation based on published evaluation priorities and publicly available information – not a claim about the authorities’ undisclosed reasons for selection.

How the Rules of Competition Are Changing

1. Building capability is the baseline

Construction capability, financial strength, PUE, reliability and delivery speed remain essential. The differentiators are increasingly the additional outcomes a project can deliver: deployed computing capacity, local R&D, talent development, international connectivity, energy innovation and measurable long-term economic value.

2. Energy strategy begins at the design stage

Energy can no longer be treated simply as a utility connection added after the facility is built. Sources, storage, microgrids, low-carbon fuels, load management and thermal utilisation must be considered from the earliest stages. Green-energy claims will also need to be specific, scalable and genuinely deliverable.

3. Efficiency must extend beyond PUE

PUE measures the relationship between total facility energy and IT-equipment energy, but it does not show how much useful computing output is produced. Future assessments are likely to consider the efficiency and utilisation of servers and accelerators, the workload value created, and the digital output supported per unit of energy, water and carbon.

4. Cooling architecture must match the workload

The future is unlikely to be a simple replacement of air cooling with liquid cooling. High-density GPU systems may require direct liquid or immersion cooling, while network equipment, storage and lower-density devices may remain air-cooled. Hybrid cooling architectures will need to coordinate different heat densities within the same facility.⁵

5. The physical layer must remain upgradeable

AI chips, network speeds and rack power densities can evolve faster than the building itself. Data centres therefore need high-density fibre connectivity, low-loss optical links, scalable structured cabling architectures that can evolve from 400G and 800G to 1.6T and beyond, and cable pathways compatible with liquid-cooled equipment.

6. Data centres are becoming part of an industrial energy system

Locating the projects on Jurong Island creates opportunities to connect data centres with low-carbon fuels, shared utilities, energy storage, microgrids and thermal systems.^{2, 4} The long-term model is not four isolated low-carbon buildings, but a wider ecosystem in which digital infrastructure, energy and industry support one another.

What This Means for Data Centre Physical Infrastructure

The shift identified in DC-CFA2 reaches all the way down to the physical layer. If a facility is expected to accommodate changing compute platforms and cooling methods over many years, cabling and connectivity can no longer be designed only around today’s equipment layout.

• Fibre density must scale without creating unmanageable congestion inside racks and pathways.
• Optical performance must support higher speeds and tighter loss budgets as networks progress towards 800G and 1.6T.
• Cabling routes, patching systems and rack layouts must coexist with liquid-distribution infrastructure and mixed cooling architectures.
• Modularity and clear migration paths are essential to avoid disruptive physical-layer rebuilds during technology refreshes.
• Documentation, testing and lifecycle management become increasingly important as connectivity environments grow denser and more complex.

For cabling and connectivity providers the implication is clear: the physical layer must be engineered not merely for current port counts, but for the next several generations of compute, cooling and network architecture.

From Capacity Allocation to Value Creation

DC-CFA2 does not remove the fundamental requirements for power, cooling, connectivity and reliability. It changes the level at which projects must compete. A 50 MW allocation is not, by itself, the final source of value. What matters is whether that capacity can be converted into advanced computing, international connectivity, local innovation, talent development and credible low-carbon outcomes.

The four selected operators each illustrate a different capability: global platform reach, enterprise interconnection, energy integration and AI-ready operations. Together, they suggest that future data centre competition will extend across the entire technology, energy, economic and industrial ecosystem.

For companies planning the next generation of data centres, the decisive question is no longer simply how many megawatts can be built, but what irreplaceable value those megawatts can create – and whether the underlying physical infrastructure can continue supporting that value as technology evolves.

# # #

About the Author

XinYue leads digital marketing and content at Hermesys, a Singapore-based provider of fibre optic, structured cabling and data centre infrastructure solutions across Southeast Asia. She writes about data centre development, digital infrastructure policy and the physical-layer technologies behind AI-ready facilities.

Sources

• [1] IMDA – Four Data Centre Proposals Selected as Part of DC-CFA2
• [2] JTC – Why Singapore Is Building Data Centres on Jurong Island
• [3] Digital Realty – Partnership with BW Digital on Singapore-Batam Connectivity
• [4] Keppel – MOU with JTC on Low-Carbon Energy Infrastructure
• [5] STT GDC – The Evolution of Data Centre Cooling for AI Workloads
• [6] Equinix – Singapore Data Centres and Interconnection Ecosystem

The post Singapore DC-CFA2: The New Rules of Data Centre Competition appeared first on Data Center POST.

TL;DR

• AI-driven growth is intensifying competition for power, land, and capital, creating new challenges for digital infrastructure development.
• Increasing AI training and inference demands are driving changes in data center design, power infrastructure, and network connectivity.
• Fiber investment and consolidation are reshaping the connectivity landscape as demand for high-capacity networks continues to grow.
• Community opposition, cybersecurity concerns, and geopolitical considerations are influencing how digital infrastructure is developed and expanded.

# # #

The future of digital infrastructure is being determined by more than technological innovation and investment capital. As artificial intelligence (AI) drives the need for new facilities, energy resources, and connectivity, questions surrounding public acceptance, national security, and the concentration of infrastructure ownership are becoming increasingly difficult to ignore.

At PTC DC 2026, held September 17 in Reston, Virginia, industry leaders examined the economic and geopolitical implications of AI expansion alongside the practical realities of infrastructure development. From the changing investment landscape for fiber networks to growing community resistance and the emergence of sovereign AI, the discussions highlighted how business decisions, public policy, and local priorities are influencing the industry’s direction.

AI Growth Is Reshaping Infrastructure Investment and Development

The scale of AI investment is transforming the digital infrastructure landscape as developers compete for power, land and capital. Increasing demand for computing capacity is placing greater pressure on energy systems while raising questions about the industry’s ability to deliver new infrastructure at the speed and scale required.

These challenges were examined by Andy Power, President and CEO of Digital Realty, and Gio Albertazzi, CEO of Vertiv, who participated in discussions addressing the global competition for infrastructure resources and the requirements of supporting AI workloads. Topics included capital concentration, power availability, facility design, and the growing need for reliable, high-capacity infrastructure in both established and emerging markets.

As AI training and inference applications evolve, successful development strategies will require greater coordination among technology providers, utilities, infrastructure developers, and policymakers. Long-term competitiveness will depend on securing resources while building facilities capable of adapting to changing computing requirements.

Fiber Connectivity Remains Essential to AI Expansion

While data centers and power infrastructure receive significant attention, fiber networks remain fundamental to supporting AI growth. High-capacity connectivity enables data movement between computing environments and supports the increasingly distributed infrastructure required for training, inference, and mission-critical applications.

The evolving investment landscape for fiber was explored by Ben Edmond, CEO and Founder of Connectbase, who joined industry leaders examining the sector’s consolidation activity and changing capital market dynamics. Discussions highlighted the role of fiber as a critical component of digital infrastructure, even as investment trends differ from those affecting towers and data centers.

As AI applications place greater demands on network performance, continued investment in fiber infrastructure will be essential to connecting computing resources, expanding geographic reach and supporting reliable digital services.

Community Trust Is Becoming a Critical Infrastructure Requirement

Growing public scrutiny of data center development is creating additional challenges for infrastructure expansion. Concerns surrounding electricity consumption, land use, environmental impacts, and local economic benefits are contributing to community opposition and greater pressure on developers to demonstrate the value of proposed projects.

The importance of rebuilding public confidence was addressed by Buddy Rizer, Executive Director for Economic Development at Loudoun County, Virginia, during discussions examining the factors driving community resistance. Topics included aligning development with local priorities, improving communication, and establishing greater trust between infrastructure providers and the communities they serve.

These considerations reinforce the importance of engaging local stakeholders early in the development process. As projects increase in scale, transparent communication and measurable community benefits will play a greater role in securing support and advancing infrastructure development.

Global AI Expansion Raises New Security and Policy Challenges

As AI technologies become more deeply integrated into critical industries and government systems, the need for secure, resilient infrastructure is growing. Expanding AI capabilities globally introduces new considerations surrounding national competitiveness, cybersecurity, intellectual property protection, and sovereign AI initiatives.

The intersection of technology development and geopolitical policy was explored by Seth Center, Former Acting Special Envoy for Critical and Emerging Technologies at the U.S. Department of State, who participated in discussions examining how AI can expand globally without compromising security or operational integrity. These challenges highlight the need for greater coordination among infrastructure operators, technology providers, and policymakers to support innovation while addressing emerging risks.

Building a More Resilient Digital Infrastructure Future

PTC DC 2026 reinforced the importance of collaboration across the digital infrastructure ecosystem as the industry navigates AI-driven growth, power constraints, connectivity demands, and community expectations. Addressing these interconnected challenges will be essential to developing the reliable, scalable, and secure infrastructure required for the next phase of AI adoption.

The challenges highlighted at PTC DC 2026 demonstrate how closely AI growth, infrastructure investment, connectivity, and community engagement are intertwined. As the industry works to address these priorities, the focus now shifts to PTC’27, January 17 – 20, 2027, in Honolulu, Hawaii where global digital infrastructure leaders will gather to examine the opportunities and challenges shaping the year ahead.

The post Building More Isn’t Enough to Meet AI’s Infrastructure Demands appeared first on Data Center POST.

TL;DR

• The grid is a structural bottleneck: US interconnection queues hold over 2,600 GW of projects with median wait times of 4.5 years, stretching to 8 to 12 years for large-scale data centers.
• Lessons from mobile electrification: Data center operators face the exact power delays that EV fleet managers experienced a decade ago, where relying on promised utility timelines stalled operations.
• Immediate bridge power deployment: On-site solutions like Battery Energy Storage Systems (BESS) and microgrids can be commissioned in weeks to bring compute capacity online without waiting for formal grid connections.

# # #

By Josh Aviv, CEO and Founder of SparkCharge

The data center industry is facing a power problem that looks familiar to me.

For the past decade, fleet operators, autonomous vehicle companies, and logistics hubs have sought solutions at the intersection of commercial energy demand and grid infrastructure restraints. The grid has been unable to keep up with the pace at which businesses need power. These industries cannot wait years for utility connections.

Data center developers are now standing at the same crossroads. The timelines are longer, the loads are larger, and the financial stakes are higher. But the fundamental problem is identical: enterprises need power now, and the grid delivers it in years.

The lessons the mobile energy industry learned the hard way over the past decade are directly applicable to what data center operators are navigating today.

The Grid Is a Structural Problem, Not a Temporary One

The most important thing I can tell any data center developer is this: do not plan your business around the grid catching up. It will not catch up on your timeline.

The US interconnection queue holds more than 2,600 GW of projects waiting for grid connection, a figure representing more than twice the total installed generating capacity of the country, according to Lawrence Berkeley National Laboratory’s 2025 Queued Up report. The median wait time for a new grid interconnection is 4.5 years. For large-scale data center projects in constrained markets, timelines of eight to twelve years are not unusual.

This is not a temporary backlog caused by a spike in applications. It is a structural mismatch between the pace of commercial energy demand and the pace at which utility infrastructure can be built, permitted, studied, and connected. AI workloads, fleet electrification, and industrial reshoring are all drawing from the same constrained grid simultaneously. The queue is not clearing. It is growing.

The EV fleet operators who struggled most in the early years of electrification were the ones who believed the utility’s promised timeline. They signed leases, ordered vehicles, and told their operations teams to expect a charging infrastructure that arrived 18 months late. The data center operators who will struggle most in the next five years are the ones making the same assumptions.

Facilities Managers Are Being Asked to Solve an Infrastructure Problem

One of the most underappreciated dynamics in the current data center power crisis is where the pressure lands inside an organization. Prefabricated and modular data center systems have compressed construction timelines significantly. Containerized edge units can be operational weeks after on-site delivery. Room-scale prefabricated modules typically run 12 to 24 weeks from contract to installation. The build problem is largely solved.

That means the person handed a completed data center building is now also handed a power gap that no one in the procurement chain has an answer for. Facilities managers and infrastructure directors are being asked to make operational something that has no grid power and no confirmed delivery date for grid power. That is an unsustainable operating model, and it is becoming the norm rather than the exception.

In the electric vehicle (EV) and autonomous vehicle fleet world, we saw the equivalent dynamic play out at the depot level. Fleet managers were handed EVs and asked to charge them at facilities that had not been upgraded. The ones who found solutions did not wait for the utility. They found energy infrastructure that could be deployed immediately and scaled as the fleet grew.

The data center industry needs to make the same mental shift. The grid is not a planning assumption. It is a variable with a very wide confidence interval.

Mobile Energy Infrastructure Is Not a Workaround. It Is a Strategy.

At fleet depots, mobile battery energy storage has often been framed as temporary bridge power to cover the period before permanent infrastructure becomes available. In practice, these systems can serve more than one role: they can support operations during infrastructure delays and, if grid power arrives, remain part of a broader energy expansion.

Battery Energy Storage Systems deployed for bridge power do not become obsolete when a grid connection arrives. They become more valuable. A BESS system that powers a data center during the interconnection gap transitions into a permanent grid-tied asset if the utility connection goes live. It shaves peak demand, eliminates demand charge exposure, and in markets with active wholesale energy programs, participates in demand response and capacity markets, generating revenue from capacity that would otherwise sit idle.

The financial framing shifts entirely. Bridge power is not a capital expenditure liability. It is the first phase of a long-term energy infrastructure investment that continues to generate value well beyond the interconnection date.

Large-scale BESS deployments can be commissioned in weeks, delivering immediate behind-the-meter power that allows compute infrastructure to come online without waiting for formal grid interconnection. At a moment when AI compute capacity is among the most commercially valuable infrastructure on earth, every month of delayed operation represents revenue that does not materialize. Bridge power converts that waiting period into a generating period.

Mobile distributed generation systems add continuous runtime to the equation. Paired with BESS, they create a hybrid energy architecture where generation runs at its optimal load point, BESS handles peak demand and power conditioning, and intelligent energy management software coordinates both autonomously. The result is unlimited runtime, lower fuel cost per kWh delivered, and a system that scales as rack density increases.

Build Ahead of the Grid, Not Around It

The organizations getting this right are not treating the power problem as someone else’s responsibility to solve. They are treating it as a first-order infrastructure decision that gets made before the site selection, before the lease is signed, and before the equipment is ordered.

The grid is not the enemy. I want to be clear about that. On-site energy systems, including battery storage and dispatchable generation, can complement rather than replace the grid. They can provide power while utility interconnections and upgrades are pending and, where the technical and economic conditions support it, continue to help manage demand after grid service becomes available.

But the grid is also not a planning assumption any serious data center developer can rely on in 2026. The interconnection queue data, the transformer lead times running two to four years according to Wood Mackenzie’s Q2 2025 survey, and the regulatory shifts requiring data centers to fund their own grid upgrade costs all point in the same direction. The enterprises that deploy on-site energy infrastructure now will be generating revenue while their competitors are still in the interconnection queue.

That is not speculation. It is what happened in the EV fleet industry. The operators who moved first on microgrid energy infrastructure are the ones running the most competitive, most reliable, most cost-efficient fleets today. The ones who waited are still waiting.

Data center developers do not have to make the same mistake. The infrastructure exists. It deploys in days. It works before the grid arrives, and it keeps working after it does.

Do not wait for the utility to solve a problem the utility has not been able to solve for fifty years. Build ahead of the grid.

# # #

About the Author

Joshua Aviv is the Founder and CEO of SparkCharge, an intelligent microgrid provider. SparkCharge deploys microgrids that deliver grid-scale power in days, not years, for enterprises that can’t wait for the grid. Each turnkey microgrid combines SparkAI and SparkEMS with Battery Energy Storage and PowerHub generation. SparkCharge powers data centers, autonomous vehicle operations, ports and logistics hubs, construction and industrial sites, live events, and EV fleets across North America and the United Kingdom.

The post Bypassing the Grid: What Data Center Operators Can Learn From a Decade of Mobile Energy Deployment appeared first on Data Center POST.

TL;DR:

• Higher operating frequencies produce physically shorter wavelengths, and that inverse relationship increasingly shapes facility engineering.
• Operators respond with concrete, measurable changes to how they build and equip their data halls.
• Conventional data center shielding faces the most direct pressure, since enclosures sized for longer wavelengths can lose effectiveness.
• Rising rack density and its cooling needs form the second major area of architectural redesign.

# # #

Radio frequency (RF) physics now sits alongside power and compute demand as an active constraint on data center design. As operating bands climb, wavelengths shorten, and the effect shows up in places as ordinary as a cabinet seam or a cable tray.

The trend raises practical questions about why the shrinkage continues, why it calls for deliberate planning and how operators redesign their facilities in response.

The Shift Toward Shorter RF Wavelengths

Frequency and wavelength are inversely related, so a tenfold increase in one reduces the other by the same factor. Each wireless generation claims spectrum that earlier networks left largely unused. For 6G, much of that new territory sits in millimeter-wave bands and above.

Within the white space, server and switch interconnects already carry signal content at tens of gigahertz. A colocation hall that terminates fronthaul and backhaul links adds its own radio hardware. As a result, one building can host several active bands at once.

Federal measurement planning already lists traceable methods and calibration services for devices and materials as part of its 6G hardware validation deliverables. That work spans direct current to sub-terahertz frequencies. Test standards usually mature before the products that depend on them, which gives facility planners an early read on the bands headed their way.

Why Mindful Architecture Design Matters Now

Facilities that keep building around longer wavelengths open themselves to signal leakage, interference vulnerabilities and inefficient physical layouts. In practice, that can mean a radio shelf coupling noise into a neighboring storage array, or a late-added RF enclosure stranding floor space around it.

Engineers know the fixes well, and operators increasingly build them into the first drawing set. Timing is the harder problem to address. A change that slots easily into construction can mean negotiating maintenance windows row by row once a hall carries a live load.

How Data Centers Are Adapting Their Architecture

Operators move on three fronts, which are reworking shielding, managing rack density and cooling, and rethinking materials and layout. Each one answers a different pressure, yet they often compete for the same floor space and budget cycle.

Data Center Shielding Gets Reengineered

Shrinking wavelengths change the physical requirements of data center shielding. An enclosure’s performance depends on how its openings compare to the signals it must contain. Every seam and cable penetration acts as an aperture, passing energy more readily as its dimensions approach a meaningful fraction of the wavelength.

A cabinet that cleared its shielding test years ago, when the room’s radios ran in lower bands, can still pass inspection by eye. Once a Ka-band terminal moves in nearby, that same cabinet can start leaking. Ventilation panels often account for the difference, because airflow needs typically drive their perforation sizes long before anyone checks RF performance.

Wavelengths at 40 gigahertz measure 7.5 millimeters, a scale at which conventional shielding methods reveal critical weaknesses in many existing enclosures. Recovering that protection means continuous conductive seams, firmer gasket compression and waveguide-style vents held to much finer tolerances.

Upgrades like these hold up only when testing keeps pace, so rooms with lower-band certifications may need requalification once higher-frequency gear arrives. Many operators use the change log as the cue, so each new radio or antenna triggers a retest before it goes live. That timing pays off, since catching leakage at installation usually takes far less effort than tracing interference after a complaint.

Rack Density and Cooling Redesigns

While frequency scaling changes what a facility’s enclosures must contain, rising compute density reshapes its thermal load as a separate, parallel trend. The two meet on the same floor plan, and designers who solve them in isolation end up rerouting one around the other.

Power draw in high-performance computing has more than doubled since 2013, with recent deployments that exceed 125 kilowatts per compute rack at the high end. At those loads, air alone struggles to dissipate heat, so early adopters switched to cold plates and immersion tanks.

A liquid-cooled row looks different on the floor. Coolant distribution units and fluid loops claim space that raised-floor airflow once occupied, while busways bring power closer to the racks. Heavier cabinets also force a second look at slab loading, which can rule out some older halls.

Materials and Layout Choices for Higher Frequencies

Beyond shielding and cooling, material selection now starts with performance at the frequencies a facility expects to host. Enclosure alloys and conductive elastomers respond differently as wavelengths compress, and a sealed metal box supports more resonant modes. Absorbing panels increasingly turn up inside cabinets that never needed them before.

Purchase specifications have started to cover the entire operating spectrum of planned equipment. Some materials also drift electrically with temperature and age, and weight or corrosion resistance can veto an option that tests well on the RF bench.

Sites that host millimeter-wave backhaul radios or Ka-band ground-station hardware often provide dedicated zones for that gear, away from dense compute rows. Cable loss increases steeply with frequency, so equipment is typically placed close to its antennas, generally on rooftops or in perimeter spaces.

Prefabricated rooms and modular enclosures let those high-frequency areas grow or relocate as hardware mixes change. Factory crews can complete seams and grounding under controlled conditions, which is usually easier than chasing a leak through a finished wall on-site.

Building for What Comes Next in RF-Driven Design

Shorter wavelengths follow directly from the physics of higher-frequency communication, so the pressure they place on data centers should continue to mount. The cabinet retrofits and liquid loops going in today mark one stage of a longer evolution. Operators who plan for shrinking RF signals gain room to absorb each higher band without tearing out finished work.

As frequencies climb further, the data centers built around them are likely to evolve into purpose-built RF environments.

# # #

About the Author

Lou Farrell is the senior editor of AI content at Revolutionized Magazine. He has over five years of experience analyzing AI advancements across business, manufacturing, and engineering fields, providing insightful commentary on the latest breakthroughs, applications, and ethical responsibilities.

The post How Shrinking RF Signals Are Reshaping Data Center Architecture appeared first on Data Center POST.

TL;DR

• Alexander Marcus, CEO of ESI Total Fuel Management, is a speaker at the free online Kingdom Impact Summit, Nov. 16-20, 2026. IBAM’s Steve Adams hosts it.
• The summit is for Christian entrepreneurs, business owners, pastors, ministry leaders and marketplace professionals.
• The theme is “One Village Can Spark a Global Kingdom Movement.” More than 30 speakers are taking part, including Jordan Raynor, Mark Batterson, Peter Greer and Dr. Ché Ahn.
• New sessions are released daily, and each day’s sessions are free to watch for 24 hours. Closed captions are included.

# # #

By Alexander Marcus, CEO, ESI Total Fuel Management

I’m speaking at the free online Kingdom Impact Summit, Nov. 16-20, 2026. Join me for practical conversations on faith, work, and serving others.

What would change if your faith shaped the way you work, lead, and use what God has given you?

I’ve spent more than 40 years in engineering, fuels, and critical power, and 30 of them leading ESI Total Fuel Management. Along the way, I’ve come to believe that business is a stewardship: of people, resources, communities, and the environment. That’s why I’m glad to be a speaker at the Kingdom Impact Summit 2026. It’s a free online event on Nov. 16-20, hosted by Steve Adams, founder and president of IBAM.

Save your free place: summit2.ibam.org/sessions.

Who it’s for

The summit is for Christian entrepreneurs, business owners, pastors, ministry leaders, and marketplace professionals. It’s also for anyone who wants their work and resources to matter beyond the paycheck. You don’t need to know anything about business as a mission. You only need to be willing to take one faithful next step.

What to expect

The theme is “One Village Can Spark a Global Kingdom Movement.” Each day has its own focus:

• Day 1: See your work through God’s eyes
• Day 2: Lead with wisdom and steward for impact
• Day 3: Make your values visible at work
• Day 4: Use success to serve others
• Day 5: Put your gifts to work, here and around the world

More than 30 speakers are taking part, including Jordan Raynor, Mark Batterson, Peter Greer, and Dr. Ché Ahn.

How to watch

New sessions are released each day from Nov. 16-20. Each day’s sessions are free to watch for 24 hours after release, and closed captions are included. You don’t have to watch everything. Pick the conversations that fit the questions you’re carrying right now.

One thing to try first

Ask yourself: who could I serve better through the work I’m already doing? Choose one person and one action this week. That is often where faith at work begins.

Register for free here: summit2.ibam.org/?am_id=alexandermarcus

The summit is hosted by IBAM, a faith-driven organization that equips entrepreneurs in underserved communities with startup support, business training, and discipleship. After you register, you’ll see an optional All-Access Pass for people who want to keep the sessions.

The post What Would Change If Your Faith Shaped Your Work? Join Me at the Kingdom Impact Summit 2026 appeared first on Data Center POST.

TL;DR

• Mexico’s digital infrastructure market is expanding beyond established hubs such as Querétaro, creating new opportunities for data center development and regional investment.
• Public-private partnerships are playing an increasingly important role in addressing power availability, permitting, connectivity and infrastructure financing.
• New fiber routes, Internet Exchange Points (IXPs) and regional network operators are critical to strengthening Mexico’s connectivity ecosystem and cross-border digital corridors.
• Evolving telecommunications regulations and initiatives to expand rural connectivity are influencing the country’s long-term digital infrastructure strategy.

# # #

Mexico’s emergence as a strategic market for data centers, artificial intelligence (AI) and connectivity is opening new opportunities for infrastructure development across the country. While established markets continue to attract investment, emerging regions, evolving telecommunications policies, and growing demand for cross-border connectivity are changing how Mexico approaches its digital future.

These priorities shaped the agenda at Mexico Connect 2026, held September 9–10 in Mexico City. The two-day event explored the investments, partnerships, and infrastructure requirements supporting Mexico’s expanding digital economy, with discussions spanning hyperscale development, energy availability, fiber networks, regulatory reform, and digital inclusion.

Data Center Development Moves Beyond Traditional Markets

As demand for hyperscale and AI infrastructure increases, Mexico’s data center industry is exploring opportunities beyond Querétaro. Regions including Nuevo León, Jalisco, Hidalgo, San Luis Potosí, and Yucatán are attracting attention as potential locations for new facilities, supported by regional development initiatives and evolving infrastructure requirements.

Discussions examined the factors influencing site selection, including power availability, land, permitting, workforce readiness, and connectivity. The importance of coordination among developers, utilities, and regional governments was another focus, particularly as projects move into markets without established data center ecosystems.

These emerging opportunities also highlight the role of public-private partnerships in accelerating development. Aligning infrastructure investment with energy planning, regulatory support, and regional economic priorities will be essential to expanding Mexico’s data center footprint while supporting long-term growth.

Fiber Investment Strengthens Mexico’s Regional Connectivity

Mexico’s growing digital economy requires a connectivity ecosystem capable of supporting hyperscale facilities, cloud services, and increasing data traffic. Investment in long-haul fiber networks, Internet Exchange Points (IXPs), and Internet Service Providers (ISPs) is becoming increasingly important to improving network performance and expanding access to digital services.

The development of new cross-border fiber corridors was a major topic, particularly projects connecting Mexican infrastructure hubs with markets in the United States. Proposed and developing routes, including connections between Querétaro and Phoenix, demonstrate the opportunities to strengthen Mexico’s position as a digital bridge between North and Latin America.

Beyond international connectivity, discussions addressed the importance of improving domestic traffic exchange, reducing latency, and strengthening regional networks. As demand for AI and cloud services grows, efficient interconnection and a more extensive fiber backbone will be critical to supporting Mexico’s digital infrastructure expansion.

Regulatory Changes and Financing Influence Infrastructure Growth

Mexico’s evolving telecommunications regulatory environment is introducing new considerations for operators, investors, and infrastructure developers. The transition to the Comisión Reguladora de Telecomunicaciones (CRT) has raised questions surrounding spectrum policy, market competition, and future investment priorities.

The agenda also examined how financing models are adapting to support large-scale infrastructure projects. Development finance institutions, commercial banks, private investors, and international capital providers are playing important roles in funding fiber networks, backbone expansion, and connectivity initiatives.

As Mexico pursues additional infrastructure investment, regulatory clarity, and access to financing will be important factors in determining project viability. Greater alignment between public policy and private capital could help accelerate deployment while supporting competition and broader access to digital services.

Expanding Connectivity Beyond Major Urban Centers

While large-scale data center projects and international fiber corridors are important to Mexico’s digital economy, expanding connectivity to underserved communities remains a national priority. Smaller and regional telecommunications operators have an important role in reaching areas where traditional network deployment models may be less economically viable.

Discussions explored policy incentives, infrastructure sharing, and financing mechanisms designed to support these operators. Initiatives involving PROMTEL and other public-sector programs were also examined as potential ways to strengthen rural broadband deployment and improve last-mile connectivity.

These efforts emphasize the need for a more inclusive approach to infrastructure development. Supporting regional operators and encouraging collaboration between government and industry can help extend the benefits of digital investment beyond established commercial and technology centers.

Mexico’s Next Phase of Digital Growth

Mexico Connect 2026 highlighted the opportunities emerging as the country strengthens its position within the North American digital infrastructure ecosystem. From expanding data center development into new regions to improving cross-border connectivity and supporting underserved communities, the agenda reflected the importance of coordinated investment across multiple infrastructure sectors.

Mexico’s continued growth will depend on how effectively industry stakeholders address energy requirements, financing, regulatory changes, and network expansion. With public and private organizations working to align these priorities, the country has an opportunity to strengthen its connectivity infrastructure and support a more geographically diverse digital economy.

To explore more of the trends, opportunities, and developments influencing the global digital infrastructure industry, visit Capacity Media’s upcoming events calendar and discover where industry leaders will gather next.

The post Beyond Querétaro, Mexico’s Digital Infrastructure Opportunity Expands Nationwide appeared first on Data Center POST.

TL;DR

• Low public awareness and narrow misconceptions create a major hiring bottleneck for the data center industry as it struggles to build a talent pipeline to keep pace with exploding demand.
• Academic degree requirements obscure practical needs, turning away school leavers and tradespeople even though facility operations rely heavily on mechanical expertise and traditional trade disciplines.
• Recruitment must leverage transferable skills by drawing candidates from adjacent technical trades like plumbing, HVAC, and electrical work, as well as non-technical fields like hospitality and logistics.
• Bridging the workforce gap requires demystifying facilities through direct school outreach like the Data Centre STEM Ambassadors Programme and establishing modular, hands-on training pathways

# # #

By Stuart Newman, Business Development Manager at Apx Data Centre Solutions

The data centre sector forms the invisible backbone of modern life: the phone you’re holding, the video you’re streaming on the commute to work, every financial transaction, AI query and cloud-stored document. It all relies on complex physical infrastructure operating 24 hours a day, 365 days a year.

In spite of the role they play in our ‘always-on’ digital economy, the general understanding of the infrastructure inside these windowless warehouses remains a mystery. But it’s a predicament with severe consequences: the next generation simply doesn’t know we exist or what opportunities lie inside those walls. And that lack of awareness is having a major knock-on effect, with the industry facing an increasingly urgent challenge to secure the talent pipeline to keep pace with the exploding demand.

With this year’s National Data Centre Day focusing on the theme of ‘education’, it provides us, as industry professionals, with an opportunity to focus on how we recruit, train and demystify the sector for the next-generation. Because right now, our biggest bottleneck isn’t the pace at which technology is changing. It’s perception and access.

The misconception of technology

You only have to take a cursory glance at our news outlets today to see just how people feel about data centres. Around the globe, they’re lambasted for being monolithic, featureless warehouses filled with complicated computer racks.

Because of the picture that’s (rightly or wrongly) being painted, there’s a widespread assumption that working within one requires a specialised computer science degree or an elite background in software engineering.

It’s a narrow perception, often perpetuated by the mysterious aura that tends to hang over data centres,  that turns away bright young school leavers and skilled tradespeople before they’ve even considered applying.

At the same time, traditional educational pathways and standard apprenticeship frameworks often struggle to keep pace with the relentless speed of innovation in critical infrastructure. The result? A system where newcomers see only blockers rather than entry points, leaving a gap precisely where structured, hands-on training should be.

By focusing primarily on formal academic credentials, we obscure a fundamental truth: the digital world is not built solely on lines of code. Instead, it’s built on practical problem solving, high spec mechanical expertise and, crucially, on traditional trade disciplines.

In fact, we’re starting to see the latter play an important role in shoring up our industry. A recent Data Centre Dynamics Workforce report reaffirms this, by noting that organisations in the sector are “increasingly looking to adjacent technical infrastructure and utilities sectors to fill roles”.

And for me, this is something I’ve seen first-hand.

Early interest

While education-based pathways remain an important route to driving interest in the next generation of data centre professionals, more than a quarter of DCD’s respondents cited low industry awareness as a blocker to entry.

You only have to look at the official National Data Centre Day’s recent social media campaign to know that we’ve got an awareness issue. The series, which marked the industry’s second anniversary as National Critical Infrastructure, was called ‘then and now’ and aimed to document the realities of what school-age children aspire to become as adults.

Throughout the campaign, we’d seen posts about aspiring primary school teachers, pop stars and theme park enthusiasts, engineers, soldiers and athletes, among others. I, myself, wanted to be a motorsport engineer.

Besides it being a great tool to show the range of roles those individuals now hold in the industry, it’s a campaign that’s demonstrated two things: the first, that none of them knew this industry, or the roles within it, existed at a younger age. The second, just how many transferable skills exist across other industries and how interests can crossover.

All of this begs the question: without awareness at school level, how can we expect young minds to take an early interest? It starts, I think, with operators and suppliers being more open-minded; being willing to open doors to candidates with non-linear backgrounds and alternative skill sets.

The power of transferable skills

Take my own route into the industry. Having started my career as a plumber before transitioning into commercial sales and business development within the data centre sector, I’ve seen first-hand how directly these core trade skills crossover into high-tech environments.

A modern data centre is, at its heart, an exceptionally complex engineering site. Keeping mission-critical servers online requires sophisticated liquid and air cooling networks, precise environmental controls, high-voltage power distribution and continuous physical maintenance.

Skilled tradespeople such as electricians, HVAC technicians, pipefitters, plumbers and mechanical engineers already possess the core competencies these facilities need to function smoothly. So much so, it’s no surprise to see that, according to the DCD Report, as much as 19% of the workforce is being pulled from ‘expanded industries’ like utilities.

They understand pressure systems, fluid dynamics, fault diagnosis and strict safety compliance. Yet, despite possessing the exact real-world experience our industry desperately craves, many skilled workers rule themselves out of the running simply because they don’t realise how applicable their abilities are.

Beyond technical knowledge, non-technical roles across site operations, client account management, compliance and project logistics rely heavily on soft skills that are honed in entirely different sectors: think hospitality managers who excel under pressure and deliver exceptional client service, aviation cabin crew trained in rigorous safety protocols and crisis management, a logistics coordinator who wrangles complex schedules and inventory control or, perhaps, an event planner who knows how to manage vendor ecosystems and strict deadlines.

The reality though, is that none of these professionals are likely to even consider a data centre as their next logistical step, despite their core strengths being exactly what today’s modern facility management and data centre teams need.

This is precisely why we must challenge how we view candidate pools. Right now, according to the DCD Report, around 29% of the workforce is currently drawn from critical infrastructure. And this is where we risk pigeon-holing ourselves; when we look too closely at sectors that mirror our own, we aren’t creating a broader talent pipeline. We’re simply shuffling people within a very similar ecosystem.

To build genuine, long-term resilience, we must look beyond obvious structural parallels.

Redefining education for the infrastructure era

If we’re to bridge the growing skills gap, the industry can’t afford to wait for conventional education models to catch up. We must take an active role in demonstrating that university isn’t the only gateway to high-growth careers.

To truly inspire the next-generation, I think three fundamental changes need to happen.

We must demystify the facilities themselves. Industry leaders must open their doors to schools, colleges and trade academies to show students the tangible reality of these modern facilities. We’re starting to see this happen in pockets around the industry, primarily in an effort to bring down the proverbial walls and slowly break down the growing, negative perception building around the market.

In fact, just the other day, I was reading a story about DataVita boss, Danny Quinn, inviting Greens co-leader, Ross Greer and Dr Kat Jones, director of Action to Protect Rural Scotland (APRS), down to his facility in North Lanarkshire to help start that ‘demystification’ process and answer directly some of the criticism we’re seeing.

We’ve also got companies like Telehouse, a leading global data centre service provider, tapping into pop culture to bridge the gulf in public knowledge and tackle common misconceptions by creating an animated avatar character named ‘DC’.

But demystification also means getting out from behind our perimeter fences and into schools. A fantastic step in this direction is the newly-launched Data Centre STEM Ambassadors Programme, an initiative established by National Data Centre Day alongside AVK and STEM Learning. By connecting industry professionals directly with students aged 8 to 16 through classroom talks and mentoring, it gives us a structured route to show young minds what powering the digital world actually looks like. It’s an initiative that we, at Apx, are eager to throw our weight behind, because brokering these sorts of discussions and opening the doors, no matter the audience, is crucial to building a better future for the industry.

Once we’ve broken down the walls, we then need to look at the pathways to getting the next generation trained up. In any industry, starting from scratch can be extremely daunting. We need to put pathways in place to train new colleagues, providing them with the hands-on support needed to give them the confidence to seamlessly transition into critical infrastructure roles.

We also need modular, hands-on training initiatives that recognise existing baseline capabilities, whether from a mechanical trade, customer-facing role or emergency service background, and quickly bridge the gap to data centre application. By establishing clear, cross-skilling bootcamps and structured earn-while-you-learn pathways, we can lower the barrier to entry and show non-traditional talent that they already hold the building blocks for a successful career in critical infrastructure.

Opening the doors

The data centre sector sits at the absolute centre of global progress, but our continued growth relies entirely on the people behind the physical infrastructure.

Moving the needle requires us to be braver in how we recruit. By opening our facility doors, broadening our search parameters beyond traditional degrees and valuing transferable soft and trade skills, we can transform what was once a hidden, super specialised sector into one of the most accessible, rewarding career paths of the modern era.

If we want the next generation to take a chance on us, we have to be willing to take a calculated chance on them. It’s time to stop waiting for talent to find us and start showing them exactly where they fit in.

# # #

About the Author

Stuart Newman is the business development manager at Apx, specialising in bespoke Critical Cooling equipment engineered for both current and future operational demands.

The post Why The Data Centre Industry Must Rethink Its Talent Pipeline appeared first on Data Center POST.

TL;DR

• Parallel delivery runs two simultaneous workstreams, allowing site preparation and indoor module manufacturing to progress independently at the same time.
• Mod42’s off-site factory fabrication completes over 90% of construction indoors, eliminating weather delays, reducing on-site labor constraints, and enabling pre-testing before shipping.
• Mod42 targets deployment in under seven months by integrating complete volumetric modules with synchronized site preparation and utility commissioning

# # #

Originally posted on Mod42.

Traditional data center construction relies on a sequential approach where site grading, building construction, equipment installation, and testing occur one after another. This process takes between 24 and 36 months to complete. Since each phase depends entirely on the completion of the preceding step, unexpected delays in weather, labor, permitting, or utility interconnections can impact the entire plan. With demand for artificial intelligence and high-performance computing accelerating exponentially, this traditionally structured process becomes a major strategic liability.

Parallel delivery transforms scheduling by dividing deployment into two independent, simultaneous workstreams: the site track and the factory track. The site track manages civil work, foundations, grading, and utility connections, while the factory track produces complete data center modules indoors. With Mod42’s factory-built approach, over 90% of construction takes place in a controlled factory environment which allows projects to avoid common field disruptions such as severe weather and local labor shortages. Cooling, electrical, mechanical, and structural systems are fully integrated and tested off-site before shipping, allowing finished modules to be set into place as soon as the site pad is ready.

Successfully executing parallel deployment depends on complete volumetric modular construction, clear scope ownership, early alignment on pad dimensions and utility interfaces, and active risk communication. By coordinating factory fabrication alongside site preparation, Mod42 targets a total deployment schedule of under seven months covering fabrication, setting, integration, and final commissioning. This synchronized model drastically accelerates time-to-market compared to conventional multi-year timelines.

To read the original article, please click here.

The post How Parallel Delivery is Shortening the AI Infrastructure Deployment Schedule appeared first on Data Center POST.

TL;DR

• AI infrastructure development is accelerating, but power availability, site selection, and permitting challenges are influencing where and how new data centers are built.
• Growing demand for AI training and inference is driving innovation across compute, networking, memory, and storage infrastructure.
• Delivering AI capacity at scale requires greater coordination between infrastructure design, energy planning, capital investment, and operational strategies.
• As development expands, political considerations, regulatory requirements, and local community expectations are becoming increasingly important to project viability.

# # #

The rapid expansion of artificial intelligence (AI) is transforming digital infrastructure, creating new demands for computing capacity, power availability, networking, and data center development. These challenges were central to discussions at AI Infra Summit 2026, held September 15–17 in Santa Clara, California, where industry leaders explored the technologies, investment strategies, and operational considerations shaping the next generation of AI infrastructure.

Bringing together executives, technology innovators, infrastructure developers, and investors, the summit examined the increasingly interconnected systems required to support AI at scale. From high-performance computing and data movement to power constraints, site selection, and infrastructure economics, the discussions reflected an industry working to balance accelerated demand with the practical challenges of building and operating AI-ready environments.

Power Availability and Site Selection Are Reshaping Development Strategies

As AI workloads drive demand for larger and more energy-intensive data centers, securing viable development sites has become increasingly complex. Power availability, utility interconnection timelines, permitting requirements, and local political considerations are influencing project feasibility long before construction begins.

These challenges were central to discussions moderated by Ilissa Miller, CEO of iMiller Public Relations, who joined industry leaders examining the intersection of power, permitting, politics, and site selection. The discussion featured Raul Martynek, CEO of DataBank, and explored the factors influencing the race to secure suitable land for AI infrastructure development.

The growing importance of early-stage planning reflects a shift in how developers evaluate potential projects. Access to land and capital alone is no longer sufficient. Successful development strategies must also account for power delivery, regulatory approvals, infrastructure readiness, and the concerns of communities where facilities are proposed. These considerations are becoming essential to reducing project delays and improving long-term viability.

Scaling AI Compute Requires a More Integrated Infrastructure Approach

The increasing complexity of AI applications is placing new demands on computing architectures and the systems supporting them. As organizations expand their use of large language models, agentic AI, and real-time inference, infrastructure strategies must address performance, scalability, and efficiency across the entire technology stack.

The evolution of AI computing was explored by Ian Buck, Vice President, Hyperscale & HPC Computing at NVIDIA, whose participation in the summit included discussions on advancing infrastructure for agentic AI and the technologies supporting frontier AI development. These topics reflected the growing need to align processing capabilities with the networking, memory, and storage resources required to support increasingly sophisticated workloads.

The summit also highlighted the importance of designing infrastructure around specific AI requirements rather than relying on conventional computing models. As training and inference workloads evolve, organizations must consider how specialized hardware, distributed computing, and more efficient system architectures can improve performance while managing the cost and complexity of deployment.

Data Movement and Networking Are Critical to AI Performance

While computing capacity remains central to AI infrastructure development, the ability to move and access data efficiently is becoming equally important. Large-scale AI environments depend on high-bandwidth, low-latency connectivity between processors, memory, and storage systems, creating challenges that cannot be resolved through additional compute resources alone.

Among the experts addressing these challenges was Nick Harris, Founder and CEO of Lightmatter, who participated in discussions examining data movement, memory, and storage bottlenecks at scale. The session reflected the growing importance of improving how information moves throughout AI systems as organizations pursue greater performance and infrastructure efficiency.

Networking architecture, optical connectivity, and data transfer efficiency are increasingly important considerations for AI infrastructure investment. As distributed computing environments expand, improvements across these systems will be essential to supporting larger workloads, reducing performance limitations, and maximizing the value of computing resources.

Building AI Infrastructure for Long-Term Operational Efficiency

The rapid pace of AI infrastructure investment is creating pressure to deliver new capacity quickly while maintaining reliability, flexibility, and cost control. As facilities become larger and more technically complex, operators must evaluate how design decisions affect long-term performance, energy consumption, and the ability to adapt to future workloads.

The importance of integrated infrastructure planning was addressed by Ram Nagappan, Vice President, AI Infrastructure and Lead Architect, Gigawatt-Scale Data Centers at Oracle, whose presentation focused on the design of AI campuses at gigawatt scale. The topic highlighted the need to coordinate power infrastructure, computing requirements, and facility architecture when planning increasingly large AI deployments.

These considerations extend beyond initial construction. Operational strategies must account for changing power demands, evolving hardware requirements, and the integration of systems that support continuous, high-performance computing. Greater coordination between infrastructure developers, technology providers, and operators will be essential to managing complexity and supporting sustainable growth.

Building the Infrastructure Behind AI’s Next Phase

AI Infra Summit 2026 highlighted the scale of the challenges and opportunities facing the digital infrastructure industry. While advances in computing technology continue to accelerate AI adoption, the ability to support that growth depends on coordinated investment across power, networking, data center development, and operational systems.

The discussions reinforced that AI infrastructure cannot be planned or evaluated in isolation. From securing viable development sites to addressing data movement bottlenecks and designing gigawatt-scale campuses, the industry’s next phase will require closer alignment between technology innovation, infrastructure investment, and the communities supporting new development.

As demand for AI continues to expand, the strategies explored in Santa Clara provide a framework for addressing the practical requirements of scalable, resilient, and efficient digital infrastructure. Industry leaders will have another opportunity to continue these discussions at AI Infra Summit 10X 2027, scheduled for August 30 – September 2, 2027, in San Jose, California.

The post AI Infrastructure Growth Challenges Traditional Approaches to Data Center Development appeared first on Data Center POST.

TL;DR

• First connections: Empire Fiber Internet is now serving the first customers on Madison County’s publicly owned fiber network, which is expected to reach more than 3,600 homes, farms and businesses when completed.
• Project details and timeline: The $31 million public-private initiative spans nearly 389 miles, with construction expected to finish by early 2027. Empire is also expanding service beyond the publicly funded footprint.
• Continued momentum: The milestone builds on Empire’s recent network expansions in Ithaca, New York, and Berwick, Pennsylvania.
• Expanding access: The project expands access to connectivity that supports education, telehealth, remote work and local businesses while demonstrating how government and private providers can collaborate to address rural broadband gaps.

# # #

Empire Fiber Internet, a leading fiber optic internet service provider serving communities across New York and Pennsylvania, has begun serving the first customers on Madison County’s publicly owned fiber broadband network, marking a significant milestone in a multiyear effort to expand internet access across the rural New York county. Announced October 14, the initial connections bring the nearly 389-mile network closer to its goal of reaching more than 3,600 homes, farms and businesses, many of which previously had limited or no service.

The first serviceable portions of the network extend between the towns of Cazenovia, Sullivan, and Lenox. Additional locations will become eligible for service as crews complete construction and activate the remaining infrastructure, with construction expected to conclude by early 2027. Empire Fiber Internet will operate the network and provide internet service, while Madison County retains ownership of the infrastructure.

The $31 million Madison County Connect Project is supported by the USDA Rural Development ReConnect Program, New York State’s ConnectALL Municipal Infrastructure Program, Empire Fiber Internet, and Madison County. Its origins date to a $10.1 million USDA ReConnect grant awarded in 2021, subsequently increased to $12.7 million, to reach 1,087 unserved homes and businesses. Construction began in November 2023, and the broader project now encompasses more than 3,600 potential service locations.

Empire’s involvement also extends beyond operating the publicly funded network. The company is expanding service to additional addresses along the county’s network backbone in Cazenovia, Chittenango, and Canastota, including select neighborhoods outside the funded footprint. These extensions broaden the project’s reach while introducing another provider option for more residents and businesses.

The Madison County activation follows other recent developments for Empire Fiber Internet across New York and Pennsylvania. In August, the company expanded its Ithaca network to more than 700 additional homes and businesses, bringing its availability there to more than 5,400 locations. In September, Empire launched its fiber network in Berwick, Pennsylvania, following a $1.4 million local investment. The company also received its fourth consecutive FLX Finest Gold award for Best IT / Office Technology in 2026. Together, these milestones highlight a period of network expansion alongside continued community recognition.

With construction nearing completion, attention now turns to activating the remaining areas. Residents and businesses can check their addresses through Empire Fiber Internet’s availability tool and follow county updates as additional portions of the network open for service.

To learn more about Empire Fiber Internet, visit empirefiber.com.

The post Empire Fiber Internet Connects First Customers on Madison County’s Publicly Owned Broadband Network appeared first on Data Center POST.

TL;DR

• Standard cages and power feeds rarely fit modern enterprise workloads, so effective data center environments are now built around the customer.
• Four forces drive the shift: AI-driven density (some deployments need 50–220 kW per cabinet), mixed air- and water-cooling, diverse workloads, and data security and sovereignty requirements.
• One-size-fits-all forces a choice between overbuilding and paying for unused capacity, or underbuilding and hitting limits as demand grows. Hybrid environments across colocation, private cloud, public cloud and on-premises add fragmented vendors, contracts, and support models.
• When evaluating providers, look at power and cooling, connectivity, flexibility, compliance and support. Start with your workloads, then work backward to the infrastructure.

# # #

By Sam Zurzolo, VP of Data Center Services, 365 Data Centers

For years, data center decisions came down to a short list: how much space, how much power, and how much redundancy. Those fundamentals still matter today, but no two enterprises run the same workloads. As a result, a standard cage and a standard power feed rarely fit in most modern scenarios.

After more than three decades in telecom and data center operations, from U.S. Navy electronics to running facilities across the country at 365 Data Centers, one of the biggest changes I’ve seen is the move from standardized offerings to environments built around the customer.

What Changed

Several forces are driving the shift:

• Density: AI and data-intensive applications are pushing power requirements well past what legacy facilities were designed for. Some deployments now call for anywhere between 50 to 220 kW per cabinet.
• Cooling: Higher densities often mean air-cooled and water-cooled approaches working together, and the right mix depends on the workload.
• Workload diversity: Inference, analytics, regulated data, and traditional applications each place different demands on power, latency, and connectivity.
• Data control: Security, privacy, and sovereignty requirements increasingly determine where, and how, workloads can run.

Why One-Size-Fits-All Falls Short

A standardized environment forces compromises. Organizations either overbuild and pay for capacity they don’t use, or they underbuild and run into limits just as demand grows. Neither is efficient. Teaching fiber design early in my career taught me that the network path matters as much as the rack it connects to.

Hybrid complexity adds to the problem, because many enterprises now operate across colocation, private cloud, public cloud, and on-premises systems. Without a plan for how those pieces connect, teams end up managing fragmented vendors, contracts, and support models.

From an operations standpoint, customization also affects reliability. Power distribution, cooling design, maintenance schedules, and network paths all work best when they match the actual load. A well-matched environment is easier to maintain and easier to scale.

What to Evaluate

When assessing a provider, consider these questions:

1. Power and cooling: Can the facility support your density today, and can it grow with you?
2. Connectivity: Are carrier options, cloud on-ramps, and bandwidth available where your workloads live?
3. Flexibility: Can you start with colocation and add private cloud or other services later, rather than committing to everything at once?
4. Compliance: Does the environment meet the security standards your industry requires?
5. Support: Will experienced people work with you on design and ongoing operations, not just hand you a rack?

Start with your workloads, then work backward to the infrastructure. A clear picture of your current and projected power, latency, and compliance needs makes every other decision easier.

Planning for What’s Next

Customization doesn’t mean complexity. The goal is an environment that fits today’s requirements and can adapt as they change. Enterprises that treat infrastructure as a tailored foundation, rather than a commodity, are better positioned to deploy new technologies with confidence.

At 365 Data Centers, we work with customers to design colocation, connectivity, and cloud environments around their specific needs, with the support of a team that has operated critical infrastructure for decades. If you’re reassessing your data center strategy, we’re glad to help you think it through. Contact us today at 365datacenters.com/contact.

The post The Customization Imperative: Why Data Center Solutions Are No Longer Standardized appeared first on Data Center POST.

TL;DR

• AI-driven growth is reshaping Europe’s data center industry, placing greater demands on power availability, infrastructure design and operational flexibility.
• Battery energy storage is expanding beyond traditional backup applications to support grid stability, energy management and increasingly dynamic workloads.
• Rising compute densities are accelerating innovation in cooling technologies while challenging conventional approaches to data center design and efficiency.
• Europe’s digital infrastructure expansion increasingly depends on coordinated energy planning, connectivity investment and stronger community engagement.

# # #

As artificial intelligence (AI) accelerates demand for digital infrastructure, Europe’s data center industry faces mounting pressure to expand capacity while addressing power constraints, sustainability requirements, and increasingly complex workloads. These challenges were central to discussions at DCD Connect London 2026, held September 15-17, where industry leaders explored the technologies, strategies, and partnerships shaping the next phase of data center development.

Bringing together operators, technology providers, investors, and infrastructure experts, the event highlighted how the industry is adapting to a changing energy landscape. Discussions examined everything from battery storage and advanced cooling to AI-ready facility design, grid modernization, and the growing importance of community acceptance.

Rethinking Power Infrastructure and Energy Storage

Power availability and reliability emerged as major considerations as operators seek to support increasingly energy-intensive AI workloads. With grid connection delays and electricity costs affecting development strategies, the industry is exploring new approaches to energy storage, onsite generation, and grid participation.

The evolving role of battery storage was a key area of discussion with Liz Wallace, Senior Product Manager at ZincFive, among the industry experts examining how Battery Energy Storage Systems (BESS) are moving beyond traditional backup applications. Topics included power quality, peak shaving, load shifting, and demand response, along with opportunities to better manage fluctuating energy requirements.

As data centers become more active participants in the energy ecosystem, storage technologies may offer opportunities to improve operational flexibility and support renewable energy integration. These capabilities are becoming increasingly relevant as operators balance uptime requirements with energy costs and sustainability objectives.

Adapting Infrastructure Design for AI Workloads

The rapid evolution of AI hardware is challenging traditional data center design assumptions. Higher rack densities, changing workload profiles and closer integration between IT and operational technology are creating demand for facilities that can adapt without requiring extensive redesign.

Among the industry leaders addressing these challenges was Ivy Tavershima, Senior Technical Program Manager – AI Delivery at Microsoft, who participated in discussions examining whether standardized facility designs can remain effective as computing requirements continue to change. Topics included modular infrastructure, adaptable power, cooling systems, and the challenge of balancing repeatable designs with workload-specific requirements.

Cooling innovation was another important focus. As operators prepare for higher-density deployments, discussions explored liquid cooling, hybrid thermal architectures, and opportunities to improve efficiency without compromising reliability. The growing complexity of AI infrastructure is reinforcing the need to consider flexibility and long-term operational requirements during the earliest stages of facility planning.

Addressing Europe’s Energy and Connectivity Constraints

Europe’s digital infrastructure expansion is increasingly influenced by where sufficient power and network connectivity can be secured. Traditional data center markets face growing grid and permitting challenges, prompting developers to evaluate emerging regions with more favorable energy conditions.

The evolving relationship between data centers and energy providers was examined by industry experts including Goran Arya, Director, Energy Origination & Structuring, EMEA at Equinix. Discussions addressed grid capacity allocation, demand-response opportunities, and new models for integrating large-scale digital infrastructure into Europe’s electricity systems.

Connectivity investment must also evolve alongside changing development patterns. Joshua Seawell, Director of Product Management at Sumitomo Electric Lightwave, joined discussions exploring how fiber infrastructure, long-haul routes, and metropolitan networks must adapt as data center development expands into new geographic markets. Coordinated planning between energy providers, network operators, and developers will be essential to supporting reliable, high-capacity digital infrastructure.

Building Community Support for Sustainable Growth

Beyond technical and operational challenges, the industry is confronting increased scrutiny surrounding the environmental and economic impacts of data center development. Community concerns about energy consumption, infrastructure demands, and local benefits are influencing project planning and approval processes.

The importance of establishing a social license to operate was explored by industry representatives including Michael Blake, EMEA Community and Workforce Development, Regional Manager – Global Data Centers at Google. Discussions focused on incorporating community value into development strategies, improving public understanding, and engaging local stakeholders before opposition becomes a barrier to progress.

These considerations reinforce the importance of treating community engagement as an integral part of infrastructure development rather than a final step in the approval process. As Europe’s data center footprint expands, transparent communication and measurable community benefits will play an increasingly important role in supporting responsible growth.

Preparing for the Next Phase of Digital Infrastructure

DCD Connect London 2026 highlighted the interconnected challenges facing Europe’s data center industry. AI-driven demand is accelerating infrastructure investment, but the ability to deliver new capacity depends on more than computing technology alone. Energy availability, storage innovation, cooling efficiency, connectivity, and community acceptance are all influencing how and where facilities are developed.

The discussions underscored the need for greater collaboration across the digital infrastructure ecosystem. As operators and technology providers work to meet growing demand, flexible infrastructure strategies, and coordinated planning will remain essential to building resilient, efficient, and sustainable data centers. The industry will have another opportunity to reconnect and continue these important conversations at DCD Connect London 2027, taking place September 14-15, 2027. 

The post The Forces Reshaping Europe’s Data Center Future appeared first on Data Center POST.

TL;DR

• Ecosystem over Infrastructure: Data center value is increasingly driven by surrounding carrier density, cloud on-ramps, and Internet Exchanges rather than power and space alone.
• Compounding Business Value: Carrier-neutral environments leverage network effects to offer enhanced choice, vendor redundancy, and efficient paths for AI and hybrid cloud workloads.
• Future-Proofing Strategy: Interconnection serves as a core driver of resilient infrastructure, enabling organizations to adapt to evolving multi-cloud and distributed data needs

# # #

Written by iMiller Public Relations on behalf of 1547 Critical Systems Realty.

Modern data centers are evaluated not just by physical power and cooling capabilities, but by the strength and diversity of their surrounding connectivity ecosystems. While physical infrastructure forms the essential baseline, access to dense networks, cloud platforms, and Internet Exchanges transforms a facility into a key business driver. Organizations increasingly prioritize carrier-neutral environments where multiple providers compete, giving businesses greater freedom of choice, reduced dependence on single vendors, and improved cost flexibility.

Interconnection ecosystems operate on compounding network effects, where each added network or cloud platform creates exponential value for existing tenants and attracts further participants. This density becomes crucial for modern workloads, such as distributed artificial intelligence tasks and hybrid cloud deployments, which require direct cloud on-ramps and resilient multi-route connectivity. Consequently, facility connectivity is no longer viewed as an optional add-on, but as a central component of long-term infrastructure strategy and geographic redundancy.

Real-world deployments within 1547’s portfolio demonstrate how these ecosystem principles function in practice. At Chase Tower in McAllen, Texas, 1547 operates MCT-IX, a carrier-neutral Internet Exchange connecting over 46 networks across two meet-me rooms at speeds up to 100G, serving as a hub for cross-border traffic between the United States and Latin America. Additionally, the PTOR1 facility at Portland’s Pittock Block delivers extensive carrier access and AWS Direct Connect capabilities, allowing enterprises to link colocation environments directly to cloud platforms.

To read the original article, please click here. 

The post The Business Value of Interconnection Ecosystems appeared first on Data Center POST.

TL;DR

• Fiber operators are reassessing network expansion strategies as AI workloads increase demand for capacity, low-latency connectivity and stronger returns on infrastructure investments.
• Modular data center designs and edge computing are gaining attention as operators seek faster, more flexible ways to deploy AI-ready infrastructure.
• Community engagement and public trust are becoming increasingly important to project approvals, investment decisions and long-term development.
• Across fiber, data centers and energy infrastructure, greater coordination is essential to overcoming deployment challenges and supporting sustainable growth.

# # #

Digital infrastructure leaders gathered in Austin, Texas, September 1–3 for Metro Connect Fall 2026, where discussions focused on the opportunities and challenges shaping the future of fiber networks, data centers, energy, and infrastructure investment. Held alongside Datacloud USA and related industry forums, the event examined how accelerating artificial intelligence (AI) demand is influencing infrastructure development, financing strategies, and operational priorities.

From expanding fiber capacity and accelerating modular data center deployments to overcoming power constraints and strengthening community relationships, discussions reflected an industry navigating increasingly complex development requirements. While demand for digital infrastructure continues to grow, greater emphasis is being placed on strategic investment, operational efficiency, and collaboration across the infrastructure ecosystem.

Fiber Investment Priorities Continue to Evolve

As AI applications place greater demands on network capacity and performance, fiber operators are reevaluating how and where they invest. Timothy Parker, Chief Growth Officer at Assured Communications, joined industry leaders in examining the suitability of existing fiber infrastructure, evolving customer requirements, and the economics of new network deployments.

With investors applying greater scrutiny to capital allocation, operators face increasingly complex decisions about whether to build new networks or acquire existing assets. The growing importance of AI inference, distributed computing, and low-latency connectivity is also influencing network planning, creating opportunities for fiber providers that can align infrastructure investments with measurable customer demand.

Modular Infrastructure Supports the Growth of AI

Speed, scalability and adaptability are becoming increasingly important as data center operators respond to rising AI workloads. Valerie Crafton, Chief Revenue Officer at Mod42 and Doug Recker, CEO of Duos Edge AI, were among the industry leaders addressing how modular construction and prefabricated solutions can help accelerate infrastructure delivery.

As demand grows for both hyperscale campuses and distributed edge facilities, modular approaches offer potential advantages in deployment timelines, standardization, and flexibility. Discussions explored how prefabrication, digital design tools, and manufacturing innovations can help operators respond to changing capacity requirements while addressing construction and supply chain challenges.

Community Trust Becomes a Critical Infrastructure Priority

Public perception, local stakeholder expectations and community opposition are increasingly influencing where and how digital infrastructure projects move forward. Ilissa Miller, CEO of iMiller Public Relations, led a discussion with representatives from data center development, energy, municipal government and community affairs about the challenges of securing community support and maintaining the social license to operate.

As concerns surrounding energy consumption, environmental impacts, and local development continue to influence public opinion, developers face greater pressure to demonstrate the value of proposed projects. Discussions emphasized the importance of engaging stakeholders early, understanding community priorities and establishing credible relationships before development plans advance. These considerations are also becoming more relevant to investors as permitting delays, political risk, and community opposition can influence project timelines, market selection, and asset valuations.

Infrastructure Growth Requires Greater Industry Coordination

Across fiber, data center, energy, and investment discussions, a common challenge emerged as infrastructure development becomes increasingly interconnected. Power availability influences data center site selection, fiber connectivity determines network performance, and construction capacity affects how quickly projects can move forward.

From emerging financing structures and workforce development to behind-the-meter power strategies and network interconnection, discussions reinforced the importance of aligning infrastructure planning across traditionally separate sectors. As operators navigate more demanding technical requirements and tighter investment expectations, collaboration among developers, utilities, network providers, suppliers, and communities will play an increasingly important role.

Metro Connect Fall 2026 reflected an industry moving beyond the race to build capacity toward a more disciplined approach to growth. Meeting the demands of AI will require not only investment in technology and infrastructure, but also stronger coordination, sound financial strategies, and the public trust necessary to bring projects to completion.

As the industry continues to address these opportunities and challenges, the conversations will continue at Metro Connect 2027, February 8-10 in Hollywood, Florida. Join digital infrastructure leaders once again to exchange insights, strengthen industry connections and explore the priorities shaping the next phase of infrastructure growth. 

The post AI, Infrastructure Investment, and Community Trust Take Center Stage at Metro Connect Fall 2026 appeared first on Data Center POST.

TL;DR

• Simplified Service Delivery: Centralizing service configuration at the point of presence removes the need for repeated WAN changes, complex multicast design, and premises-level NAT adjustments.
• Scalable Multi-Service Platform: Once the initial Layer 2 segment is established, operators can easily attach new cloud, partner, or AI infrastructure services to the customer site.
• Commercial Availability: Managed by a cloud-based Path Computation Engine, the product line launches with the PBC 124-C to expand addressable markets for off-net sites

# # #

Written by iMiller Public Relations on behalf of MaiaEdge.

Connecting remote customer premises over existing IP transport allows service providers to expand their reach and service delivery without constructing new fiber infrastructure. By pairing a customer-premises device like the PBC 100 Series with a PoP-side PBC 2000, operators establish an encrypted Layer 2 SD-Ethernet extension over any available transport, including fiber, dedicated internet access, or satellite. This approach delivers a consistent Ethernet handoff directly to the customer site, bypassing the need for complex, manually negotiated network-to-network interfaces.

Traditional last-mile connections often complicate service expansion, requiring repeated WAN modifications, network address translation adjustments, and complex multicast routing configurations for every new offering. Extending an isolated Layer 2 Ethernet segment back to the operator’s point of presence solves this bottleneck by centralizing service management. Operators can seamlessly attach additional offerings, such as cloud connectivity, partner networks, or artificial intelligence infrastructure, directly at the central network level without modifying customer-premises hardware or WAN settings.

Driven by a cloud-based Path Computation Engine, the solution automates path selection and measures service performance across the encrypted connection. The initial deployment in this MaiaEdge product line, the PBC 124-C, enables operators to turn existing connectivity into an extensible platform for recurring service growth. By simplifying last-mile delivery, service providers can rapidly unlock new revenue streams and address off-net customer locations that were previously difficult to serve.

To read the original article, please click here.

The post MaiaEdge Introduces the PBC 100 Series appeared first on Data Center POST.

TL;DR

• 3C and Datalec Precision Installations have merged to form 3C Datalec, combining liquid cooling, engineering, manufacturing, connectivity, and facilities management expertise.
• The combined company forecasts more than $400 million in 2026 revenue, with approximately 1,000 employees and active projects across 16 countries.
• Backed by new investment, 3C Datalec plans to hire 200 engineers and open facilities in Poland, Canada, and the United States over the next 12 months.

###

As AI workloads increase power and cooling requirements across data centers, operators are placing greater emphasis on integrated infrastructure solutions that support projects from initial design through ongoing operations.

3C, the liquid cooling infrastructure company founded by former CoolIT Systems CEO Steve Walton, has merged with Datalec Precision Installations (DPI) to form 3C Datalec, bringing together expertise in liquid cooling, mechanical and electrical (M&E) engineering, manufacturing, connectivity, and facilities management. The combined company will provide integrated services for AI and high-density data center environments.

The newly formed organization has approximately 1,000 employees and forecasts more than $400 million in revenue for 2026. With active projects in 16 countries totaling hundreds of megawatts, 3C Datalec serves four of the five largest hyperscalers, more than 25 colocation providers, neoclouds, and enterprise customers across international markets.

Backed by new investment from Longbow Capital, S2G Investments, Apogem Capital, and QIC, the company plans to expand its global operations and technical capabilities. Over the next 12 months, 3C Datalec expects to hire 200 engineers and establish new facilities in Warsaw, Poland; Calgary, Canada; and Austin, Texas.

The company’s services will be organized around three core areas: Create, covering design and engineering; Construct, encompassing liquid cooling, M&E, manufacturing, and connectivity installation; and Care, supporting commissioning, maintenance, operations, and facilities management. This approach brings together services that are increasingly interconnected as operators address higher rack densities and more demanding cooling requirements.

“Rack densities have outgrown air, and liquid cooling is no longer optional,” said Steve Walton, CEO of 3C Datalec. “At CoolIT, we spent years preparing for this moment and became a global leader in liquid cooling. The hard part now is making cooling, power, construction and operations work as one system, while running at scale. We’re combining our liquid cooling expertise with Datalec’s engineering, manufacturing and FM capabilities, so customers have one partner from design to operation.”

Workforce development will also be a focus as the company expands. Through apprenticeships, technical training, and career development programs, 3C Datalec aims to develop additional expertise in liquid cooling and critical infrastructure while supporting the growing demand for skilled data center engineers.

The merger reflects the increasing importance of coordinated design, construction, cooling, and operational services as AI infrastructure requirements become more complex. By combining technical expertise with an established international delivery presence, 3C Datalec is expanding its ability to support both new data center developments and upgrades to existing facilities.

To learn more about the merger and 3C Datalec’s expanded capabilities, read here: www.imillerpr.com/news/3c-merges-with-datalec-to-create-us400m-revenue-liquid-cooling-and-data-centre-services-platform

The post 3C and Datalec Merge to Form $400M+ AI Data Center Infrastructure Platform appeared first on Data Center POST.

TL;DR

• AI workload demands: Rapidly accelerating rack densities and power requirements mean operators must establish accurate visibility into existing capacity before committing capital to new deployments.
• Unified operational visibility: Disconnected spreadsheets and software create visibility gaps, whereas infrastructure intelligence links facility, IT, and connectivity dependencies into a single operational model.
• Proactive planning with digital twins: Integrating operational data allows teams to simulate changes, identify power or cooling constraints early, and uncover genuine available capacity.
• Pragmatic implementation: Operators can begin by validating critical assets against physical reality, mapping system dependencies, and strengthening change management processes to support compressed AI timelines.

# # #

Artificial intelligence is pushing data center infrastructure into unfamiliar territory. Rack densities are increasing, power and cooling requirements are becoming more demanding, and AI deployments are moving at a pace that can make traditional planning cycles feel outdated.

But the challenge facing data center operators is not simply how to provide more power, more cooling, or more space. Before operators can determine what they need next, they need to know what they already have. That makes infrastructure intelligence a critical foundation for AI.

Infrastructure intelligence is the ability to maintain an accurate, connected view of the data center environment, from physical facilities and power systems to IT equipment, connectivity, dependencies, and available capacity. It gives operators the context they need to understand not only what exists, but how everything is connected, what it can support, and what could happen when something changes.

AI Raises the Cost of Inaccurate Information

In a traditional data center environment, an outdated record or incorrect equipment specification may create an operational inconvenience. In an AI environment, the consequences can be much greater. AI racks can place significantly higher demands on power and cooling than conventional workloads. A deployment that appears to have sufficient capacity on paper may encounter constraints once the actual power draw, heat load, cooling configuration, and upstream dependencies are considered.

At the same time, infrastructure information is often distributed across multiple systems, including DCIM platforms, CMDBs, facility monitoring tools, CAD drawings, spreadsheets, and other operational records. Each may contain valuable information, but they do not necessarily provide a connected picture of the environment.

The result is a fundamental visibility problem: teams may know what individual systems contain without having a reliable understanding of how those assets relate to one another. For AI infrastructure, that distinction matters.

From Asset Visibility to Infrastructure Intelligence

Knowing that a rack exists is not enough. Operators need to understand what is in the rack, how much power it actually consumes, where that power comes from, how it is cooled, what equipment it connects to, and which workloads or services depend on it. This is where infrastructure intelligence moves beyond inventory management.

By connecting facility, IT, and connectivity information, operators can establish a shared operational model of the data center. That model can help answer questions such as:

1. Can this location support another high-density AI rack?
2. What power and cooling resources would a new deployment require?
3. Which systems, services, and dependencies could be affected by a planned change?
4. Where is there genuine capacity, rather than capacity that only appears available in a disconnected system?
5. What happens if a cooling unit or power feed is taken offline for maintenance?

These are not theoretical questions; they are the questions operators need to answer before committing additional capital, deploying new infrastructure, or making changes to an increasingly complex environment.

Digital Twins Must Reflect Reality

Digital twins can play an important role in building this intelligence, but a digital twin is only useful if it remains aligned with the physical environment. A static 3D representation of a data center may provide an attractive visualization, but it does not necessarily provide the operational context required for AI infrastructure planning. A useful digital twin needs to connect the physical environment with equipment, capacity, dependencies, and operational data. 

That enables teams to move from simply seeing the environment to modeling potential changes. For example, before deploying a high-density AI rack, an operator could model its expected impact on power, cooling, and related infrastructure. Instead of discovering constraints after the hardware arrives, teams can identify them during the planning process.

Maintaining that value requires discipline. When equipment is moved, circuits are changed or infrastructure is upgraded, those changes need to be reflected in the model. Otherwise, the digital twin gradually becomes another source of outdated information rather than a trusted representation of the environment.

Infrastructure Intelligence Enables Better Decisions

The objective is not to replace every system already used by a data center operator. It is to connect the information those systems provide so teams can understand the relationships between infrastructure components. A data center infrastructure management platform such as FNT Command bridges this gap by connecting facility, IT, and connectivity data into a shared operational model. Rather than simply providing another visualization layer, it gives teams the infrastructure relationships and capacity context needed to plan deployments, understand dependencies, and make decisions based on the environment as it actually exists.

This becomes particularly important as AI accelerates the pace of change. Traditional infrastructure planning often assumes relatively predictable refresh cycles and gradual capacity growth. AI deployments can compress those timelines, requiring organizations to evaluate infrastructure decisions much faster.

The organizations that can respond effectively will not necessarily be those with the most hardware. They will be the ones that can quickly determine what their infrastructure can actually support.

Start With the Infrastructure That Matters Most

Building infrastructure intelligence does not require operators to model every asset in a data center on day one. A more pragmatic approach is to begin with the areas where AI deployments create the greatest operational and financial consequences. From there, operators can take four practical steps:

1. Reconcile records with physical reality:
   Start by validating existing infrastructure records against what is actually deployed in the data center. This creates a reliable foundation for planning and decision-making.
2. Connect assets and dependencies:
   Establish the relationships between equipment, power, cooling, and other infrastructure. Understanding these dependencies provides the context needed to evaluate the impact of a change.
3. Strengthen change processes:
   Ensure that moves, adds, changes, and upgrades are consistently captured in the infrastructure model. This helps prevent the gap between documented infrastructure and physical reality from growing over time.
4. Expand intelligence across the environment:
   Once the most consequential AI infrastructure is accurately modeled and operational processes are established, organizations can extend that intelligence across the broader data center environment.

The goal is a continuous operational discipline in which infrastructure plans are validated before changes are made, changes are reflected in the infrastructure model, and the model is routinely reconciled against reality. This creates a foundation that can scale as AI infrastructure evolves.

AI Performance Depends on More Than GPUs

The AI infrastructure conversation often focuses on processors, power availability, cooling technologies, and data center capacity. All those elements are important, but the ability to operate them effectively depends on something less visible: the quality of the information connecting them.

AI may be the technology transforming the data center, but infrastructure intelligence is what allows operators to understand, plan and manage that transformation. The question is no longer simply, “How much more power, cooling or space do we need?” It’s also: “What do we have, what can it actually support, and what happens if we change it?” Without reliable answers, even the most advanced AI infrastructure can be constrained by the environment supporting it. 

With accurate, connected infrastructure intelligence, operators have the visibility and context needed to make faster, more informed decisions as AI continues to reshape the data center.

# # #

About the Author

Oliver Lindner has over 30 years of experience in IT and the management of IT infrastructures with a focus on data centers. He has worked for many years at FNT Software, a leading provider of software for managing digital infrastructure. In his current position as Director of Product Management, he is responsible for the strategic direction and continuous improvement of the software products for data centers. The aim is to support customers in the efficient and transparent design of their IT infrastructure.

Oliver Lindner attaches great importance to customer focus, innovation and quality. His expertise also includes the development and provision of Software as a Service (SaaS) solutions that offer customers maximum flexibility and efficiency. To this end, he works closely with his own team, partners and customers to create sustainable and innovative software solutions.

The post AI Is Only as Strong as the Infrastructure Intelligence Behind It appeared first on Data Center POST.

TL;DR

• Drawings Reflect Design Intent, Not Installed Reality: Equipment records extracted from engineering drawings show intended tag names and ratings, but they cannot confirm actual field swaps, serial numbers, or procurement updates.
• Separate Document Data from Field Verification: Every record entry should cite its source drawing number and revision letter, maintaining a clear distinction between drawing-derived data and field-confirmed status.
• Preserve Exact Tag Names and Explicitly Flag Uncertainty: Record tag names precisely as printed on drawings while logging operational designations separately, and explicitly mark ambiguous or unreadable values instead of making assumptions.

# # #

Before a data center is handed over, someone has to produce the equipment record that operations will live with: every switchboard, UPS module, generator, chiller, pump, CRAH unit and valve, with a tag, a location and a source. Most of that record starts in the drawings. One-line diagrams, mechanical piping schematics and cooling P&IDs name the equipment long before it is installed. Errors can enter the record when information is copied out of those drawings into a list.

A drawing can establish less than people assume. A one-line diagram shows that a breaker exists, what it feeds and what its tag is. It often shows a rating. It does not show the serial number, the firmware version, or whether the unit installed is the unit that was drawn. A cooling schematic shows a chilled water pump, its tag and the loop it serves. It does not show that the pump was swapped for a different model during procurement. A list built from drawings reflects what those documents record. It does not, by itself, verify what is installed. Treating it as a verified record of installed equipment is an easy mistake to make because the two can look identical in a spreadsheet.

The practical fix is to keep document-derived information and field verification distinct. Every row extracted from the documents carries the drawing number and the revision it was read from. A second step, done on the floor or from commissioning records, confirms each row against the installed equipment and adds what only the field can know. When the two disagree, the record should show the disagreement, with both values, until someone resolves it. A field marked “not shown on the drawing” is more useful to a facilities team than a plausible value nobody can trace.

Revisions are where document-derived lists drift. Power and cooling drawings change through design, construction and commissioning, and a list typed from revision B is quietly wrong by revision E. The tag of a renumbered panel changes on the drawing and not in the list. A generator added for a later phase appears on the one-line and never reaches the asset register. The habit that prevents this is simple and tedious: every row keeps the revision it came from, and each new drawing issue is compared with the last one, tag by tag, so that additions, deletions and renumbering are found as a set and not discovered one at a time after handover.

Tags deserve their own care. The same piece of equipment is often named three ways: the designer’s tag on the drawing, the manufacturer’s designation on the nameplate, and the name the operations team wants in the maintenance system. Problems start when a list silently translates one into another. Keep the tag exactly as the drawing prints it, hyphens and all, and add the other names beside it as separate fields. Anyone who later has to trace a work order back to a drawing will be able to.

Uncertainty should be written down, not smoothed over. Scanned drawings have characters that cannot be read with confidence. A tag may be partly hidden by a revision cloud, or a rating may be legible on one sheet and different on another. The person preparing the list will be tempted to pick the likely answer and move on. A better record marks the row as unconfirmed, says why, and points at the sheet. The commissioning team can then prioritize flagged rows while carrying out the required verification of the wider record.

A short checklist covers most of it. Every row names its source drawing and revision. Tags are recorded as printed. Document-derived information and field verification status are distinguishable. Unreadable or conflicting values are flagged. Each new drawing issue is compared against the list before the list is reissued.

The principle is straightforward. Every line in an equipment record should show where it came from and whether it has been verified. Data centers are built quickly and handed over on a date, and that record is one of the few things from the project that the operations team will still be using ten years later.

# # #

About the Author

Muaawia Jannat is the founder and CEO of Tagsight Technologies Inc., a Calgary-based company developing engineering software for instrumentation and controls teams. His work focuses on turning engineering drawings into structured equipment and instrument records that teams can review, trace to their source and use in downstream workflows.

The post An Equipment List Is Only as Useful as Its Source appeared first on Data Center POST.

TL;DR

• Power Readiness Scheduling: Grid interconnection wait times reaching five to ten years in many regions require maintaining a dedicated power-readiness schedule that tracks utility upgrades, energization permissions, and phased deployment dependencies.
• Rack Level Cooling and Water Strategy: Facility engineering must look beyond hall averages to accommodate high rack densities with direct liquid cooling, traceable heat paths, and site-specific water management plans.
• Shared Acceptance Decisions: AI platform, facilities, and energy procurement leaders must jointly approve site readiness under real-world operating conditions, explicitly pricing the financial risk of delayed or restricted operations.

# # #

Picture an AI project with everything lined up: accelerators secured, lease signed, installation window booked. Then the utility’s service date slips, or the cooling design cannot support the specified racks through a hot afternoon with a pump unavailable. The hardware exists, but the computing capacity the business expected does not. Buying the machines was never the same as buying the ability to run them.

That distinction should change how AI infrastructure is planned. Chip availability still matters, but a site can become the binding constraint long before its floor space is full. Power delivery, heat removal and workload commitments need to be settled together, while there is still time to change the location or deployment plan.

Even without pinning every data center kilowatt-hour on AI, the scale is substantial. The IEA’s April 2026 outlook estimates global data center electricity consumption at about 485 terawatt-hours in 2025 and projects roughly 950 TWh in 2030, nearly double. Those totals cover data centers broadly; the report separately projects electricity use by AI-focused facilities to triple over that period. The 2030 figures are projections, not measured demand.

Start with a deliverable service date

A power discussion needs a location, a date and operating conditions, not just a megawatt figure. In its 2026 report, the IEA identifies grid-connection waits reaching five to ten years in many jurisdictions. That is a warning about constrained markets, not a universal lead time to paste into every project schedule.

Queue statistics also require care. Berkeley Lab’s 2026 Queued Up analysis reports a median exceeding five years from interconnection request to commercial operation for projects completed in 2025, in regions with available data. It tracks generation and storage projects, not data center load connections. Treating that number as the average wait for an AI campus would misstate the evidence.

The rules for connecting large loads are also in flux. In June 2026, FERC directed six regional grid operators to justify or revise their tariffs for large loads, addressing studies, cost allocation, co-location and flexible service. That action was not an energization approval for any individual campus. Developers still need to establish what their particular service agreement permits and what infrastructure must be completed first.

My recommendation is to maintain a power-readiness schedule alongside the construction schedule. It should name the dependencies for each deployment phase: studies, utility upgrades, equipment delivery, protection testing and permission to energize. Ask which dates are commitments, which are estimates and who carries the cost of slippage. A smaller first phase with power you can count on may beat a larger phase whose power date is still conditional.

Design around the demanding rack rather than the average hall

Once electricity reaches the building, the challenge becomes more local. Two hypothetical halls could have identical IT megawatts while placing very different demands on individual rows, electrical branches and cooling loops. Averaging across the hall would hide that difference. Require the proposed rack layout and expected workload profile before accepting a cooling-capacity claim.

Direct liquid cooling moves heat into fluid close to where it is generated, rather than first moving it through room air. The Department of Energy’s 2024 design guide describes its ability to serve higher heat densities. But selecting liquid-cooled servers is not the same as specifying the facility’s entire heat-removal system.

Temperature compatibility deserves particular attention. ASHRAE’s liquid-cooling guidance explains that increasing component power can require lower supply-water temperatures, potentially reducing opportunities to cool without mechanical refrigeration. A loop suitable for one hardware generation is therefore not automatically suitable for the next. Validate the actual equipment envelope rather than assuming that “liquid-ready” settles the question.

For design reviews, I would ask for a traceable heat path from the chip to the outside environment. Which components remain air-cooled, what temperatures and flow rates are required, and how does capacity change during maintenance? Include coolant distribution units, pumps, piping, controls and leak response in commissioning scenarios. A cooling system should earn its operating limits through testing, not through a label on the sales specification.

Keep efficiency claims within their measurement boundaries, too. DOE defines power usage effectiveness, or PUE, as annual facility energy divided by annual IT energy; it measures infrastructure overhead, not useful computing output. A favorable annual PUE does not establish peak electrical or thermal capacity. For an AI deployment, I would review it alongside completed work, service quality, total energy and peak demand.

Water belongs in the siting decision

Water accounting has two distinct operational boundaries. Berkeley Lab’s 2024 U.S. data center report estimates 2023 direct water consumption at about 66 billion liters and the indirect footprint from electricity at nearly 800 billion liters. These are sector-wide estimates, not AI-only measurements. The indirect estimate uses regional electricity mixes and excludes facility-specific power-purchase agreements and behind-the-meter generation.

Liquid cooling does not, by itself, tell you how much water a facility consumes. In December 2024, Microsoft described a new design using recirculating water without evaporation for cooling, while acknowledging increased energy demand when mechanical cooling replaces evaporation. That was a design announcement, not evidence that its whole fleet had eliminated cooling-water use. Nor does zero evaporation at the site erase water consumption associated with electricity generation.

The local source matters as much as the annual total. Google’s published water-risk framework considers community demand, available supply, delivery infrastructure and changing climate conditions. It describes choosing air cooling for Mesa, Arizona, after identifying elevated risk to the proposed water source. That is a site-specific decision, not a universal endorsement of one cooling architecture.

For a new project, request a water plan before making the final siting commitment. Separate withdrawal from consumption, identify potable and reclaimed supplies, and test the plan against seasonal restrictions and extreme heat. Ask what happens to electrical demand if water availability falls. A design that saves electricity by using water needs a credible operating plan for the days when that water is hardest to obtain.

Buy flexibility only when the workload can deliver it

There is public evidence that some computing demand can move. In August 2025, Google announced utility agreements targeting machine-learning workloads for demand response and reported reducing associated demand during three grid events in 2024. It also described shifting non-urgent tasks. These are specific operator disclosures, not proof that every AI workload can be interrupted economically.

I would qualify flexibility workload by workload. A batch job with deadline slack might be deferrable; an interactive service with a tight response-time commitment may not be. For training, examine checkpoint overhead, restart time and the cost of leaving accelerators idle. Moving work elsewhere also requires spare capacity, sufficient connectivity and permission to move the relevant data.

The useful deliverable is a tested operating specification: how many megawatts can be reduced, with how much notice, for how long, and at what recovery cost? Measure the reduction at the facility boundary rather than assuming that an application-level pause produces an equal grid benefit. Test the recovery period as well; deferred work still needs somewhere and some time to run. Do not promise flexibility to an energy supplier before the service owners have accepted those constraints.

Make site readiness a shared acceptance decision

Before approving an accelerator purchase, I would bring the AI platform lead, facilities engineer and energy-procurement lead to the same review. Their shared acceptance record should answer three questions:

• What capacity can be delivered at commissioning, and which electrical or thermal conditions reduce it?
• What evidence supports the next expansion date, and what happens if a dependency slips?
• Who owns each unresolved constraint, including decisions that cut across facilities, software and energy procurement?

Compare candidate locations against the same workload, including the earliest usable power date, supported rack envelope, water restrictions, curtailment obligations and network constraints. Then price a delayed deployment and a restricted operating year, rather than comparing only steady-state electricity costs. If the preferred site wins only when every upgrade arrives on its earliest estimated date, examine the downside before discarding the alternative. This comparison also gives the business a concrete choice: wait, deploy less, change the workload or move.

The central planning question is straightforward: what work can this site reliably deliver under the conditions in which it will actually operate? A hardware count cannot answer it. Neither can an annual energy contract, a good PUE number or a promise of future cooling upgrades. AI capacity becomes real when those pieces work together.

The views expressed here are my own. This article does not represent the views, policies, practices, or technology stack of my employer, LinkedIn. All examples are based on public sources or explicitly hypothetical scenarios. 

# # #

About the Author

Silu Panda is a staff software engineer at LinkedIn, focused on AI infrastructure and model serving. His interests include reliable inference, distributed systems, and open source. He writes about evaluating, deploying, and operating AI applications. 

The post AI Infrastructure Is Running Into Power Grid and Cooling Limits appeared first on Data Center POST.

TL;DR

• Expanded network reach: Enables service providers to deliver private Layer 2 Ethernet connectivity beyond their existing fiber footprints without additional fiber construction.
• Automated connectivity: Pairs with the MaiaEdge PBC 2000 to establish encrypted Ethernet connections over available fiber, dedicated internet access, satellite and other IP transport.
• New service opportunities: Supports cloud on-ramps, distributed AI access and additional services through existing customer connections.
• Faster customer onboarding: Reduces manual network configuration for service providers, data center operators and AI infrastructure companies.

###

MaiaEdge has introduced its PBC 100 Series, compact customer premises equipment (CPE) designed to help fiber operators and service providers extend private Ethernet connectivity beyond their existing fiber networks. The new solution enables operators to reach off-net customers without additional fiber construction or manually negotiated network-to-network interfaces (NNIs).

Part of MaiaEdge’s Path Border Controller (PBC) product family, the PBC 100 Series works alongside the company’s PBC 2000, deployed at an operator’s point of presence (PoP), Metro Ethernet edge or data center. Together, the devices enable automated, encrypted Layer 2 connectivity across available fiber, dedicated internet access, satellite and other IP transport.

Built on MaiaEdge’s SD-Ethernet architecture, the PBC 100 Series extends automated path provisioning directly to customer premises. A centralized interface simplifies routing complexity and provides operators with visibility across the connection, reducing the time typically required to establish services from weeks or months.

“For decades, where your fiber ended, your service opportunity ended with it,” said Abilash Menon, CEO and Co-Founder of MaiaEdge. “The PBC 100 Series breaks that link. Providers can reach off-net customers without building fiber or negotiating a new NNI, then turn that connection into private connectivity, cloud access, AI infrastructure, and more. We’re giving operators a way to expand their addressable market without expanding their physical network.”

The PBC 100 Series also enables operators to introduce additional services through the same customer connection, including cloud on-ramps and distributed AI access. For data center and AI infrastructure operators, the technology can accelerate tenant onboarding by reducing manual network configuration while creating opportunities to deliver services through their own networks or MaiaEdge-enabled partners.

The launch expands MaiaEdge’s Path Border Controller portfolio, allowing operators to extend private Ethernet services directly to customer locations while maintaining their existing physical network infrastructure. The first device in the series, the PBC 124-C, is available now.

To learn more about MaiaEdge and the PBC 100 Series, visit maiaedge.io

The post MaiaEdge Introduces PBC 100 Series to Extend Private Ethernet Beyond Fiber Networks appeared first on Data Center POST.

TL;DR

• Growing Facility Complexity: Rapidly rising rack densities and liquid cooling demands are outpacing traditional construction and labor availability.
• Cascading Delay Costs: A missed milestone affects financing costs, contractual penalties, and future capacity commitments.
• Discipline Around Time: Achieving completion certainty requires early risk visibility and treating schedule management with the same discipline as financial capital.

# # #

Data center development has always been complex, but what has changed is the scale.

During a recent Yotta panel, Dr. Atif Ansar, Executive Chairman and Co-Founder of Foresight, alongside industry leaders, discussed how projects are becoming larger, denser and more technically demanding while many of the resources required to deliver them have not expanded at the same pace. Skilled labor remains constrained, power is increasingly difficult to secure, equipment lead times remain significant and infrastructure requirements continue to evolve.

As a result, the consequences of a missed milestone are growing. A delay rarely stays isolated to one part of a project. It can ripple across contractors, suppliers, customers, financing and future capacity.

Complexity Is Increasing Faster Than Delivery Capacity

Not long ago, a 100 or 150 MW data center represented significant scale. Today, the industry is developing campuses measured in hundreds of megawatts and even gigawatts. Rack densities are also climbing rapidly, while direct liquid cooling and other new infrastructure requirements are reshaping facility design.

Meanwhile, the labor pool has not grown at the same rate as construction demand. Supply chains and power availability add further pressure.

This leaves less room for delays, design changes or slow decision-making.

Managing the Entire Delivery Chain

Data center schedules extend well beyond construction. Land, power, design, procurement, manufacturing, logistics, commissioning and customer readiness all contribute to the final ready-for-service date.

Panelists discussed how vertical integration, modular construction and offsite manufacturing can help reduce field labor and provide greater control over schedules. But improving delivery also requires stronger coordination between owners, contractors, customers and suppliers.

For owners, that does not mean managing every detail of construction. It means maintaining visibility into the milestones that matter and understanding what needs to happen today to protect a delivery date months into the future.

Time Has a Direct Financial Impact

The consequences of delay extend beyond construction costs. Facilities that come online late can face lost revenue, contractual penalties, increased financing costs and delayed customer capacity.

During the panel, Dr. Atif Ansar illustrated the potential impact using a 100 MW data center, estimating approximately $20 million in monthly revenue exposure before additional delay-related costs are considered.

That makes early visibility critical. The sooner teams identify schedule risk, the more opportunity they have to adjust resources, resequence work or coordinate with customers and suppliers.

Delivery Certainty Is Becoming a Competitive Advantage

The industry cannot eliminate uncertainty. Technology will continue to evolve, labor and power constraints will remain, and supply chains will face disruption.

The opportunity is to manage those challenges earlier and more effectively.

Successful delivery will increasingly depend on shared visibility, faster decision-making and stronger alignment across the project ecosystem. At this scale, time deserves the same discipline and attention organizations already give to capital.

The post Why Data Center Scale Is Raising the Stakes for Project Delivery appeared first on Data Center POST.

TL;DR

• More than 1,500 new data centers are in various stages of development across the U.S., but developers are encountering growing opposition related to electricity costs, water and energy use, environmental impacts and artificial intelligence.
• States and municipalities are responding with new requirements, moratoriums and bans, forcing developers to reconsider when and how they engage local communities.
• Industry leaders say transparency, early outreach and tangible community benefits are becoming essential to advancing responsible data center projects.

###

More than 1,500 new data centers are in various stages of development across the United States, but growing community opposition is changing how projects move forward. According to Inside Towers Intelligence’s 2026 Volume 3 Digital Infrastructure Industry: Market Analysis Report, concerns about electricity costs, energy and water use, environmental impacts, and artificial intelligence are driving greater public scrutiny of proposed developments.

Public resistance is widespread. A Gallup poll conducted in March 2026 found that seven in ten Americans opposed data center construction in their local communities including more than half who strongly opposed it. Environmental concerns are particularly significant with 55% of U.S. adults saying they worry a great deal about data centers’ environmental impact and another 23% expressing a fair amount of concern.

Government officials are responding with a growing patchwork of regulations, moratoriums, and bans. Several states have considered restrictions on new developments, while numerous municipalities have enacted or explored local pauses. These actions are already affecting project timelines and encouraging developers to reevaluate their site-selection and community-engagement strategies.

As opposition increases, early community outreach is becoming a prerequisite for development. Instead of securing land, power, fiber, and water before speaking with residents, developers are increasingly engaging local leaders and communities earlier in the planning process. These conversations allow companies to explain their projects, address local concerns, and identify benefits that can be delivered to the surrounding area.

Several developers are pairing transparency with tangible community investments. Examples highlighted in the report include support for schools, roads, workforce-development programs, and environmental initiatives. Duos Edge AI also hosts open houses and guided facility tours that allow residents, educators, officials, and businesses to see its modular edge data centers and ask questions directly about their size, cooling systems, and electricity requirements.

Clearer communication will remain central as the industry responds to concerns about power, water, jobs, and the environment. Ilissa Miller, CEO of iMiller Public Relations, recommends approaching community engagement more like an election campaign than a traditional project-marketing effort. She also advocates for engaging communities before filing conditional-use applications and providing clear information about what developers can offer at each stage of the process.

The report indicates that responsible development will increasingly depend on transparency, thoughtful planning, and measurable community value. Operators that engage early, correct misinformation, and work constructively with residents and local officials will be better positioned to move projects forward. Explore the Inside Towers Intelligence report for additional insights into the community and regulatory pressures shaping data center development.

The post Early Community Engagement Becomes Critical as Data Center Opposition Grows appeared first on Data Center POST.

TL;DR

• Unified Infrastructure Platform: MOD Mission Critical’s PaaSPort consolidates fragmented vendor relationships for colocation, bare metal, connectivity, and cloud access into a single global platform.
• Comprehensive Global Services: The platform offers on-demand hardware, IP transit scaling from GigE to 10G across six continents, virtual hosting, IPv4 space, and network monitoring through an open provider ecosystem.
• Addressing Edge Expansion: Unified edge solutions help enterprises manage growing complexity as the global edge computing market expands from $84 billion to over $1.5 trillion by 2035.

# # #

Written by iMiller Public Relations on behalf of MOD Mission Critical.

Traditional digital infrastructure deployment often forces organizations to stitch together a patchwork of separate vendors for colocation, connectivity, bare metal, and cloud access. Managing these isolated services introduces vendor sprawl, multiple contracts, tracking separate SLAs, and coordinating complex escalation paths during outages. To solve these operational challenges, MOD Mission Critical developed PaaSPort, a unified platform designed to bring global infrastructure capabilities together under a single ecosystem. Building on capabilities developed since 2011, MOD enables businesses to explore, compare, and price global infrastructure options through a single partner.

The PaaSPort platform provides access to a comprehensive suite of infrastructure services across six continents. Organizations can deploy on-demand physical bare metal hardware, vendor-agnostic colocation procurement, and IP transit scaling from GigE to 10G. Additionally, the platform covers virtual hosting, dedicated hosting, IPv4 services, managed proxy, and network monitoring. By offering an open ecosystem for both customers and infrastructure suppliers, the platform streamlines multi-market sourcing into a single procurement framework.

Simplifying distributed infrastructure management is increasingly vital as enterprise edge deployments expand worldwide. Industry projections indicate that over 60 percent of enterprises globally will adopt unified edge frameworks by 2027 to control operational complexity. Furthermore, the global edge computing market is projected to surge from approximately $84 billion today to over $1.5 trillion by 2035. As infrastructure requirements grow across global markets, unified sourcing platforms help organizations scale their footprints without accumulating administrative friction.

To read the original article, please click here.

The post PaaSPort: Bringing MOD’s Global Infrastructure Services Into One Platform appeared first on Data Center POST.

TL;DR

• Grid Constraints Suppress Pipeline: Announced US data center power demands exceed peak grid capacity, but interconnection delays will prevent most projects from being built
• Behind-the-Meter Power Shift: With 40 percent of AI data centers facing power limits by 2027, operators are turning to on-site generation to bypass long queue times
• Blind Efficiency Verification: Skepticism over unverified vendor energy claims requires predictions against withheld operational data to prove performance

# # #

Written by iMiller Public Relations on Behalf of Physical Superintelligence (PSI).

Announced data center projects in the United States total roughly 780 gigawatts, surpassing the country’s all-time peak load of 759 gigawatts set in July 2025. Although market demand for compute power remains high, PSI’s Scott Charter sheds light on the factors keeping a significant portion of this planned pipeline from materializing; chief among them, critical power availability and interconnection bottlenecks. Across nearly every market outside China, the central obstacle to expanding compute capacity lies in grid transmission backlogs rather than a willingness to construct power generation. 

To avoid multi-year interconnection queues, expansion in terrestrial compute through 2029 will increasingly adopt behind-the-meter, on-site generation. This transition aligns with projections showing that 40 percent of existing AI data centers will be operationally constrained by power by 2027, with global compute infrastructure likely hitting a definitive power wall within five to seven years. 

Simultaneously, vendor energy efficiency claims face widespread skepticism due to a market for lemons dynamic, as unverified percentage metrics prompt buyers to discount assertions to zero. Overcoming this skepticism requires rigorous verification, such as submitting predictions against withheld data from specific halls or operational periods before those datasets are revealed

To read the original article, please click here.

The post America Has Announced More Data Centers Than It Has Grid appeared first on Data Center POST.

TL;DR

• Customer continuity: Service at Columbus continues without interruption under a revised five-year agreement.
• Expansion focus: Redirects capital from GPU ownership toward new modular AI colocation sites.
• Continued service: Keeps the Columbus customer supported without interruption under an enhanced five-year agreement.
• Financial flexibility: Eliminates approximately $98.1 million in prospective GPU equipment financing, preserving capacity for site development.

# # #

Duos Technologies Group, Inc. (“Duos” or the “Company”) (Nasdaq: DUOT) has completed the sale of its GPU-as-a-Service entity, Duos Edge AI – GPUaaS, LLC, to Axe Compute, Inc. (Nasdaq: AGPU). The transaction closed on September 30, 2026, advancing Duos’ shift toward a focused AI colocation platform.

Duos will continue to own and operate its Columbus, Georgia colocation facility, providing power, cooling, security, and managed infrastructure services. Rather than owning and financing the GPUs, the Company will lease the underlying compute capacity from Axe Compute. Its existing Columbus customer will continue receiving service without interruption under a revised five-year agreement expected to increase the revenue Duos recognizes over the contract’s life.

Under the agreement, Axe Compute acquired all outstanding equity interests in the entity holding Duos’ GPU compute cluster and associated equipment supply and financing positions. Duos will receive $42.9 million through evenly scheduled monthly payments over the next 60 months, while approximately $98.1 million in prospective equipment financing obligations have been removed.

“Selling the GPU fleet sharpens Duos into a pure-play edge data center operator,” said Chris DeAlmeida, Duos CFO. “We keep the customer and the economics, we shed roughly $98.1 million of prospective equipment debt, and we free the balance sheet to fund sites instead of servers. Our outlook for 2026 is unchanged: full year revenue above $50 million and positive adjusted EBITDA.”

The transaction allows Duos to direct capital toward developing and operating 10 to 30 MW modular colocation sites across the United States. Its broader strategy centers on owning the sites, power capacity, and infrastructure that support AI workloads, while customers bring and operate their compute equipment.

The sale reflects Duos’ continued focus on building long-lived colocation assets and reinvesting capital into new sites, campuses, and customer deployments.

To learn more about Duos Technologies Group, visit duostechnologies.com.

The post Duos Technologies Group, Inc. Sells GPU-as-a-Service Entity to Axe Compute appeared first on Data Center POST.

TL;DR

• Introducing SD-Ethernet: Software-Defined Ethernet transforms Ethernet into an automated, cross-domain service that is encrypted by default and programmable via APIs.
• MaiaEdge Architecture: By pairing Path Border Controllers with a cloud-based Path Computation Engine, MaiaEdge slashes service activation times from 60 to 90 days to seconds while offering real-time telemetry across segments.
• Built for AI Demands: SD-Ethernet enables enterprises to instantly extend their private network directly to remote GPU clusters and clouds, converting static network pipes into a dynamic platform for AI workloads.

# # #

Written by iMiller Public Relations on behalf of MaiaEdge.

Over five decades, the networking industry introduced a succession of protocols and architectures, from ATM and MPLS to SD-WAN and SASE, to address complex connectivity, routing, and security challenges. Despite these evolving abstractions and overlay networks, Ethernet remained the universal foundation at the port level, natively used by servers, switches, clouds, and GPU clusters. However, traditional Ethernet services suffered from severe operational bottlenecks, requiring manual cross-connect contracts, spreadsheets, and lengthy negotiations that delayed multi-domain provisioning for 60 to 90 days.

To eliminate these legacy friction points, Software-Defined Ethernet (SD-Ethernet) introduces a programmable, automated layer that extends secure Ethernet handoffs across independent operators, fiber networks, and cloud providers. Utilizing MaiaEdge’s Path Border Controllers at network boundaries and a cloud-based Path Computation Engine, the architecture dynamically computes cross-domain routes, encrypts traffic end-to-end, and measures performance along every segment. This approach slashes connection activation times from months down to seconds while delivering complete path visibility, allowing organizations to pinpoint the exact provider domain responsible for packet loss.

This paradigm shift is heavily driven by the demands of AI infrastructure, where distributed training models, inference nodes, and sovereign datasets require fast, private, and flexible interconnections. Rather than forcing enterprises to move into entirely new networks or manage fragmented security stacks, SD-Ethernet allows existing private networks to seamlessly extend wherever GPU resources reside. Consequently, networking transitions from static, slow-to-provision pipes into an agile platform built for high-performance AI data movement.

To read the original article, please click here.

The post Why Ethernet Never Left: A 50-Year Networking Mystery appeared first on Data Center POST.

TL;DR

• Broadband strategies are expanding beyond network construction to address adoption, affordability, digital skills, and meaningful use.
• Community-led partnerships are helping align infrastructure investment with local priorities, including Tribal, rural, and historically underserved markets.
• Public and private funding must be structured around long-term operations, accountability, and sustainability, not simply initial deployment.
• Multifamily connectivity decisions increasingly depend on resident expectations, practical technology planning, policy requirements, and measurable performance.

# # #

Broadband Communities Summit 2026 placed a clear emphasis on what happens after infrastructure reaches a community. Held August 25-27 in Houston, Texas, the event examined how broadband providers, municipalities, Tribal organizations, property owners, investors, and nonprofit leaders can turn network availability into reliable, affordable, and useful connectivity. Across the Communities, Finance & Frameworks, Infrastructure and Multifamily tracks, the agenda reflected an industry that is balancing ambitious deployment goals with the practical work of adoption, operations, and long-term value.

The summit also highlighted that broadband challenges vary widely by market. Rural communities may need flexible technology and funding models, multifamily owners must weigh resident experience against infrastructure readiness and regulatory demands, and local leaders must build trust among providers, utilities and community organizations. The strongest strategies discussed in Houston treated connectivity as an ecosystem, one that depends on capital, technology, policy, and community participation working together.

Community Needs Must Shape Broadband Strategy

Access remained essential, but summit programming repeatedly looked beyond coverage maps. Sessions explored affordability, device availability, digital literacy, and the role of local organizations in helping residents use broadband for education, healthcare, employment, and economic opportunity. That broader view positions adoption and measurable community benefit as core parts of network planning rather than secondary considerations after construction.

In the keynote “Digital Justice: Reimagining Connectivity for All,” Francella Ochillo, Adjunct Professor and Consultant at Johns Hopkins University, addressed the evolving broadband landscape and the strategic approaches required to overcome persistent industry hurdles. The event’s community-focused sessions reinforced the same principle: successful broadband initiatives begin with accurate local data and sustained engagement, then connect infrastructure decisions to the needs of the people the network is intended to serve.

Local and Tribal Partnerships Create New Infrastructure Pathways

Partnership models were especially prominent in sessions focused on rural, regional and Tribal connectivity. Rather than relying on a single organization to finance, build and operate every component, communities are assembling coalitions that combine local knowledge, public resources, private-sector expertise and shared infrastructure. These arrangements can expand access while giving communities a stronger voice in how networks are designed and governed.

Joel Ogren, Chief Executive Officer of Assured Communications, joined Tyson Johnston of Toptana Technologies for “Building Connectivity from the Coast,” a presentation on a Tribal partnership advancing subsea and backhaul infrastructure. The session illustrated how major connectivity projects can link technical deployment with regional priorities. It also highlighted the value of partnership when projects must coordinate difficult geography, long-haul infrastructure and the economic goals of the communities involved.

Financing Must Support the Full Network Lifecycle

The summit’s finance programming made clear that securing capital is only one part of a viable broadband plan. Workshops and panels addressed private equity expectations, commercial debt, public-private partnerships, grant compliance, and approaches to layering public and private investment. Across these topics, the central question was whether a project’s financial structure could support construction, ongoing maintenance, service quality, and future growth.

Kyle Zimbelman, Chief Operating Officer of Golden State Connect Authority, participated in “Public-Private Partnerships in Practice,” which examined how partners can align state, local, and provider priorities while balancing accountability, flexibility, and speed. The discussion framework emphasized careful partner selection and sustainable operating models which are important considerations as communities move from funding awards and project announcements to execution and long-term stewardship.

Community-Led Models Can Turn Ambition Into Adoption

Several sessions offered practical examples of communities pursuing new approaches where conventional deployment models had fallen short. These case studies showed that local leadership, coordinated outreach, and a clearly defined public mission can help projects navigate political, financial, and operational barriers. They also demonstrated why success should be evaluated through adoption and community outcomes, not only through miles of fiber or premises passed.

Joshua Edmonds, CEO of DigitalC, presented “From Impossible to Improbable to Impactful: Exploring the Rise of the Cleveland Model.” His session traced the development of an approach designed to deliver quality and affordability in one of the nation’s least-connected cities. The Cleveland experience provided a useful example of how a community-centered operating model can evolve when technology, political commitment, and execution capacity are built around a long-term effort to close the digital divide.

Multifamily Connectivity Requires Practical, Data-Informed Decisions

The Multifamily track addressed another rapidly changing part of the broadband market. Property owners and service providers examined Wi-Fi standards, resident expectations, bulk broadband and video adoption, implementation challenges, fee transparency, and the operational effects of new policies. The sessions reflected a growing need to translate technology and regulation into decisions that make sense at the property level.

Elizabeth Parks, President and CMO of Parks Associates, contributed to the Multifamily Town Hall and a data-focused session on resident satisfaction and adoption trends. That market perspective supported one of the track’s clearest messages: owners need realistic performance benchmarks and credible adoption data when planning conversions, evaluating infrastructure upgrades, and assessing the resident experience. Technology choices create lasting value when they match building conditions, operational capacity and demonstrated customer needs.

Building Broadband That Lasts

Broadband Communities Summit 2026 reflected an industry entering a more demanding phase. Funding programs and new technologies continue to create opportunities, but lasting results depend on whether communities and providers can coordinate deployment, earn trust, support adoption, and operate networks effectively over time. The most durable plans will connect financial discipline and technical performance with clearly defined community outcomes.

As broadband stakeholders move forward, the Houston discussions offer a useful standard for evaluating progress: infrastructure matters, but its value is ultimately measured by the people, properties, and local economies it serves. To learn more about Broadband Communities Summit and future events, visit the official Broadband Communities Summit website.

The post Broadband Communities Summit 2026 Highlights the Shift From Network Access to Lasting Community Impact appeared first on Data Center POST.

TL;DR

• Traditional backup is no longer sufficient for modern data centers due to rapid data growth, complex hybrid IT environments, slow recovery times, and sophisticated ransomware threats.
• Resilient protection enhances core frameworks like the 3-2-1 backup strategy by incorporating immutable storage, air-gapped backups, encryption, and automated recovery testing.
• Purpose-built appliances simplify operations by combining storage hardware, software integration, and hardened security into a single platform for faster deployment and reliable recovery.

# # #

Data centers are the foundation of modern digital businesses. From financial services and healthcare to SaaS platforms and manufacturing, organizations depend on always-available infrastructure to run critical workloads.

However, as data volumes continue to grow and cyber threats become more advanced, traditional backup approaches are no longer enough. Organizations need backup strategies that provide fast recovery, ransomware protection, operational simplicity, and long-term data availability.

A modern data center backup strategy must address not only hardware failures but also cyberattacks, accidental deletion, insider threats, and increasingly complex hybrid IT environments.

The Growing Challenges Facing Data Center Backup

Data protection has become more complicated because organizations now manage data across multiple environments, including:

• On-premises data centers
• Private clouds
• Public cloud platforms
• Virtualized environments
• Remote offices and edge locations

At the same time, ransomware attacks have changed the requirements for backup infrastructure. Simply having copies of data is no longer enough. Organizations need protected backups that attackers cannot easily modify or encrypt.

This shift has moved backup from being a simple storage function to becoming a critical component of business continuity planning.

Why Traditional Backup Methods Are No Longer Enough

Legacy backup systems were designed primarily for recovery after hardware failures or accidental data loss. While they remain useful, they often lack the capabilities required for modern threats.

Common limitations include:

Slow Recovery Times

Large enterprises may need to restore terabytes or even petabytes of data after an incident. Traditional backup systems can result in long recovery windows, increasing downtime costs.

Limited Protection Against Ransomware

If backup copies remain connected and writable, attackers may target them during an attack. Organizations increasingly require immutable and isolated backup copies.

Complex Management

Many enterprises operate multiple backup tools across different environments. This creates management overhead and increases the possibility of configuration errors.

Building a Strong Data Backup Strategy

A reliable backup architecture should follow established data protection principles. One widely adopted approach is the 3-2-1 backup strategy, which recommends maintaining three copies of data, storing them on two different types of media, with one copy kept offsite.

This approach helps organizations reduce risks caused by hardware failures, disasters, and cyber incidents.

However, modern enterprises often extend this model by adding additional protection layers such as:

• Immutable storage
• Air-gapped backups
• Encryption
• Automated recovery testing
• Continuous monitoring

The Role of Purpose-Built Backup Appliances

As data environments become more complex, many organizations are moving toward purpose-built backup infrastructure instead of relying only on general-purpose storage systems.

A backup appliance combines storage hardware, backup software integration, and data protection features into a single platform.

For data centers, this approach provides several advantages:

• Faster deployment
• Simplified management
• Optimized backup performance
• Better integration with enterprise applications
• Improved recovery reliability

Purpose-built solutions can also support advanced protection features such as immutable backups and ransomware-resistant storage architectures.

Immutable Backup for Modern Data Protection

One of the biggest changes in enterprise backup is the adoption of immutable storage.

Immutable backups prevent stored backup data from being altered or deleted during a defined retention period. This capability is especially important for organizations facing ransomware threats.

For environments using Veeam, organizations often deploy a Veeam immutable backup appliance to create protected backup repositories that support secure recovery after cyber incidents.

Combining backup software with hardened storage infrastructure creates a stronger defense against modern threats.

Disaster Recovery Requires More Than Backup Copies

Backup and disaster recovery are closely connected but serve different purposes.

Backup focuses on protecting copies of data, while disaster recovery focuses on restoring business operations after an outage or failure.

A complete disaster recovery strategy should consider:

• Recovery Point Objectives (RPO)
• Recovery Time Objectives (RTO)
• Application dependencies
• Infrastructure availability
• Geographic risks

Many organizations now use a disaster recovery appliance to simplify recovery processes and reduce downtime during unexpected events.

Cloud and Hybrid Recovery Models

Cloud infrastructure has introduced new possibilities for disaster recovery. Organizations can now combine local backup infrastructure with cloud-based recovery options.

A hybrid approach allows businesses to keep frequently accessed recovery data locally while using cloud resources for additional resilience.

However, cloud recovery strategies should still include careful planning around:

• Data transfer costs
• Recovery speed
• Compliance requirements
• Security controls

Cloud-based disaster recovery works best when integrated into a broader business continuity strategy.

Choosing the Right Backup Architecture for Your Data Center

Every organization has different requirements based on workload size, compliance needs, recovery objectives, and infrastructure design.

Before selecting a backup solution, IT teams should evaluate:

Data Growth

Rapidly increasing datasets require scalable backup infrastructure.

Recovery Requirements

Critical workloads may require near-instant recovery capabilities.

Security Requirements

Organizations should prioritize solutions supporting encryption, immutability, and access controls.

Operational Complexity

Simplified management helps reduce administrative overhead and improves reliability.

The Future of Data Center Backup

The future of backup will focus on automation, cybersecurity, and intelligent recovery.

Organizations are moving toward backup platforms that can:

• Detect unusual activity
• Automate recovery workflows
• Protect against ransomware
• Support hybrid environments
• Scale with growing data demands

For modern data centers, backup is no longer just a secondary storage function. It has become a core component of infrastructure resilience.

Conclusion

As organizations continue to depend on digital infrastructure, data protection must evolve beyond traditional backup methods. A modern approach combines proven strategies like the 3-2-1 backup strategy, immutable storage, disaster recovery planning, and purpose-built backup infrastructure.

By investing in resilient backup architectures, data centers can improve availability, reduce downtime risks, and maintain business continuity even during unexpected disruptions.

# # #

About the Author

George Williams, Senior Technical & Product Marketing Specialist, StoneFly Inc.

George Williams has over 13 years of experience in the data storage, backup and disaster recovery, air-gapping, immutability and archiving markets. A true geek with love for ease and simplicity in data storage, George has been working for StoneFly Inc. for over a decade. Ever since StoneFly started shipping products since 2006, George has been working to ensure that technical information is relayed in a simple and effective way to customers and targeted audience. George helps curate content and works with numerous publishers and technology blogs to spread awareness and knowledge of data storage technology.

The post Why Modern Data Centers Need a Resilient Backup Strategy Beyond Traditional Protection appeared first on Data Center POST.

TL;DR

• Equal Speeds Matter: Symmetrical internet delivers matching upload and download speeds, ensuring data sent out travels as fast as incoming data.
• Eliminates Bottlenecks: Strong upload capacity prevents video call freezing, slow file sharing, and background cloud backup delays when multiple family members are online.
• The Empire Advantage: Empire Fiber Internet provides equal upload and download speeds across connected household devices, and residents can check local address availability to upgrade.

# # #

Written by iMiller Public Relations on behalf of Empire Fiber Internet.

While traditional home internet plans prioritize download speeds for loading webpages and streaming shows, upload speed plays an equally critical role in household connectivity. Upload capacity governs everything sent from a home out to the internet, including participating in video calls, submitting school assignments, and backing up photos to the cloud. Symmetrical internet connections provide equal maximum speeds for both uploading and downloading data, delivering balanced bandwidth that fiber internet infrastructure inherently supports.

On busy school nights when multiple family members are online simultaneously, insufficient upload speeds create noticeable bottlenecks. Video calls can freeze or drop audio, group project uploads take longer, and automated cloud backups stall background productivity across smart household devices. Having equal upload and download capacity prevents high-bandwidth activities from interfering with one another, ensuring smooth performance for learning, working, and streaming at the same time.

Unlike traditional cable, DSL, or variable 5G wireless services that limit upload speeds, Empire Fiber Internet offers matching upload and download speeds to eliminate upload lag. This symmetrical capability ensures clearer video conferencing, faster file transfers, seamless photo and document backups, and reliable multi-device support for modern connected homes. Households can check service availability at their address to find an Empire Fiber Internet plan that matches their connectivity needs.

To read the original article, please click here.

The post Why Symmetrical Internet Speeds Matter for Your Home appeared first on Data Center POST.

TL;DR

• Expanding a Strategic Investment: Ares Secondaries funds have increased their investment in Sabey Data Center Properties (SDCP), bringing their total commitment to the platform to more than $500 million.
• Supporting Long-Term Growth: The additional investment builds on Ares’ initial minority equity investment announced in July 2026 and supports SDCP’s continued growth as demand for digital infrastructure increases.
• Scaling for Future Demand: With 275 MW across its existing campuses and an expansion pipeline expected to reach approximately 737 MW by 2033, SDCP is positioned to continue serving growing cloud and AI infrastructure needs.

# # #

As demand for digital infrastructure continues to grow, Sabey Data Center Properties (SDCP) is building on its long-term strategy with an expanded investment from Ares Secondaries funds.

The additional commitment follows Ares’ minority equity investment in SDCP announced in July 2026 and brings its total investment in the platform to more than $500 million. The follow-on investment further strengthens the relationship among Sabey Corporation, National Real Estate Advisors and Ares, bringing together three long-term partners focused on supporting SDCP’s continued growth.

SDCP currently owns and operates hyperscale and enterprise data center campuses in top-tier U.S. data center markets totaling 275 megawatts (MW). Looking ahead, the company’s expansion pipeline is expected to reach approximately 737 MW by 2033, positioning the platform to support the next generation of cloud and artificial intelligence demand.

“This follow-on investment reflects continued conviction in SDCP’s long-term value proposition and growth trajectory,” said Kevin Verdi, Executive Vice President & Chief Investment Officer at National Real Estate Advisors. “We remain focused on disciplined growth and delivering the digital infrastructure needed to support the next generation of cloud and AI demand.”

For SDCP, the expanded investment reinforces confidence in a platform built around developing and operating mission-critical infrastructure for customers across the United States.

“We are pleased to deepen our relationship with Ares as we continue executing on our long-term growth strategy,” said Tim Mirick, President of Sabey Data Center Properties. “This additional commitment reflects confidence in the strength of our platform and our ability to develop and operate the mission-critical infrastructure our customers rely on.”

The investment also reflects Ares’ continued confidence in SDCP’s growth potential.

“We are pleased to expand our investment in SDCP and build on our relationship with the Sabey and National teams as we continue to see significant runway for growth,” said Jamie Sunday, Co-Head of Real Estate Secondaries at Ares.

With a portfolio of more than four million square feet of mission-critical space, SDCP is one of the largest privately owned multi-tenant data center owners, developers and operators in the United States. As the digital economy continues to evolve, the expanded Ares investment provides additional support for SDCP’s long-term growth and its ability to serve customers with scalable, mission-critical data center infrastructure.

The post Sabey Data Center Properties Builds on Growth Strategy with Expanded Ares Investment appeared first on Data Center POST.

TL:DR

• Growing Grid Constraints: Expanding AI and digital workloads have made reliable electric grid access a major bottleneck for new data center capacity
• Repurposing Stranded Infrastructure: Utilizing orphaned and end-of-life oil and gas wells via in-situ hydrogen generation enables clean, low-cost fuel production near major power loads
• Zero-Loss Liquid Storage: Integrating onsite production with liquefaction and active zero-loss storage eliminates fuel boil-off and preserves energy reserves until needed
• Independent & Resilient Dispatch: Converting stored hydrogen through fuel cells provides clean, dispatchable power for backup reserves, peak demand management, or primary generation independent of electric grids

# # #

Data centers face a fundamental power challenge. As AI, cloud computing, and digital services drive electricity demand, access to reliable power is becoming one of the biggest constraints on new capacity. At the same time, operators are under growing pressure to reduce emissions, improve resilience, and limit dependence on increasingly constrained electric grids. The answer may be closer than many developers realize: thousands of existing oil and gas wells that are no longer economically viable.

Across the United States, documented orphaned oil and gas wells represent a significant underutilized infrastructure resource. Rather than viewing these wells solely as environmental liabilities, they could serve as platforms for producing hydrogen near where energy is needed. A new approach to in-situ hydrogen generation provides a path to do exactly that.

Turning stranded wells into hydrogen resources

Traditional hydrogen production requires bringing feedstocks and energy to a centralized facility, then transporting hydrogen to the end user. In-situ hydrogen generation turns that model on its head. The process uses existing end-of-life hydrocarbon reservoirs and well infrastructure. Oxygen is injected through existing or newly developed wells, triggering controlled downhole reactions that oxidize residual hydrocarbons and produce a hydrogen-rich syngas. At the surface, hydrogen can be separated and directed to the application that requires it.

This approach has the potential to produce hydrogen at very low cost while making productive use of existing subsurface resources and infrastructure. The process can also provide a pathway to lower-carbon hydrogen, with carbon management strategies determining the ultimate carbon intensity.

For data center developers, the significance is not simply how hydrogen is produced. It is where it can be produced. Locating hydrogen production near major loads can reduce reliance on long-distance fuel transport and provide a dedicated energy resource for power generation. In regions where suitable wells are near growing electricity demand, this could create an entirely different model for developing data center power infrastructure.

From hydrogen production to reliable power

Producing hydrogen onsite is only part of the opportunity. The hydrogen must also be available when the data center needs it. That makes storage critical.

Liquid hydrogen offers a highly energy-dense way to store substantial quantities of hydrogen onsite. Advanced zero-loss storage systems can actively remove heat from the tank, preventing hydrogen from warming and converting to gas. Instead of allowing boil-off to become a fuel loss, the stored hydrogen remains available for future use. That changes the economics and operational value of hydrogen storage.

For a data center, stored hydrogen can serve as a strategic energy reserve to meet extended power needs when grid power is unavailable or constrained. Hydrogen can be converted back to electricity via fuel cells, providing a clean, dispatchable power source without the combustion emissions associated with conventional diesel generation. The result is a potential closed-loop energy architecture: produce hydrogen at or near the site, liquefy it, store it without losses, and convert it to electricity when the data center requires additional or backup power.

A different approach to data center power

The conventional model assumes the grid will supply most electricity and that backup generators will protect against outages. That model becomes more difficult as data center loads grow and grid interconnection timelines lengthen.

Hydrogen offers another option: build power resilience around a fuel that can be produced, stored, and dispatched independently of the grid. The opportunity is particularly compelling when three technologies are considered together. In-situ hydrogen generation can unlock stranded subsurface resources. Liquefaction can make hydrogen practical for high-energy-density storage. Zero-loss liquid hydrogen storage can preserve that fuel until it is needed.

Together, these technologies could transform otherwise unproductive wells into assets supporting critical digital infrastructure. The broader opportunity extends beyond backup power. As hydrogen production and storage systems scale, the same infrastructure could support primary or supplemental power generation, helping data centers manage peak demand, improve energy resilience, and reduce reliance on constrained grid capacity.

Data centers need more than electricity. They need reliable power, fuel security, scalability, and predictable operating costs. The next generation of data center development may therefore require a different question. Rather than asking only where the grid can deliver more power, developers should also ask where existing energy resources can be converted into new, resilient power sources.

Orphaned wells may be part of the solution. By combining in-situ hydrogen generation with liquefaction and zero-loss liquid hydrogen storage, an overlooked energy resource could become a foundation for powering the digital infrastructure of the future.

# # #

About the Author

Cullen Hall is Vice President of Product Development at GenH2, a subsidiary of Path2 Hydrogen AG and technology leader in liquid hydrogen infrastructure systems, including Zero-Loss Controlled Storage and advanced hydrogen liquefaction. He has more than 20 years of experience in engineering, cryogenics, and hydrogen technologies, specializing in liquid hydrogen infrastructure. His career includes leadership roles with Linde’s Praxair, Plug Power, and several hydrogen companies throughout Asia, with expertise in cryopumps, compressors, dispensers, and hydrogen refueling systems. Learn more at genh2.com.

The post Powering Data Centers with Hydrogen from an Overlooked Energy Resource appeared first on Data Center POST.

TL;DR

• Accountability Beyond Consumption: Data centers must be judged by the physical systems they replace as well as what they consume, requiring a testable impact ledger tied to a defined, specific service
• Full-Lifecycle Burden Accounting: The ledger must record direct impacts, including embodied materials, electricity, cooling, water, backup power, and local grid constraints, allocated across the entire delivery network
• Verified Downstream Outcomes & Rebound: Claimed environmental benefits must be proven through measured downstream physical changes rather than predictive models, crediting participants conservatively while factoring in rebound usage

# # #

My previous article argued that a data centre should be judged by what it replaces as well as what it consumes. That principle is useful only if it changes decisions. An operator seeking planning consent, a customer buying capacity and an investor assessing a campus need a ledger that can be tested after the promise is made.

Start with a defined service, not a general claim that digital infrastructure makes society more efficient. A building optimization workload, for example, can be assessed against a specified portfolio, period, service level and physical outcome. The comparison must be the same buildings delivering a comparable service without that intervention. An empty field is not a credible baseline for an operating digital service; neither is an implausibly wasteful building.

Put the burdens on the record

The direct account should cover the site’s construction, embodied materials, electricity, cooling, water, backup generation, hardware, networks, replacement and end of life. It should also identify local constraints. A contracted renewable supply does not by itself establish that the demand causes no additional burden on a constrained grid. Water consumption needs its catchment and seasonal context. A single power efficiency ratio cannot answer either question.

The record should be granular enough to allocate a reasonable share to the workload being assessed. That allocation will be uncertain in a shared facility, so publish the method and a range. The account cannot quietly exclude the sensors, access networks, edge equipment or validation systems needed to deliver the claimed service.

Measure the physical change

The benefit account begins downstream. Did the controls reduce measured building energy use after adjusting for weather, occupancy and operating hours? Did leakage detection result in repaired pipes and less water lost? Did predictive maintenance extend useful equipment life, or merely produce alerts? Separate an action taken from a model’s forecast of what might have happened.

Credit must be shared conservatively. A saving in a building depends on the owner’s investment, commissioning, facilities team, sensors, software, connectivity and computing. It cannot be claimed in full by every participant. The customer may report a reduction in its own inventory, while the service provider separately describes its contribution to an avoided-impact assessment. These are different accounts and should remain labelled as such.

Rebound belongs on the same page. Lower operating cost can increase use. New computing capacity can create demand that was absent from the baseline. If plausible additional demand reverses the result, the claim has not passed the decision test.

Give evidence an expiry date

A sound submission records the functional unit, baseline, measurement period, source data, allocation method, uncertainty and independent assurance status. Physical sensors require identity, calibration and maintenance records. An immutable digital record may protect provenance after data entry; it cannot make a faulty reading true.

The baseline will also change. Buildings are retrofitted, grids decarbonise, regulations tighten and alternative services improve. A claimed benefit should be reviewed against current conditions rather than carried indefinitely from an early pilot. The TUTUM framework proposes a review at least every three years and expiry after five unless the baseline is independently re-established. Those periods are a proposed governance rule, not an external reporting standard.

Use the ledger as a gate

For each impact category, show the direct burden, attributable avoided impact and rebound separately. Carbon, water, materials and local infrastructure effects require their own units and their own decision. A favourable carbon result cannot compensate for an unacceptable water or community burden by arithmetic.

Before approval, the parties should agree who owns each measurement, who can challenge the baseline, which claims may be used publicly, and what happens if measured performance falls short. Procurement can require evidence access and periodic recalculation. Planning and investment committees can distinguish demonstrated benefits from scenarios and place conditions on the latter.

The conclusion may be positive, negative or unresolved. That is the point of an honest ledger. It gives the industry a route to show where computing delivers a verified physical improvement, while making it harder to use a distant possible saving to excuse a present local impact.

# # #

About the Author

Mark Burns is Executive Director of TUTUM Group and a chartered development and infrastructure leader. His work focuses on asset governance, programme assurance, sustainability and the relationship between physical infrastructure and digital capital.

The post Data Centres Need a Net System Impact Ledger appeared first on Data Center POST.

TL;DR

• AI is reshaping data center operations, driving higher rack densities, increased power requirements, and a greater demand for liquid-to-chip cooling expertise.
• Resilience depends on more than  redundant infrastructure, requiring proven processes, real-time monitoring, strong documentation, reliable partners, and well-trained operational teams.
• Datalec Precision Installations (DPI) is helping customers navigate  these challenges across the full data center lifecycle, from deployment and fit-outs to maintenance and specialized cooling support

# # #

In episode 70 of NEDAS Live!, host Ilissa Miller sat down with Scott Longhurst, Data Center Services Director at Datalec Precision Installations (DPI), to discuss the evolving operational demands of AI-ready data centers. Drawing on more than 35 years in mission-critical environments, Longhurst explores what resilience truly means as facilities become more dense, power-intensive, and complex.

For Longhurst, resilience extends far beyond redundancy. It requires proven processes, strong documentation, effective monitoring, dependable supply chains, and well-trained teams that can respond intelligently as infrastructure and customer requirements change. However, the industry faces a growing workforce challenge as data centers are being built faster than skilled operations professionals can be trained.

Artificial intelligence is accelerating the shift toward higher-density deployments and liquid-to-chip cooling. Longhurst notes that rack requirements are rising quickly, putting new pressure on facility power, cooling, structural capacity, maintenance practices, and deployment timelines. Liquid cooling also introduces a new operational boundary between landlord-managed and customer-managed systems, making coordinated monitoring and clear accountability essential.

Proactive maintenance was another key focus of the conversation. Longhurst explains how a maintenance program should not simply follow a calendar. Operators need to use predictive monitoring, equipment trend data, lifecycle planning, and root-cause analysis to identify and address issues before they disrupt service. The approach he outlines can reduce risk, support customer commitments, and guide smarter investment in aging infrastructure.

The two also discussed the industry’s continued focus on efficiency. Higher allowable operating temperatures, more efficient UPS systems, variable-speed equipment, and updated cooling strategies are helping operators lower energy use while maintaining reliable performance. For DPI, the opportunity lies in supporting customers across the data center lifecycle, from fit-outs and modular deployments to commissioning, maintenance, and specialized AI cooling services.

Tune in to episode 71 of NEDAS Live! for the full discussion on how resilient operations, skilled teams, and proactive maintenance strategies will shape the data centers powering the AI era.

The post Building Resilient Data Centers for the AI Era with Scott Longhurst appeared first on Data Center POST.

TL;DR

• Shift to AI-Ready Infrastructure: Rising AI workloads are driving rack densities up to 227kW, requiring early alignment of power distribution, cooling, and connectivity.
• Execution and Repeatability: Navigating diverse APAC markets demands standardized deployment strategies paired with local regulatory and operational expertise.
• Quality Safeguards Speed: Rigorous installation, structured cabling, and comprehensive system testing prevent post-launch operational issues as facilities pre-lease rapidly.

 # # #

Written by iMiller Public Relations on behalf of Datalec Precision Installations (DPI).

Asia-Pacific’s data centre market is experiencing continued expansion, with its development pipeline reaching a record 26.5 GW in the first half of 2026 after growing by 7.1 GW in six months. However, roughly 82 percent of this total remains in the planning phase, compared to 4.8 GW actively under construction. As colocation vacancy rates fall and pre-leasing accelerates across the region, the industry’s primary bottleneck has shifted from securing demand to executing construction and delivery effectively.

Artificial intelligence is reshaping infrastructure requirements, placing greater power and thermal demands on facility design. High-density hyperscale deployments now reach rack densities as high as 227kW, requiring early coordination across power distribution, cooling, connectivity, and modular infrastructure. To navigate complex regional dynamics across different APAC markets, operators must combine global delivery standards with local market intelligence, utilizing standardized deployment procedures to ensure consistent quality and scalability.

Speed to market cannot compromise operational reliability. Maintaining strict quality controls during equipment installation, cable management, labeling, and integrated system testing protects long-term facility performance. Ultimately, the region’s digital ecosystem will be defined not by the sheer size of its planned megawatts, but by its capability to convert complex development pipelines into resilient, AI-ready operational capacity.

To read the original article, please click here.

The post From Pipeline to Performance: How APAC Data Centres Can Deliver AI-Ready Capacity at Scale appeared first on Data Center POST.

TL;DR

• Power as the Primary Bottleneck: Driven by AI and cloud expansion, digital infrastructure demand is at an all-time high, but grid capacity, interconnection timelines, and power availability now determine project timing and success.
• Practitioner-Led Advisory & Truth-Telling: Founded by former operators from Digital Realty, CoreSite, and Equinix, ASG pairs hands-on development experience with deep grid and regulatory expertise to deliver honest early feasibility assessments.
• Convergence of Power & Digital Infrastructure: Over the next decade, ASG foresees power and digital infrastructure merging into a single integrated asset class, where data centers are planned directly alongside energy generation and transmission assets

# # #

Data Center POST had the opportunity to connect with Jarrett Appleby, CEO & Managing Partner at ASG. Jarrett has spent his career on every side of the table: from running operations at Digital Realty and CoreSite, building go-to-market at Equinix, and leading strategy at global network operators like FLAG Telecom and Reliance Globalcom. When he founded Appleby Strategy Group, Jarrett wanted to give developers and investors the practitioner’s view that he would have wanted as an operator. That became ASG, a team of senior executives who have been in the trenches of data center, power and fiber development.

What is ASG’s core mission?

ASG helps clients turn digital infrastructure projects from concept into assets that are financeable and ready to transact. Power, land, permitting and capital now decide which projects succeed, and many projects stall between strategy and delivery. We bring strategy and execution under one roof. We also tell clients the truth about which projects are feasible and worth doing.

How would you describe the current state of the global digital infrastructure market in terms of challenges and opportunities for your business?

Demand has never been stronger, driven by AI and cloud. The main limit is now power, not demand. Interconnection timelines, grid capacity, permitting and supply chain decide who can deliver and when. That’s a challenge for the industry. It’s also our biggest opportunity: clients who secure power and site readiness early set themselves apart.

Who do you see as your primary customer base today, and how has that evolved over the past few years?

We work with data centers, power companies, developers, neoclouds, operators, PE/VC investors and hyperscalers. A few years ago, most engagements centered on traditional colocation and enterprise. Today, more of our work involves large AI campuses, new neocloud platforms, and investors entering digital infrastructure who need operating expertise.

What are the top priorities or concerns you hear from your customers, and how is your company addressing them?

• Speed to power: We provide power and interconnection advisory, including grid planning, queue strategy and cost allocation.
• Site certainty: We support site selection, land control, permitting and transaction readiness.
• Finding the right buyer or partner: We help identify buyers, set valuations and close deals.

Are there emerging customer segments or industries that you see as growth areas for your business?

AI and neocloud operators, energy and utility companies monetizing land and generation, and institutional capital entering the sector.

How are you aligning your company’s strategy to address sustainability and regulatory requirements in the coming years?

We build grid, energy-market and regulatory strategy into site development from day one. That includes interconnection cost allocation, on-site and hybrid generation, and working with utilities and regulators. Projects that are good grid citizens get approved and financed faster.

How is ASG fostering partnerships to create value for customers and enhance your competitive position?

We connect utilities, landowners, hyperscalers, operators and capital providers. Our team’s relationships across the ecosystem let us put the right counterparties together early, which lowers risk and shortens time to close.

What are ASG’s core products or services?

Strategic development and execution; site development strategy and leadership; go-to-market and partner facilitation; power and interconnection advisory.

What markets do you offer ASG’s solutions to?

We offer solutions to the global market. ASG is headquartered in Northern Virginia and advises on data center, power and fiber projects across North America and international markets.

What long-term goals or initiatives are you most excited about, and how do you envision the industry evolving over the next decade?

Over the next ten years, power and digital infrastructure will merge into a single asset class. Data centers will be planned alongside generation and transmission. We want ASG to be the go-to partner for that integrated development.

What upcoming industry events will you be attending?  

Over the next six months, ASG will be attending DCD Connect VA, Neocloud Summit (an invite- only ASG Event), PTC Hawaii, and CERAWeek by S&P Global.

Is there anything else you would like our readers to know about your company and capabilities? 

ASG was built by practitioners, for practitioners. Our leaders have run operations, raised capital, developed sites and negotiated with utilities, so we know what it takes to get projects built. That depth runs across disciplines: our team pairs data center operating experience with grid and regulatory expertise, including Allison Clements, a former FERC Commissioner. That combination matters more than ever now that power, not demand, sets the pace of development.

We also hold ourselves to a simple standard: we only take on work we believe is feasible and worth doing, and we tell clients the truth about it early, before time and capital are spent. Whether a client is a developer securing power and land, an investor underwriting a platform, or a hyperscaler or neocloud looking for capacity, we bring strategy and execution together to get projects to a financeable, transaction-ready outcome.

Strategic Partners. Industry Builders. Truth Tellers.

# # #

About ASG

ASG is the outcome-driven advisory partner behind leading data center, power and fiber projects. Founded by practitioners who have been developers, investors, operators and advisors, ASG provides strategic development, site development, go-to-market and power and interconnection advisory to developers, neoclouds, operators, investors and hyperscalers. ASG is headquartered in Northern Virginia. Learn more at asg-team.com or reach out to asg@asg-team.com.

The post Executive Profile: Jarrett Appleby on Power, AI, and Digital Infrastructure appeared first on Data Center POST.

TL;DR

• Rightfiber has completed its acquisition of Fastwyre Broadband’s Nebraska business, adding 26 communities to its footprint.
• Former Fastwyre customers and employees now join a provider serving 20 states, with more than 28,000 route miles of fiber and over 300,000 FTTH passings.
• Integration has begun, and customers will receive ongoing updates on products, services, billing, and future enhancements.

# # #

For more than a century, Nebraska has been an important part of Rightfiber’s story. Now, the fiber provider is expanding that presence as it enters its next chapter of growth.

Rightfiber, formerly Great Plains Communications and a portfolio company of Grain Management, has completed its acquisition of Fastwyre Broadband’s Nebraska business, bringing 26 additional communities across the state into its growing network.

With the deal done, former Fastwyre customers in Nebraska are now part of one of the nation’s largest independent fiber broadband providers. Rightfiber’s network spans more than 28,000 route miles of fiber across 20 states and supports over 300,000 Fiber-to-the-Home (FTTH) passings.

For customers, the focus now shifts to the transition. Rightfiber has begun integrating Fastwyre’s Nebraska operations and will continue communicating with customers about products, services, billing and future enhancements.

“We’re excited to welcome Fastwyre’s Nebraska customers and employees to Rightfiber,” said Heath Simpson, Chief Executive Officer of Rightfiber. “We look forward to serving these communities with reliable connectivity, responsive local support, and the customer-first experience that define Rightfiber. Our focus is on making this transition as seamless as possible.”

The acquisition also builds on a history in Nebraska that stretches back more than 115 years. While the Rightfiber name is new, the company’s connection to the state is not.

“Nebraska has been an important part of our story for more than 115 years, and completing this acquisition reinforces our long-term commitment to the communities we serve across the state,” said Todd Foje, Executive Chairman of Rightfiber. “We’re excited to build on that legacy by strengthening our relationships across Nebraska and continuing to support the people, businesses, and communities that have been the foundation of our success for generations.”

That combination of history and growth helps put the acquisition into perspective. Rightfiber is expanding the scale of its network while building on an established presence in Nebraska. As Rightfiber continues its next chapter, the Fastwyre acquisition represents another step in that evolution: expanding the network, welcoming new communities. and carrying its Nebraska legacy forward under the Rightfiber name.

Read the full release here: www.imillerpr.com/news/rightfiber-completes-acquisition-of-fastwyre-broadbands-nebraska-business.

The post Rightfiber Builds on Its Nebraska Roots With Continued Fiber Expansion appeared first on Data Center POST.

TL;DR

• Fuel readiness depends on more than having enough fuel on-site. Operators need visibility into how fuel is performing and how quickly it is being consumed.
• Real-time fuel data can help teams compare expected runtime with actual conditions and make more informed replenishment and emergency planning decisions.
• Operational intelligence can help identify developing fuel-quality or system-performance issues before they affect generator readiness.
• By connecting fuel condition, consumption, inventory, and system performance across the lifecycle, organizations can reduce uncertainty and make more confident operational decisions.

# # #

For decades, fuel has often been treated as a relatively straightforward part of backup power planning. As long as there is enough fuel on-site and the system is maintained, backup power is expected to be ready when needed.

For mission-critical operations, however, fuel readiness requires a clearer understanding of what is actually happening across the system. Operators need visibility into fuel performance and consumption so they can better understand how long available supply will support operations and whether changing conditions could affect readiness.

This is where operational intelligence becomes valuable. Instead of viewing fuel as a static asset sitting in a tank, organizations can use data from across the fuel lifecycle to better understand system performance and make more informed decisions before an outage puts those assumptions to the test. ESI Total Fuel Management’s approach reflects this shift by treating fuel as an actively managed operational variable.

From Fuel Data to Operational Intelligence

Collecting data is only useful when that information can help guide decisions. A fuel system may provide visibility into inventory, consumption, and equipment performance, but those data points become more valuable when they are viewed together and used to understand overall readiness.

Knowing how much fuel is stored on-site, for example, does not necessarily tell operators how long that supply will last. Actual consumption can change under load, and runtime assumptions may not always reflect real operating conditions. ESI’s fuel consumption monitoring capabilities help provide that added visibility by tracking engine supply and return flow rates in real time.

Moving Beyond Runtime Assumptions

Backup power planning often starts with expected runtime, typically based on tank capacity and estimated generator consumption. While those calculations are useful, actual operating conditions can change how long available fuel will support the facility.

Consumption may shift as generator load changes, and fuel may be distributed across different parts of the system. During an extended outage, those variables can make planned runtime different from actual runtime.

Operational intelligence helps give operators a clearer picture of what is happening in real time. By comparing current fuel consumption with available inventory, teams can make more informed decisions about replenishment and emergency fuel needs before supply becomes a concern.

That visibility is especially important during prolonged outages or other high-stress events, when understanding actual runtime can directly support continuity planning. ESI’s approach is designed to help mission-critical operators move from estimated runtime to a more informed view of how long their systems can actually perform.

Seeing Risk Before It Becomes a Failure

Fuel intelligence goes beyond knowing how much fuel is available. Fuel condition and overall system performance also affect whether backup power will perform as expected when it is needed.

In standby applications, fuel may remain stored for long periods, creating opportunities for contamination or other issues to develop over time. Scheduled inspections and testing can help identify problems, but an intelligence-driven approach provides a more continuous view of system conditions so emerging risks can be addressed earlier.

ESI’s model is built around that broader view of readiness. Instead of evaluating fuel quality or equipment performance in isolation, ESI treats the fuel system as an interconnected whole and emphasizes continuous verification. This gives operators greater visibility into developing concerns before they have the chance to affect generator performance.

Turning Visibility Into Better Decisions

Operational intelligence is most valuable when it helps teams take action with greater confidence. Visibility into fuel condition, consumption, inventory, and system performance gives operators a stronger foundation for decisions across the fuel lifecycle.

Operational intelligence can also improve lifecycle efficiency by giving teams better information about how fuel is being used and where attention may be needed. This can support more informed maintenance decisions, help extend fuel life, and reduce unnecessary waste without compromising readiness.

During an extended outage, current operating information becomes even more important. Facility teams and fuel partners can use that information to coordinate replenishment and response based on actual conditions rather than relying only on predetermined plans.

Across each of these areas, the goal is the same: Reduce uncertainty and give operators clearer, more timely information to support better decisions.

Connecting Intelligence Across the Fuel Lifecycle

A single data point cannot provide a complete picture of fuel readiness because each part of the fuel lifecycle is connected. From sourcing and delivery through storage, conditioning, transfer, consumption, and replenishment, conditions in one area can influence performance in another.

That is why operational intelligence becomes more valuable when it is part of a broader fuel management strategy rather than treated as a standalone monitoring function. ESI brings together engineered systems, lifecycle services, operational intelligence, fuel access, and emergency readiness to manage fuel as one integrated system.

This creates an important distinction between simply seeing what is happening and understanding what to do next. Monitoring can provide visibility into current conditions, while operational intelligence helps operators interpret that information in context and use it to support decisions across the fuel lifecycle.

From Fuel Visibility to Operational Confidence

Mission-critical operators do not need more data simply for the sake of having it. They need information that gives them a clearer understanding of current conditions, helps identify what may be changing, and supports timely action when it is needed.

That is the value of actionable fuel intelligence. When fuel condition, consumption, inventory, and system performance are viewed together, teams can make decisions based on a more complete picture of readiness rather than relying on assumptions.

For ESI, this means managing fuel as more than something stored on-site for an emergency. Fuel is treated as a performance-critical asset that can be monitored and actively managed throughout its lifecycle, giving operators greater visibility and confidence in how the system is performing.

When backup power is essential to uptime, better information can support better preparation and more informed decisions before an issue has the chance to affect operations.

To learn how ESI Total Fuel Management can help turn fuel-system data into actionable operational intelligence, contact ESI Total Fuel Management today.

The post Transforming Fuel into Actionable Intelligence for Smarter Decisions appeared first on Data Center POST.

TL;DR

• Space Is Not the Constraint, Usable Space Is: A 5 MW requirement for rack-scale GPU systems went to more than ten operators. Nearly all had space; almost none could support the density on the timeline
• Liquid-Cooling Ready Is Not Liquid-Cooled: The date that matters is when facility cooling reaches the white space, and it often sits well behind the building
• Credit Can Matter as Much as Capacity: Two operators raised, unprompted, whether the tenant was investment grade
• Buyers Have More Flexibility Than They Use: Phasing, multiple sites, air-cooled platforms and an earlier start all widen the field

# # #

By James Mercer, Principal, Metro Colo Advisory

North American data center vacancy has never been lower. CBRE put it at a record 1.4 percent across primary markets in the first half of 2026. It is easy to read that as “there is no space.” For AI buyers, the more accurate reading is harder: there is space, but very little of it can run what they are trying to deploy.

This year we took a requirement to market that tested that distinction directly: 5 MW of capacity for rack-scale GPU systems, liquid-cooled, commissioning in early 2027, with a path to roughly twice that within a year. We took it to more than ten operators across North America. Nearly every operator had space. Almost none could support the requirement. Here is why, and what buyers can do about it.

The Gap Between Space and Usable Space

Current rack-scale systems such as NVIDIA’s GB300 NVL72 draw 140 to 160 kW per rack. Most of the market was not built for that. According to the Uptime Institute, rack density growth is concentrated between 10 and 30 kW, about one facility in eight reports any racks at 30 to 59 kW, and racks above 100 kW remain rare.

Three things decide whether a hall can take this hardware, and none of them can be changed quickly:

Power density per rack. A hall designed for 15 kW racks cannot be averaged up to 150. The electrical distribution to each position has to exist.

Where the cooling loop ends. Direct liquid cooling needs facility water delivered to the rack row. A building with chilled water at the plant room but not in the white space is not a liquid-cooled building.

Floor loading. A fully populated rack-scale system weighs about 3,000 pounds. Many raised floors were never rated for it.

An operator can have megawatts of vacant, powered space and still have nothing that clears all three.

“Liquid-Cooling Ready” Is Not Liquid-Cooled

The most common answer we heard was a variation of “our facilities are liquid-cooling ready.” In practice that meant the building could support liquid cooling once equipment was procured and installed, and one operator put the lead time for facility-side cooling equipment at nine to twelve months. In one case, a building was due in early 2027 while its liquid-cooled space was not due until months later.

It also helps to separate two kinds of equipment that often share a name. The coolant distribution units inside the racks usually ship with rack-scale systems. The facility-side units that connect the building’s water to the rows belong to the operator. When an operator cites a long lead time, it is worth asking which of the two they mean.

The Gate Nobody Mentions: Credit

Two of the operators that engaged most seriously raised the same question, unprompted: was the tenant investment grade? One made a meeting about its future capacity conditional on the answer.

That makes sense from the operator’s side. A multi-year, multi-megawatt commitment is a large receivable, and in a market this tight an operator can choose its tenants. But it means some declines framed as “no capacity” may be credit decisions. AI buyers without investment-grade credit should expect the question, and arrive with an answer: a parent guarantee, a deposit, or a stronger entity on the lease.

What Buyers Should Ask Before Shortlisting a Facility

1. What density can you support per rack, sustained, in the hall you are offering? Not the campus maximum.
2. On what date will facility cooling be live at the rack row? Not the building handover date.
3. What is the floor rating in that hall?
4. What is the expansion path, and is it approved or on a roadmap? Many operators can start a deployment. Far fewer can commit to doubling it.
5. What will you need to see on credit? Asking early saves weeks.

Where the Flexibility Is

Buyers have more room than the headline requirement suggests. Starting smaller with a committed expansion path opens more facilities than insisting on the full load on day one. Splitting a deployment across two sites works when the workloads are independent. And where the workload allows, air-cooled GPU servers at 30 to 50 kW per rack fit a far larger share of the market than rack-scale systems do.

The biggest lever is time. Capacity for 2027 is being committed now: CBRE found 80.4 percent of capacity under construction already preleased. A buyer who starts twelve to eighteen months ahead, with credit support lined up and flexibility on phasing, is choosing from a market. A buyer who starts six months ahead is choosing from whatever is left.

# # #

About the Author

James Mercer is Principal of Metro Colo Advisory, an independent data center advisory practice based in New York. He places colocation, wholesale and AI infrastructure requirements with operators across North America, and advises developers and investors on how facilities will compete for tenants.

The post Vacant Is Not Usable: What a 5 MW AI Capacity Search Found appeared first on Data Center POST.

TL;DR

• Following its 2026 relaunch, IGF-USA is exploring how it can evolve beyond an annual conference into a sustainable, year-round organization.
• Building institutional memory and separating administrative operations from substantive programming could help IGF-USA maintain continuity while bringing new participants into the organization.
• A decentralized national model could expand Internet governance discussions beyond Washington and involve universities, state and local governments, Tribal nations, businesses and communities.
• Youth and academic engagement could bring the next generation directly into Internet governance through student programming, curriculum integration and active participation.
• Digital infrastructure issues including land use, energy, community engagement and infrastructure development are increasingly part of Internet governance and demonstrate the need for broader participation.

# # #

By Ilissa Miller, CEO of iMiller Public Relations

One of the most interesting parts of IGF-USA 2026 happened after our data center discussion. The final session turned the governance conversation inward.

The question became, after relaunching IGF-USA following a multi-year hiatus, what should the organization itself become?

It’s a timely question. The global Internet Governance Forum now has a permanent mandate from the United Nations. Meanwhile, AI, cybersecurity, digital infrastructure, privacy and other technology issues are evolving rapidly, and increasingly playing out far beyond Washington.

Learning From the Past

Christian Dawson brought important institutional history to the conversation. Nearly a decade ago, he co-chaired an IGF-USA Sustainability Working Group focused on creating processes and governance structures that could make the organization sustainable.

One of the lessons from that experience was the importance of institutional memory.

Organizations built heavily around volunteers can become dependent upon a relatively small number of people who know how everything works. When those people move on, knowledge can disappear with them.

Christian captured the challenge beautifully: IGF-USA needs “continuity without dependency and renewal without exhaustion.”

That’s not only a useful principle for IGF-USA. It’s a pretty good governance principle for almost any multistakeholder initiative.

Separate Operations From Ideas

Ron da Silva, Chair of the IGF-USA 2026 Steering Committee, discussed another practical idea: separating the administrative responsibilities of running IGF-USA from the substantive programming.

A steering committee can handle the mechanics necessary to keep the organization functioning while a broader programming community focuses on the issues, participants and conversations that matter.

Leadership can rotate. Institutional knowledge can be documented. New people can participate without requiring the organization to reinvent itself every year.

And, I think the bigger idea emerged as the discussion continued.

Don’t Just Move the Conference. Move the Conversation.

If IGF-USA is truly a national Internet Governance Forum, why should its center of gravity remain primarily in Washington? Internet governance is happening everywhere.

It’s happening in state legislatures. Universities. Tribal nations. Planning boards. Utilities. Technology companies. Economic development organizations. And, as our earlier panel demonstrated, communities deciding whether and how digital infrastructure should be developed.

The idea discussed was not simply rotating an annual conference around the country. It was creating a decentralized national network.

Changing geography should change who participates, which institutions become involved and, importantly, which questions get asked.

That could mean universities helping shape programming. Local businesses bringing regional technology issues forward. State and local officials participating directly. Civil society organizations introducing concerns that might never surface in Washington.

Bring the Next Generation With Us

Chelsea Horn of Lehigh University added another critical component: youth and academic engagement.

Her ideas included academic affiliates, student scholars, youth-led programming, curriculum integration and involving students directly as moderators, rapporteurs and participants.

If we’re discussing how the Internet and AI should evolve over the next several decades, the people who will inherit those systems shouldn’t simply be invited to watch the conversation.

The Data Center Panel Proved the Point

What I found particularly interesting was that the data center session I was part of earlier in the day became something of a proof point for this proposed model.

During the final discussion, organizers specifically noted that the panel brought people and perspectives into the IGF conversation that aren’t normally part of the traditional Washington Internet-policy world: Local planning; Tribal sovereignty; Community engagement; Land use. Energy; and  Infrastructure development.

Those aren’t separate from Internet governance anymore. They are Internet governance.

And that’s why I left Philadelphia convinced that IGF-USA shouldn’t simply be an annual event.

It has an opportunity to become a decentralized, year-round ecosystem where industry, government, academia, technical experts and communities continually work through the difficult questions accompanying our digital future.

The Internet itself is decentralized.

Perhaps the conversation about how we govern it should be too.

And for those of us working across digital infrastructure, there is every reason to participate. Because the next generation of Internet governance isn’t only about what happens online. Increasingly, it’s about what happens in our communities.

Curious about what you missed? Not to worry, you can still view the live stream online here: isoc.live/21744.

Stay tuned for IGF-USA upcoming events: igfusa.us.

The post Beyond the Annual Conference: What’s Next for IGF-USA? appeared first on Data Center POST.

TL;DR

• Data Fragmentation Challenge: The data center industry suffers from widespread data fragmentation, with critical information scattered across energy markets, land records, regulatory systems, supply chains, capital markets, and sustainability frameworks.
• Sovereign Intelligence Layer: Rather than merely storing isolated datasets, organizations need a sovereign intelligence layer to connect analytical silos into a governed, decision-ready environment.
• Critical Role of Data Provenance: As AI accelerates analytical speed, maintaining verifiable source evidence and traceable data provenance is essential for high-stakes, multimillion-dollar capital decisions.
• Data as Decision Infrastructure: Market advantage will belong to organizations that move beyond data accumulation to build connected, scalable decision infrastructure.

# # #

The strategic case for connecting fragmented infrastructure data into governed, decision-ready intelligence. 

The data center industry does not suffer from a shortage of data. It suffers from fragmentation. 

Operators, developers, investors, infrastructure providers, and technology companies increasingly make decisions using information scattered across energy markets, land records, regulatory systems, supply chains, capital markets, sustainability frameworks, and geospatial databases. 

Each dataset may be useful independently. The larger challenge is turning them into a coherent picture quickly enough to support consequential infrastructure decisions. 

As artificial intelligence accelerates demand for digital infrastructure, this distinction becomes increasingly important. The next generation of data infrastructure should not merely store information. It should help organizations understand the relationships between information, establish its provenance, and convert it into decision-ready intelligence. 

THE FRAGMENTATION PROBLEM 

Consider a hypothetical data center development. 

Before construction begins, decision-makers may need to understand land ownership, zoning, power availability, grid constraints, water resources, fiber connectivity, environmental requirements, financing conditions, construction costs, supply-chain exposure, and local regulations. 

Those questions frequently reside in different systems maintained by different institutions. 

Traditional analytics can answer individual questions. Decision infrastructure must address the relationships among them. 

A parcel of land may appear attractive until energy constraints are incorporated. An energy source may appear economical until regulatory requirements are considered. A development opportunity may appear financially sound until infrastructure costs or supply-chain dependencies are incorporated.

The value therefore lies not simply in possessing more information, but in connecting information that previously existed in separate analytical silos. 

AI IS ONLY PART OF THE ARCHITECTURE 

Artificial intelligence offers extraordinary capabilities for analyzing large amounts of information, but AI does not eliminate the need for reliable source data. 

An intelligence system is only as useful as its underlying evidence. 

Organizations deploying AI-assisted decision systems should therefore consider several questions: Where did the information originate? When was it collected? Which jurisdiction does it concern? Can the underlying source be verified? Has the information changed? What assumptions were introduced during analysis? 

These questions become particularly important when analytical outputs influence capital allocation, infrastructure development, or regulatory decisions. 

The objective should be an architecture in which AI accelerates analysis while preserving a clear path back to the underlying evidence. 

TOWARD SOVEREIGN DECISION INFRASTRUCTURE 

A useful concept emerging from this environment is the sovereign intelligence layer. 

Here, sovereignty does not necessarily mean isolation. It means maintaining control over how information is sourced, governed, interpreted, and used. 

For data-center organizations, such an intelligence layer could combine several domains: infrastructure and energy intelligence; property and geospatial information; regulatory and jurisdictional data; supply-chain information; corporate and counterparty intelligence; financial and capital-market information; and sustainability and environmental indicators. 

The resulting system becomes more than a dashboard. 

It becomes an analytical environment capable of showing how changes in one domain affect decisions elsewhere. 

PROVENANCE BECOMES INFRASTRUCTURE 

As AI-generated analysis becomes commonplace, provenance may become as important as processing power. 

Decision-makers need to distinguish among verified facts, estimates, assumptions, historical information, and machine-generated interpretations. 

That requires systems designed around evidence rather than merely outputs. 

Every significant analytical conclusion should ideally have a traceable path back to its underlying information.

This principle becomes especially important in infrastructure environments where a seemingly small error can influence multimillion-dollar capital decisions. 

WHAT COMES NEXT 

The rapid expansion of AI infrastructure will create enormous quantities of operational and market information. The organizations that gain the greatest advantage may not simply be those possessing the largest datasets. 

They may be the organizations capable of connecting disparate datasets while maintaining provenance, governance, and context. 

That requires moving beyond the traditional idea of data as something organizations merely collect. 

Data must increasingly become decision infrastructure. 

The future data center will therefore be shaped not only by processors, power systems, cooling technology, and connectivity. It will also depend upon the intelligence architecture surrounding those physical systems. 

As digital infrastructure becomes more interconnected with energy, capital, property, regulation, and public policy, the ability to see those relationships clearly could become one of the industry’s most valuable capabilities. 

# # #

About the Author

Safwan Bey is Chairman and CEO of Trinlah Sovereign Holdings Corporation (TSHC), where his work encompasses data intelligence, infrastructure, private capital, sustainability, and technology. He leads the development of the TSHC Sovereign Data Foundry, an initiative focused on transforming fragmented public, institutional, and commercial information into structured, decision-grade intelligence. 

The post From Data Overload To Decision Infrastructure: Why Data Centers Need A Sovereign Intelligence Layer appeared first on Data Center POST.

TL;DR

• Executive Appointment: Datalec Precision Installations (DPI) appointed Dominik Dziarczykowski as Business Development Manager to bolster its high-density and liquid cooling capabilities.
• Proven Track Record: Dziarczykowski brings more than 20 years of industry experience, including previous liquid cooling roles at CoolIT Systems and Vertiv. 
• AI & High-Density Focus: Joining Richard Dobbie’s Business Development team, he will focus on providing practical cooling solutions for AI-driven computing needs.

# # #

Written by iMiller Public Relations on Behalf of Datalec Precision Installations (DPI).

Datalec Precision Installations (DPI) has appointed Dominik Dziarczykowski as Business Development Manager to strengthen its capabilities in liquid cooling and high-density data center infrastructure. Joining the Business Development team under Sales and Marketing Director Richard Dobbie, Dziarczykowski will support customers and partners as DPI continues to expand its direct liquid cooling solutions and associated services. 

Dziarczykowski brings over 20 years of experience across the IT, consulting, and data center industries. His professional background includes roles at major global organizations such as IBM, Dell, Huawei, Deloitte, and Capgemini, along with direct liquid cooling leadership at CoolIT Systems and Vertiv. In his new position, he will assist clients with practical liquid cooling deployments, debuting as a company representative with a fireside chat at Data Centre World Asia 2026 in Singapore. 

Datalec’s leadership highlighted that his technical and market expertise aligns directly with the company’s mission to deliver end-to-end, future-ready infrastructure capable of supporting high-density workloads and the AI revolution. 

To read the full press release, please click here.

The post Datalec Strengthens its Liquid Cooling Capabilities with the Appointment of Dominik Dziarczykowski appeared first on Data Center POST.

TL;DR

• Shift Focus from Deployment to Long-Term Operations: While the urgency to bring AI capacity online drives rapid installation, long-term success depends on treating liquid cooling as an ongoing operational discipline rather than just a quick procurement project.
• Holistic Resilience & Blast Radius Control: Evaluating system resilience requires looking beyond redundant Cooling Distribution Units (CDUs) to minimize the overall failure “blast radius” across the entire fluid distribution and control network.
• Proactive Performance Management & Adaptability: Leveraging real-time operational data allows operators to catch performance drift before component failure occurs, ensuring cooling systems continuously adapt as AI hardware and dynamic workloads evolve.

# # #

by Kevin Roof, Director of Offer & Capture Management, LiquidStack

Much of today’s discussion around liquid cooling centers on deployment. Which CDU should I buy? Which cooling technology should I choose? What rack density can I support? These are important questions, but they represent only the beginning of the liquid cooling journey.

There is an understandable urgency to bring AI capacity online as quickly as possible. But in some cases, the pace of deployment is beginning to outstrip the operational maturity surrounding liquid cooling. Many organizations are implementing these systems for the first time, often without the decades of established practices that exist around more traditional data center infrastructure.

As AI infrastructure grows more critical the question therefore needs to shift from “How do we deploy liquid cooling?” to “How do we keep these systems operating efficiently for the next decade?”

Installation is measured in weeks. Operations are measured in years.

Liquid Cooling Operations Are an Engineering Discipline

Many operators still approach liquid cooling primarily as a procurement and deployment project: find the right CDU, integrate it into the environment, bring the GPU clusters online and move on to the next capacity challenge. But liquid cooling systems are infrastructure platforms requiring careful and continuous management.

Design establishes the conditions for long-term performance. Commissioning validates that those decisions work in practice. Operations determine whether that performance can be sustained as workloads, hardware and operating conditions change.

That last point is increasingly important. Traditional data center operations were largely built around maintaining equipment. With liquid cooling, the technology cooling system, facility infrastructure, controls and IT equipment operate as a highly interconnected system. Changes to rack configurations, accelerators, control strategies or even routine maintenance can affect behavior elsewhere in the cooling loop.

That is why liquid cooling operations increasingly need to be viewed as an engineering discipline rather than simply a maintenance function.

Fluid Management Means More Than Filling the Loop

One of the biggest misconceptions surrounding liquid cooling is that coolant is simply another utility, much like chilled water flowing through a facility.

In reality, fluid management is one of the most important determinants of long-term system reliability. The fluids circulating through a liquid cooling system must remain compatible with every material they encounter, from piping and seals to cold plates and heat exchangers.

Maintaining proper chemical balance helps minimize corrosion, prevent scaling, reduce biological growth and preserve consistent thermal performance. Small deviations may not create immediate problems, but over time they can gradually sap efficiency and shorten the lifespan of critical components. Internal channels on advanced coldplates can be as small as 100µm, which is about the diameter of a human hair. Even minor biological growth or debris from scaling can easily clog coldplate channels or the filters in place to protect them. 

The important mindset shift, then, is to stop treating fluid quality as something checked only during scheduled service. It is an ongoing operational condition that can affect virtually every component in the system.

Resilience Is About More Than Redundant CDUs

High availability has always been a cornerstone of data center design, but redundancy becomes even more important as AI workloads grow in value and computational intensity.

Most operators understand concepts such as N, N+1 or N+2 configurations. Similar principles apply to liquid cooling, but redundancy should extend well beyond the CDU itself.

One mistake is evaluating resilience too narrowly at the component level. A redundant CDU does not necessarily create a resilient cooling architecture if another shared element can interrupt cooling across an entire compute pod.

A better way to think about resilience is in terms of blast radius: if something does fail, how much of the environment does that failure affect?

A resilient system considers pumps, heat exchangers, sensors, valves, controls, the broader fluid distribution network, as well as the upstream power architecture and TCS loop layout. The goal is not simply to eliminate individual points of failure. It is to contain failures and ensure that maintenance or component replacement can occur without unnecessarily disrupting compute operations.

Maintenance Is Becoming Performance Management

Liquid cooling also creates an opportunity that operators should take advantage of: these systems generate a wealth of operational data.

Flow rates, pressure differentials, supply and return temperatures, pump performance, fluid quality and leak detection collectively provide a detailed view into overall system health. The question is whether organizations use that information simply to react to alarms or to understand how system performance is changing over time.

Preventive maintenance has traditionally centered on fixed intervals. Fluid is sampled, filters are inspected, pumps are verified, sensors are calibrated and valves are tested according to a schedule. Those activities remain essential, but they no longer tell the whole story.

In increasingly dynamic AI environments, understanding how the system behaves between maintenance intervals can be just as important.

Small deviations in pressure, pump current, temperature or flow stability may not indicate an immediate problem. But they can provide early evidence that conditions are drifting. By the time a component fails, the system may have been signaling deterioration for weeks.

That turns maintenance into performance management. Instead of waiting for failures, operators can identify trends, investigate abnormalities and schedule service before small changes become production issues.

There is an organizational dimension here as well. Liquid cooling increasingly blurs the boundaries between facilities, IT teams, tenants and equipment vendors. If different stakeholders can make changes that influence flow, pressure or thermal behavior, organizations need clear ownership around who monitors the system, who can change parameters and how visibility is shared.

Operational resilience increasingly depends on coordination as much as hardware.

The System You Install Is Not the System You Will Operate

Perhaps the biggest mistake organizations can make is designing and operating exclusively for today’s realities.

AI infrastructure is evolving at an extraordinary pace. Cooling systems deployed today may be expected to support several generations of accelerator technology, each bringing different thermal characteristics, flow requirements and operating behaviors. And it is not simply a question of supporting more heat. 

Training environments can introduce rapid swings in compute demand, requiring cooling systems to respond dynamically. Inference environments may create steadier, sustained loads. Hardware refreshes can change hydraulic requirements even when the underlying cooling infrastructure remains physically unchanged.

That means successful commissioning does not establish a permanent operating state.

When accelerators change, racks are reconfigured or significant maintenance alters the cooling loop, operators should reassess whether the assumptions under which the system was originally commissioned still apply. In some cases, changes may warrant new control strategies, different setpoints or even renewed system validation.

A well-operated liquid cooling system should not be defined by its ability to remain unchanged. It should be defined by its ability to adapt while continuing to deliver predictable thermal performance.

Installation Is Only the Beginning

Liquid cooling has become one of the foundational technologies enabling the AI era, but technology alone does not guarantee success.

The true measure of a deployment is not how quickly the equipment is installed or how well the system performs on commissioning day. It is whether that infrastructure continues to deliver reliable, efficient cooling years later as workloads expand, hardware changes and operating conditions evolve.

That requires a different mindset. Installation is not the finish line, and commissioning does not freeze a cooling system in time. Liquid cooling must be continuously observed, maintained, adjusted and, when conditions materially change, revalidated.

The industry has spent enormous energy learning how to deploy liquid cooling for AI. Now we need to become equally good at operating it.

# # #

About the Author

Kevin Roof is the Global Director of Offer & Capture Management for LiquidStack. A mechanical engineer and PMP with over a decade of experience in data center cooling, Kevin brings invaluable insights and thought leadership to the liquid cooling space.

The post The Real Challenges of Liquid Cooling Begin After Installation appeared first on Data Center POST.

TL;DR

• Cameras used as visual sensors can turn physical conditions into useful operational data, monitoring equipment between inspections and reducing routine manual checks.
• Connecting visual and thermal data with BMS and DCIM platforms gives engineers context to assess alerts sooner, provided hardware and integration suit the task.
• Start with a familiar monitoring task and test whether it reduces verification time and unnecessary visits before extending deployment.

# # #

By Peter Dempsey, EMEA Key Account Manager & Data Centre Lead, Axis Communications

Data center engineers spend part of their working day collecting information that equipment could, in some cases, report for itself. Reading a gauge or checking equipment with a handheld thermal camera remains valuable work, but repeating those tasks across a large facility takes time. Much of that effort is necessarily spent confirming that conditions remain normal – work that increasingly lends itself to automation.

Used as visual sensors, cameras can assist with much of this routine observation. They can provide a continuous view of physical conditions, while suitable analytics turn a gauge reading or thermal change into data that operational systems can use. In doing so, they connect what is happening at the equipment with the platforms responsible for monitoring it, including conditions that might otherwise require someone to visit and verify.

The opportunity is therefore not simply to add more monitoring, but to make physical evidence useful within the systems operators already trust. That means deciding where visual sensors can add value, what information they should provide and how that data will integrate with the wider monitoring strategy. 

Make the information useful where it arrives

For most operators, that means starting with the building management system (BMS) and data center infrastructure management (DCIM) platform. These are already central sources of operational oversight. Visual and thermal information adds value when it appears within those workflows; if it remains in a separate interface, engineers still have to find it and manually match it to the equipment concerned.

Consider a temperature alert associated with a particular cabinet. Linking it to the relevant thermal view allows the operator to remotely examine the area where the abnormal condition is concentrated. A suitable visible-light view may provide further context, helping them assess what needs investigating before they attend.

The connection should work in both directions. Status information from a rack or management platform can accompany the view of that asset, so the engineer can relate what they see to what the equipment is reporting. 

BMS and DCIM platforms retain their management roles, but benefit from additional information available to support decisions. That information can also generate, or be attached to, a work order or help a remote engineer advise colleagues on site. 

The practical test is to determine whether the addition of visual data actually reduces the work needed to understand the event. If staff still have to search several systems before they can assess it, the integration has left an important part of the job unfinished.

Monitor what changes between inspections

A scheduled inspection confirms conditions only at the time of the visit. It cannot account for a battery string that starts running hot shortly afterwards, or airflow that deteriorates following maintenance. In this case, detection depends on a secondary monitoring system or, often, on the next person arriving.

Increasing round frequency reduces that interval, at the cost of additional staff time. For teams responsible for expanding facilities, this becomes difficult to sustain. Operators are now in the position of identifying which checks would benefit from continuous observation and which still require an engineer’s physical presence.

Fixed thermal monitoring can complement periodic surveys by watching selected equipment between visits. Where the device and application support it, a temperature exception can trigger an alert accompanied by a view of the affected area. The engineer can then assess the incident’s location and extent when deciding how urgently to respond.

Advancements in monitoring do not replace expertise; a heat spike still needs interpretation. It may indicate a failing component, obstructed airflow or a condition that has already corrected itself. While thermal and visual information can help quickly narrow the investigation, neither will explain every fault.

Gauge monitoring offers another practical application. If the current process requires someone to read a dial and enter the result into a report, a suitable camera and reading application can instead supply that value directly to the management platform. The facility gains an auditable record that can be compared with defined limits, without waiting for the next round.

Choose hardware that supports the task

The monitoring task should determine the hardware, not the other way around. Reading a gauge requires a camera and lens that capture enough detail at the installed distance, with a clear viewing angle and lighting that avoids glare across the dial. Direct temperature monitoring requires a thermal device designed to provide temperature measurements across the requisite range, with sufficient accuracy to meet defined thresholds.

Processing capability also needs to match the application. If analytics are to run on the device, it must support that software and have the resources to perform the required task. The task itself must be clearly defined before procurement. Monitoring a thermal area against agreed limits, for example, gives operators something they can test against actual equipment conditions.

Specify the connection as well as the device

Before selecting equipment, operators must also establish how its output will reach the receiving platform. The system needs to recognize what a reading represents and which asset it belongs to. Putting devices on the same network does not, by itself, make their information useful to one another.

Protocols such as MQTT, Modbus TCP and OPC UA can support data exchange between visual sensors and management platforms, so operators should confirm which interfaces are supported and whether integration software is needed between them. But a protocol listed on a specification sheet is only part of the answer. An open platform gives operators greater freedom to build the integrations they need without the hardware itself becoming the limiting factor. 

The integration must also be secured, monitored and maintained throughout its operational life. Clear failure alerts are essential: an automated check should never give operators misplaced confidence in a reading that is no longer accurate.

Start with a task the team knows well

A useful first deployment should address a recurring task that the operations team can describe in detail. Beginning with selected thermal checks or manually recorded gauges offers a manageable scope and existing known routines against which to compare performance. The trial should cover the equipment, but also what happens after detection, including how the engineer receives the information and responds.

The team can then assess whether the change reduces verification time or unnecessary visits, as well as whether it detects the intended conditions reliably. They should also check what happens when a view is obstructed or a reading cannot be obtained. Those situations establish where manual inspection must remain and how staff will know it is needed.

Before extending the deployment, operators should ask engineers what work the system has removed and what work it has added. Their experience, alongside the trial results, should guide the next investment. If the result is simply another screen to monitor, the integration has not yet delivered its intended value.

# # #

About the Author

Peter Dempsey has spent 25+ years in the security sector undertaking various roles that have seen him take responsibility for key elements of mission critical, high-profile projects across a number of different vertical markets. For the last two years Peter has focused his attention on how security technologies best support a business’s security strategy, whilst driving the adoption and elevating the importance of cybersecurity and emerging technologies for physical security practitioners in the Data Centre vertical. As the Industry Liaison Manager for Axis Communications, Peter has been instrumental in driving the Axis strategy for Data Centres across Northern Europe and the wider EMEA region. Peter’s passion for Data Centres can be traced back to the huge success in Ireland with Hyper Scale and Multi-Tenant Data Centre (MTDC) deployments. The key to this success was Peter’s commitment to deliver a reliable sustainable solution that will mitigate downtime.

The post Turn Visual Monitoring Into Operational Intelligence in the Data Center appeared first on Data Center POST.

TL;DR

• Performance-Driven Adoption: Growth is largely fueled by AI/ML workloads, big data analytics, and real-time transactions that require maximum compute performance without virtualization overhead.
• Security & Hybrid Integration: Strict privacy regulations in BFSI and healthcare favor single-tenant isolation, while hybrid/multi-cloud models let enterprises pair bare metal seamlessly with virtual public cloud environments.
• Global On-Demand Infrastructure: MOD Mission Critical’s MOD PaaSPort™ delivers dynamically programmable physical hardware and colocation from core to edge, offering low-latency connectivity across an ecosystem of cloud and network providers.

# # #

Written by iMiller Public Relations on Behalf of MOD Mission Critical.

Demand for bare metal infrastructure is experiencing significant expansion driven by the rigorous requirements of modern workloads, including artificial intelligence, machine learning, big data analytics, and real-time financial transactions. These data-intensive applications require consistent high performance, predictable control, and dedicated resources without the performance overhead inherent in virtualization.

On top of that, strict compliance and data privacy regulations in sectors like healthcare and banking, financial services, and insurance are accelerating the preference for isolated, single-tenant environments. Combined with the rise of hybrid and multi-cloud strategies, organizations are actively combining bare metal hardware with public cloud environments to optimize performance according to specific workload demands.

Financial projections underscore a steep growth curve across the global bare metal cloud services market. Fortune Business Insights estimates market value at over $14 billion in 2025, forecasting it to surpass $78 billion by 2034 at a compound annual growth rate (CAGR) of roughly 21%. Straits Research reflects a similar trend, projecting growth from just over $12 billion in 2025 to nearly $62 billion by 2034, with a projected 21% CAGR between 2026 and 2034. Although research methodologies differ, both analyses indicate that dedicated physical infrastructure is becoming a core strategic component for mainstream enterprise workloads rather than remaining confined to legacy or niche use cases.

MOD Mission Critical addresses this market demand through its MOD PaaSPort™ platform, delivering dynamically programmable, on-demand physical infrastructure and colocation from core to edge across global markets. MOD’s footprint spans major data centers to supply direct or ultra-low-latency connectivity to cloud, neocloud, and network providers. By integrating bare metal compute with managed colocation, networking, and cloud access, the platform enables businesses to flexibly source, deploy, and scale dedicated infrastructure suited for AI processing, regulatory compliance, and high-control environment needs.

To continue reading, please click here.

The post Why Bare Metal Infrastructure Is Gaining Ground appeared first on Data Center POST.

TL;DR

• Since 2011, the firm has supported companies across data centers, networks, subsea systems, infrastructure investment, technology, and AI.
• iMPR’s services extend beyond traditional PR to include strategic advisory, crisis communications, and community engagement.
• The firm’s Groundswell™ methodology helps organizations build transparency, alignment and trust with communities impacted by digital infrastructure development.
• Founder and CEO Ilissa Miller also leads the OIX Association Digital Infrastructure Framework Committee.

# # #

As artificial intelligence drives greater demand for data centers, power and connectivity, communications, and community engagement are becoming increasingly important to digital infrastructure development. iMiller Public Relations (iMPR) is marking its 15th anniversary as the industry enters its next phase of growth.

Founded in 2011 by Ilissa Miller, iMPR has grown alongside the digital infrastructure sector, supporting data center developers, network operators, subsea cable systems, infrastructure investors, technology providers, and AI innovators. Over the past 15 years, the firm has worked with companies representing hundreds of billions of dollars in infrastructure investment while generating thousands of media placements and thought-leadership contributions.

Beyond Traditional Public Relations

As the industry has evolved, iMPR’s capabilities have expanded beyond traditional media relations. The firm provides strategic advisory, message platform development, crisis communications, media and analyst relations, event and trade show marketing, and community engagement programs.

“At iMPR, our success has always been rooted in industry knowledge, relationships, and the ability to see where the digital infrastructure sector is heading,” said Ilissa Miller, Founder and CEO of iMiller Public Relations. “As AI, power, connectivity, and large-scale data center development reshape the global economy, communications, messaging accuracy, and community engagement are more important than ever. We are proud to celebrate 15 years supporting the companies building the infrastructure that powers the digital world.”

Community engagement has become a growing part of that work as data center developers navigate increased attention surrounding land use, power, resources, and local impact. iMPR’s Groundswell™ methodology helps organizations build transparency, alignment, and trust with communities impacted by digital infrastructure development.

Supporting Responsible Infrastructure Development

That focus extends beyond individual projects through Miller’s work with OIX Association, where she leads the Digital Infrastructure Framework Committee (DIFC). The committee is focused on advancing responsible infrastructure development by providing community planners with a framework to integrate digital infrastructure into their master plans.

Behind iMPR’s work is a team of more than 20 communications, marketing and digital specialists supporting client strategy, media relations, events, and partnerships. The leadership team includes Danielle DeMicco DiGisueppe and Jaclyn Litwa, Directors of Accounts; Brian Meeley, Director of Media & Analyst Relations; Kwan Fai Cheng, Director of Digital Operations; and Dillan Conn, Director of Strategic Initiatives and Business Platforms.

Over the past 15 years, iMPR has also built relationships with industry organizations, conferences, and associations across the digital infrastructure ecosystem. These connections help the firm support clients as they engage with industry stakeholders, policymakers, and local communities.

As AI, energy constraints, and community expectations continue to reshape digital infrastructure development, iMPR is focused on helping organizations communicate their vision, build trust with stakeholders, and navigate the evolving digital infrastructure landscape.

Learn more about iMiller Public Relations and its strategic communications, public relations and community engagement services by visiting www.imillerpr.com.

The post iMiller Public Relations Marks 15 Years Supporting Digital Infrastructure appeared first on Data Center POST.

TL;DR:

• AI workloads are pushing rack power densities from roughly 12 kW in 2024 to 26–27 kW today, with some AI racks already running 50-100+ kW.
• Unlike steady-state IT loads, AI training and inference create rapid, repeated power swings – not just backup events.
• Traditional backup batteries and capacitor-based systems each address different requirements: one prioritizes runtime, while the other prioritizes rapid response.
• Rack-level energy storage is emerging as its own design category, distinct from centralized UPS systems.
• Battery chemistry choices at the rack are now a safety and reliability variable, not just a capacity spec.

# # #

For most of the last decade, data center power design centered on one question: How long can the UPS ride through an outage? That question hasn’t gone away, but now it has company. ZincFive^®, known for nickel-zinc (NiZn) battery technology in UPS systems, announced an expansion into in-rack energy storage with its new NiZn In-Rack Power Solutions, which includes dedicated Battery Backup Unit (BBU) and AI Dynamic Power Module (DPM) configurations built for AI’s faster, more dynamic power draw. The move points to a broader shift underway across the industry: Rack-level power is becoming its own design discipline, separate from the UPS layer.

Average rack densities have more than doubled in two years, climbing from around 12 kW in 2024 to roughly 26-27 kW in 2026, according to recent industry benchmarking. Some AI training racks already exceed 50 kW, and next-generation GPU architectures are being designed around 100 kW or more per rack. That’s the headline most operators already know.

What’s less understood is how that power behaves. AI training and inference don’t draw a steady current; they spike and drop in fast, repeated bursts as GPU clusters ramp up, pause, and resume. These transient loads can stress busbars, power supplies, and upstream infrastructure in ways a traditional steady-state design never had to account for.

Backup Runtime and Rapid Response Are Different Problems

This is where the engineering gets interesting. Traditional battery backup units are built to provide runtime, with enough stored energy to ride through a utility disturbance for seconds to minutes. Capacitor-based systems, by contrast, respond in milliseconds but hold very little energy, so they smooth short spikes without offering meaningful backup duration.

Most facilities have defaulted to choosing one or blending both, accepting the tradeoff. A capacitor bank might absorb a millisecond-scale current step from a GPU cluster starting up, but it can’t cover a multi-second grid disturbance. A traditional BBU can cover that disturbance but may not be designed to manage rapid, repeated load transitions.

“AI is changing not only how much power data centers require, but how that power is consumed,” said Tod Higinbotham, CEO of ZincFive. “As power densities increase and workloads become more dynamic, the industry needs new approaches that can deliver rapid power response without compromising safety, reliability, or flexibility.”

As racks get denser and workloads more dynamic, that tradeoff becomes harder to accept. Operators need infrastructure that can do both jobs without over-provisioning space or cooling capacity at the rack, which is already at a premium in AI deployments.

Why Battery Chemistry Matters at the Rack Level

As energy storage moves physically closer to compute, the chemistry inside that storage becomes a bigger factor in facility risk management. Lithium-ion remains dominant across the industry, but proximity to high-density, high-heat compute racks raises the stakes around thermal behavior.

Nickel-zinc is one proven chemistry gaining attention specifically for this reason, already validated at scale in UPS-level deployments before moving closer to the rack. It supports high-rate charge and discharge capabilities, which suit repeated transient events, while carrying no risk of thermal runaway. For rack-level deployments, where space constraints can limit fire suppression and containment options, that safety profile is becoming as relevant as energy density or cycle life.

What This Means for AI Infrastructure Planning

The practical takeaway for facility planners and OEMs is that rack-level power is no longer a single-spec decision. It requires evaluating backup runtime, transient response speed, thermal safety, and physical footprint together, not in isolation.

ZincFive is now developing dedicated in-rack solutions – BBU configurations for established backup applications and AI DPM configurations for transient and pulse power mitigation – rather than a single configuration meant to address every requirement. That approach reflects a broader industry shift: As AI reshapes how power is consumed, the tools used to manage it are having to specialize rather than generalize.

For an industry accustomed to sizing infrastructure around steady-state assumptions, that’s a meaningful adjustment, and one that’s likely to define rack design conversations as AI infrastructure continues to evolve.

To read the full press release, visit here.

The post Rethinking Rack-Level Power for AI Data Centers appeared first on Data Center POST.

TL;DR

• Data centers support users globally, but their impacts on land, electricity, water, tax base and environmental resources are experienced at the local level.
• Communities should incorporate digital infrastructure into comprehensive planning before individual development applications arrive.
• Responsible planning should account for data centers as well as fiber routes, network nodes, substations, transmission infrastructure and other systems supporting the digital economy.
• Effective governance requires early community engagement, transparency, regional planning and consideration of a project’s full lifecycle.
• The OIX Digital Infrastructure Framework is designed to help municipalities proactively evaluate digital infrastructure needs and development considerations.

# # #

By Ilissa Miller, CEO of iMiller Public Relations

We talk about “the cloud” as though it exists somewhere above us: intangible, limitless and disconnected from the physical world. Of course, it isn’t.

The cloud is buildings, servers, fiber, substations, cell towers, transmission infrastructure, cooling systems, land, power, water, and increasingly, communities trying to understand what all of this means for them. That reality was at the center of the Data Centers and AI panel I recently participated in at IGF-USA 2026 in Philadelphia.

Moderated by Daniel Castro, President of the Information Technology and Innovation Foundation (ITIF), the discussion included Tana Fitzpatrick of the University of Oklahoma, Chris Amato of the Lehigh Valley Planning Commission, Maya Schmidt of Third Way and me, Ilissa Miller, representing iMiller Public Relations and the OIX Digital Infrastructure Framework.

Daniel framed the issue well from the outset: data centers support users around the world, but their impacts: land, electricity, water, tax base and environmental considerations are experienced locally.

So the question isn’t simply whether we need data centers. The question is how we govern their development.

Plan Before the Application Arrives

My position is that communities shouldn’t be forced to figure this out one development application at a time. Digital infrastructure needs to become part of comprehensive community planning.

That means identifying appropriate land uses for different types of data centers, but it also means looking beyond data centers to fiber routes, network nodes, substations, transmission infrastructure and the other systems communities will need to participate in an increasingly digital economy.

Not every data center belongs everywhere. And not every community needs the same infrastructure. But communities should understand their needs before a developer arrives with a proposal.

That’s one of the reasons we developed the OIX Digital Infrastructure Framework: to give municipalities a way to proactively understand these issues rather than continually reacting to them.

Start With the Community

Tana Fitzpatrick added an incredibly important dimension to the conversation: tribal sovereignty.

Federally recognized tribal nations are governments, not simply another community stakeholder. As Tana explained, engagement shouldn’t begin by asking what land, water, energy or other resources a tribal nation has that a project might need.

It should begin by understanding what that nation and its community actually want. Her question was wonderfully simple: What does being a good neighbor look like? You can read more about her work at the University of Oklahoma Native Nations Center here.

Look at the Regional Picture

Chris Amato brought the planning perspective from Pennsylvania’s Lehigh Valley, where communities are already confronting multiple proposed developments.

His perspective reinforced something I believe strongly: site and setting matter.

Brownfields, existing industrial properties and adaptive reuse may offer very different opportunities than greenfield development. Communities also need to evaluate infrastructure capacity, workforce benefits, environmental resources and net fiscal impact — not simply the large investment number announced with a project. You can read more about the Lehigh Valley Planning Commission here.

Make the Benefits and Responsibilities Tangible

Maya Schmidt of Third Way focused on another important component: ensuring communities have the information and leverage necessary to negotiate effectively.

Transparency, community engagement, infrastructure costs, tax contributions, workforce development and community benefit agreements can all be part of that conversation. You can read Maya’s Third Way data center research here.

I would add one more element: the entire lifecycle. What happens if ownership changes? Do commitments made to the community transfer to the next owner? What happens when equipment becomes obsolete? What are the decommissioning responsibilities? What happens when better technology becomes available?

Responsible development can’t end when the permit is issued.

Social License Starts With Listening

Ultimately, perhaps the simplest takeaway from our discussion was this: Developers need to listen before they propose. And, they need to start that listening process much sooner than they think.

Go into the community. Meet people. Understand their concerns. Learn what they value. Then demonstrate how that input changed the project. As I said during the discussion, development needs to become of the community, for the community and by the community.

The data center industry has an opportunity right now to redefine how large-scale infrastructure development happens in America. Not by dismissing opposition, and certainly not by assuming every project belongs everywhere. But by planning earlier, listening better, being transparent and creating guardrails that communities and industry can understand together.

That isn’t simply better community engagement. It’s better governance.

Curious about what you missed? Not to worry, you can still view the live stream online here: isoc.live/21744.

Stay tuned for IGF-USA upcoming events: igfusa.us.

The post The Cloud Has a ZIP Code: Governing the Infrastructure Behind AI appeared first on Data Center POST.

TL;DR

• Surging AI Energy Demand: AI expansion is driving a massive surge in data center electricity consumption, with U.S. data center usage projected to reach between 9.5% and 15.3% of total domestic power by 2030 and global consumption expected to roughly double to 945 TWh
• Infrastructure Timing Mismatch: Traditional grid interconnection approaches are encountering severe bottlenecks because data centers can be developed in 2 to 3 years, whereas major utility power infrastructure takes considerably longer to plan, permit, and build
• Active Role for Battery Storage: Battery Energy Storage Systems (BESS) are evolving from passive emergency backups into core drivers of facility power architecture, working alongside microgrids and on-site generation (solar, fuel cells, natural gas) to manage peak demand and ensure mission-critical reliability
• Early Civil and Regulatory Integration: Deploying utility-scale storage requires early cross-functional alignment, as project success is frequently dictated by physical site conditions, civil balance-of-plant work (such as drainage and stormwater management), fire safety, and local authority reviews

# # #

By William J. Pirrone, P.E., Founding Principal, Rubicon Professional Services

Most people are aware that AI is changing and expanding the data center industry. What gets less public attention is what this growth means for the infrastructure needed to power these data centers.

I have spent more than two decades working on critical facilities and energy infrastructure, and one of the biggest changes I have seen is the amount of power developers now need and how quickly they need it. The traditional approach of designing a facility and securing a utility interconnection is becoming increasingly difficult, putting sustainable infrastructure planning at the center of nearly every new project.

According to the U.S. Department of Energy and Lawrence Berkeley National Laboratory, data centers accounted for about 4.4% of U.S. electricity consumption in 2023. Their 2024 study projected that share could reach between 6.7% and 12% by 2028. A recent Berkeley Lab update estimates that data centers could account for 11.8% of U.S. electricity use by 2030, with scenarios ranging from 9.5% to 15.3%. The two studies indicate that these issues are also playing out globally. Case in point: The International Energy Agency projects that electricity consumption by data centers will roughly double by 2030, to about 945 terawatt-hours. AI is the biggest driver of that increase.

This research doesn’t represent an abstract energy-policy discussion; it dictates where and how data centers can be built, how quickly capacity can come online and, ultimately, whether a project is economically viable. Increasingly, developers are turning to energy storage technology deployed alongside the grid rather than as a backup to address this new industry dynamic.

Battery Storage Is Moving To The Forefront

For years, battery energy storage in critical facilities was primarily associated with backup power and UPS systems, but that’s no longer the case. It is becoming a main driver of power decisions.

Today’s battery energy storage systems (BESS) play a much more active role in the facility’s power architecture. Properly designed, BESS help manage peak demand, provide rapid response when grid conditions change, and work alongside other sources of generation.

To be properly optimized, a data center’s power needs to run through integrated energy management systems that combine grid power, battery storage, and on-site generation, including solar, fuel cells, and natural gas. Today’s objective is to create a power system that is reliable enough for mission-critical computing but flexible so that the facility isn’t entirely dependent on a single power source.

A study commissioned by California’s three major investor-owned utilities found that a modeled solar facility paired with four hours of battery storage could achieve a 99.8% effective load-carrying capability. This study demonstrated how dramatically storage can improve the reliability value of intermittent generation.

Recent improvements in battery technology advancements have made systems like this possible, but they have also made them more complex. I saw that firsthand recently on a 200-megawatt, multi-hour storage project at an existing power generation site. The system wasn’t a simple containerized battery; it was closer to a distributed industrial process plant, with its own water treatment, electrolyte handling, and air management systems layered on top of the electrical framework. Evaluating it meant pricing out civil work, mechanical systems, and electrical balance-of-plant as separate, substantial line items, rather than add-ons to a battery purchase.

However, every modern data center should not simply install solar panels and batteries as this one did. Site conditions, load profiles, utility requirements, and economics vary considerably. But this 200-megawatt, multi-hour storage demonstrates that storage has moved well beyond its traditional role as an emergency backup resource.

Leveraging Batteries: Experience Installation Matters

A utility-scale battery installation is not another piece of IT equipment that can simply be added to a data center. Large battery systems call for careful attention to fire protection, electrical safety, commissioning procedures, and emergency response. They also require close coordination with utilities, fire officials, inspectors, and other authorities, and some communities may be dealing with the review of a large-scale battery installation for the first time.

On one recent project, the single biggest cost and schedule driver wasn’t the battery technology at all; it was a pair of stormwater retention ponds already on the site. Adjusting to the new installation required partial infill, regrading, and a redesigned drainage system, all before a single battery enclosure could be installed. It’s a good reminder that the constraint that decides whether a site works often isn’t the equipment—it’s what’s already in the ground.

AI Is Forcing A Power Strategy Change

A few years ago, power infrastructure was often treated as something that followed the data center design. Today, designers need to consider power much earlier. Power availability, interconnection timing, on-site generation, and energy storage will determine a facility’s success.

The International Energy Agency has pointed out the basic timing problem: a data center can potentially be developed in two or three years, while major energy infrastructure can take considerably longer to plan, permit, and build.

The Future Should Not Contain Mismatched Power

The data centers that support the next generation of AI will require enormous amounts of reliable electricity. The industry cannot assume the grid alone will always provide that power on the timetable developers need. That is why battery storage, microgrids, and behind-the-meter generation are becoming so important. The facilities that succeed will be those where engineering, construction, utilities, local authorities, and the providers of these capabilities are brought together early enough to turn a power strategy into something that thrives in the field.

# # #

About the Author

William Pirrone is founding principal of Rubicon Professional Services and Rubicon Technical Services. He brings more than 35 years of engineering design and construction, operations, and maintenance experience to the data center, telecommunications, and critical power industries, where he established a significant presence. William’s technical expertise and understanding of an owner’s perspective have been instrumental in the rapid success of both RPS and RTS on a national level.

As a founder of the Rubicon companies, William has developed a unique set of service offerings specific to mission critical environments, designed to fill the void left by traditional construction companies and methods. These services include technical program management; construction management; equipment integration; engineering, procurement, and construction (EPC); commissioning; and energy management services.

Prior to establishing the Rubicon companies, William was the director of global engineering at JP Morgan Chase, where he led the design and integration of all data, trading, and critical operating centers through multiple billion-dollar mergers and acquisitions globally. That charge carried across 60 countries and more than 70 million square feet of real estate. While at JP Morgan Chase, William also provided direct support to the real estate division by reviewing and negotiating leases for key operations centers, developing tactical and strategic facility restack/disposition planning, and line of business relocation strategies at the corporate level. He also partnered with the energy procurement team to secure substantial cost savings and pricing stability in the deregulated energy markets across the United States. William departed JP Morgan Chase to launch Rubicon. He holds a B.S. degree in electrical engineering from Pratt Institute, is a licensed Professional Engineer, and holds multiple general contractor licenses.

The post Powering the AI Economy: What the Future Holds for Data Center Energy Infrastructure appeared first on Data Center POST.

TL;DR

• The Internet Governance Forum (IGF) provides a multistakeholder platform for government, industry, the technical community, academia and civil society to discuss public policy issues surrounding the Internet.
• The United Nations General Assembly made the IGF a permanent UN forum in December 2025, shifting the conversation toward what the forum should become next.
• Internet governance now extends beyond domain names and IP addresses to artificial intelligence, cybersecurity, privacy, digital inclusion and the physical infrastructure supporting the digital economy.
• Industry has a responsibility to bring technical knowledge and operational experience to governance discussions while listening to policymakers, communities and other stakeholders.
• The OIX Digital Infrastructure Framework provides communities and developers with a common foundation for evaluating digital infrastructure and understanding responsible development.

# # #

By Ilissa Miller, CEO of iMiller Public Relations

I recently had the opportunity to participate in IGF-USA 2026 at the Museum of the American Revolution in Philadelphia. I was invited to share the OIX Digital Infrastructure Framework, an initiative I’ve been leading to help communities better understand, plan for and govern the digital infrastructure increasingly becoming part of their physical landscape.

I expected a thoughtful discussion about Internet governance. What I didn’t fully expect was how directly the conversations would intersect with many of the challenges our industry is confronting today.

First, What is the IGF?

The Internet Governance Forum, or IGF, was established by the United Nations as a multistakeholder forum for discussing public policy issues surrounding the Internet. It brings together government, industry, the technical community, academia and civil society — not to negotiate regulation, but to develop greater understanding of complicated issues and inform those who ultimately make policy and business decisions.

The Internet wasn’t built by any one government, company or organization. It evolved through collaboration among technologists, network operators, businesses, policymakers, researchers and users. It makes sense that its governance should benefit from that same diversity of expertise.

And the timing couldn’t be more important.

In December 2025, the United Nations General Assembly made the IGF a permanent UN forum. The conversation has therefore shifted from whether IGF will continue to what it should become.

As one participant memorably put it during IGF-USA, the IGF has finally “caught the bus.” After 20 years of working toward permanence, the question becomes: Now what?

Relaunching IGF-USA

That question was particularly relevant because this year’s gathering also marked the relaunch of IGF-USA.

Led by Ron da Silva, Chair of the IGF-USA 2026 Steering Committee, the Philadelphia event brought together an impressive cross-section of stakeholders both in person and virtually.

The multistakeholder participation was one of the things that impressed me most. These weren’t simply people talking about technology. They represented organizations and communities actively dealing with how technology affects policy, infrastructure, economic development, security and society.

The event was supported by Amazon, ARIN, ICANN, Identity Digital, the Internet Society Foundation and Verisign — organizations representing very different pieces of the Internet ecosystem, and it was live streamed here: isoc.live/21744

From Technology to Governance

The Internet governance conversation has also changed dramatically.

Early IGF discussions focused heavily on domain names, IP addresses and the underlying architecture of the Internet. Today, the agenda includes artificial intelligence, cybersecurity, privacy, digital inclusion and increasingly the physical infrastructure supporting our digital economy.

AI makes this conversation even more urgent.

Technology is advancing faster than many traditional policymaking processes can reasonably absorb. Yet governments are increasingly being asked to make decisions about technologies, infrastructure and markets they may not fully understand.

That’s why the multistakeholder model matters.

Industry cannot simply show up to tell policymakers what it wants. We need to bring technical knowledge, operational experience and an understanding of real-world implementation — and we also need to be prepared to listen, adapt and be challenged. That was a recurring theme throughout the day.

We Have a Responsibility to Help Govern What We Built

That was probably my biggest takeaway from IGF-USA.

Those of us who have spent our careers building the Internet and digital infrastructure have a responsibility to help establish the guardrails for what comes next.

We understand how these systems work. We understand how infrastructure gets deployed. We understand where networks, data centers, energy systems and increasingly AI intersect.

We therefore cannot wait for regulation to solve every problem.

We need to self-govern where appropriate. We need principles. We need transparency. We need accountability. And we need mechanisms for industry, government, communities and other stakeholders to understand one another before disagreements turn into policy battles.

That is also why I believe the OIX Digital Infrastructure Framework belongs in this conversation. The Framework is intended to create a common foundation for communities and developers to evaluate digital infrastructure, giving both sides better questions to ask and a better understanding of what responsible development should look like.

The IGF does not need to become another regulatory body. Its value may be precisely the opposite: creating a place where complicated issues can be understood before decisions are made.

With AI accelerating the next evolution of the Internet, we need those conversations more than ever.

Curious about what you missed? Not to worry, you can still view the live stream online here: isoc.live/21744.

Stay tuned for IGF-USA upcoming events: igfusa.us.

The post What Is IGF and Why Internet Governance Matters More Than Ever appeared first on Data Center POST.

TL;DR

• Communities should include digital infrastructure in long-term land-use and development plans before individual projects are proposed.
• Early engagement allows developers to understand local priorities and incorporate community feedback into project designs.
• Concerns about data centers increasingly include broader anxieties surrounding artificial intelligence, resource use, and community impact.
• The OIX Digital Infrastructure Framework helps municipalities identify the questions they should consider when evaluating digital infrastructure.
• Responsible development begins with determining whether a project is appropriate for a particular community and location.

# # #

Data center development is accelerating as artificial intelligence creates greater demand for computing capacity, power, and supporting infrastructure. At the same time, many municipalities are being asked to evaluate projects that were never anticipated in their zoning codes or long-range development plans.

A recent episode of the Data Center Frontier Show explored how communities can prepare for this growth before a specific project arrives. Host Matt Vincent spoke with Ilissa Miller, Founder and CEO of iMiller Public Relations and a board member of the Open Infrastructure Exchange (OIX), about proactive planning, community engagement and the OIX Digital Infrastructure Framework.

Communities regularly use comprehensive plans to determine where residential, commercial, and industrial development should occur. Digital infrastructure, however, is often missing from those discussions. Even land already designated for industrial development may not explicitly allow data centers, creating uncertainty for residents, developers, and public officials when a project is proposed.

The OIX Digital Infrastructure Framework is intended to help municipalities consider these issues earlier. Rather than advocating for the approval of specific projects, the framework helps local leaders evaluate land use, infrastructure capacity, and community needs while identifying which questions fall under municipal authority and which require coordination with utilities or other government entities. Proactive planning also allows communities to distinguish among different forms of digital infrastructure. A municipality may determine that it can support a large AI campus, while another may be better suited for colocation, edge infrastructure, or smaller facilities serving local businesses and institutions.

Community opposition to data centers increasingly extends beyond concerns about power, water, noise, construction, and taxes. Data centers have also become highly visible symbols of artificial intelligence, causing broader questions about automation, surveillance, and AI governance to enter local land-use debates.

Those concerns should be acknowledged, but communities must also understand the specific decision before them. Questions about national or statewide AI policy are different from determining whether a particular property is appropriate for digital infrastructure. Clearly explaining the proposed facility, its potential impacts, and the stages of the approval process can help residents better understand what is and is not being decided.

For developers, meaningful community engagement should begin before a conditional-use application is submitted or a project design is finalized. This process includes studying the community, understanding the history of the proposed site and identifying the priorities of residents, businesses and local officials.

Listening sessions, open houses, business forums, and community webinars can help developers gather that information. However, engagement is most effective when community feedback can still influence the project. Developers should explain what they heard, identify what changed, and provide increasingly detailed information as plans progress.

Community benefits should also reflect local priorities rather than follow a standard formula. Depending on the community, residents may prioritize workforce development, education, transportation improvements, tax revenue, environmental protections, or preservation of local land.

Community acceptance is not solely a communications challenge. Developers must also determine whether a project belongs on the proposed property. Some locations may be unsuitable because they require significant conversion of agricultural or other valued land, particularly when existing industrial properties or brownfield sites are available.

This means responsible development may sometimes involve deciding not to move forward. Evaluating land use, infrastructure availability, and community needs before acquiring or pursuing a site can reduce conflict and help ensure that projects align with local planning goals.

To hear the full conversation with Ilissa Miller on preparing communities for AI data center development, listen to the Data Center Frontier Show podcast episode here.

The post Planning for AI Infrastructure Before Data Center Projects Arrive appeared first on Data Center POST.

TL;DR

• 365 Data Centers volunteered with Boca Helping Hands, a Palm Beach County nonprofit providing food, medical, and financial assistance, as part of a community give-back day near its Boca Raton data center.
• The event highlights 365’s continued investment in the local communities surrounding its 16 U.S. data center markets, beyond its role as an infrastructure provider.

# # #

365 Data Centers, a leading provider of high-density colocation, connectivity, and cloud services, recently spent a day giving back to the community surrounding its Boca Raton data center. Team members volunteered with Boca Helping Hands, a nonprofit organization dedicated to fighting hunger and poverty throughout Palm Beach County, Florida.

Boca Helping Hands provides food assistance, medical care, and workforce development programs to help local individuals and families build long-term stability. In 2025 alone, the organization logged nearly 58,000 volunteer hours and distributed more than 3.5 million meals to residents in need.

For 365 Data Centers, which operates a facility within the Boca Raton area, the volunteer effort underscores a broader philosophy that extends beyond powering enterprise infrastructure. The company considers serving the communities where it operates a core part of its identity as a regional and national infrastructure provider.

“Great day spent volunteering with the team at Boca Helping Hands,” said 365 CSO Charlie Dankner following the event. “This organization is a well-oiled efficiency machine that provided almost 3.5 million total meals in 2025 to families in need. That included over 288,000 weekend meals for kids while generating 54 new jobs via their workforce development programs. Proud to give back to the communities we serve.”

The Boca Raton volunteer day adds to 365 Data Centers’ growing record of local community involvement across its 16 U.S. data center markets, reinforcing the company’s presence not just as an infrastructure partner, but as an engaged neighbor. For more information on 365 Data Centers’ locations and service offerings, visit 365datacenters.com.

The post 365 Data Centers Gives Back to the Boca Raton Community with Boca Helping Hands appeared first on Data Center POST.

TL;DR

• Securing AI workloads requires layered protection that covers the infrastructure, connections, data, models and the ongoing activities that support AI operations.
• AI workloads rely on specialized computing infrastructure, large datasets and interconnected systems, with demand for AI-related data center capacity growing rapidly.
• Stronger security is important because AI workloads concentrate sensitive data, valuable computing resources and interconnected systems within data center environments.
• A layered strategy combines physical security, network and access controls, data and model protection, data orchestration safeguards and continuous monitoring.

# # #

AI workloads are changing how data centers operate and how teams secure them. As these environments expand, security teams need to protect the infrastructure, data, models and connections that support operations. A layered security strategy gives data center teams a practical way to protect these workloads throughout their life cycle.

Understanding AI Workloads in Data Centers

AI workloads refer to the computing processes that support the development and operation of artificial intelligence systems. These activities include model training, fine-tuning, inference and data processing. They often rely on GPUs and other specialized accelerators that work across interconnected servers and storage systems.

The scale of this infrastructure is expanding rapidly. Global data center capacity demand will grow by three and a half times between 2025 and 2030 as AI workloads expand. The volume alone will require an estimated 156 gigawatts of capacity by 2030, representing over four times the projected level for 2025. 

AI environments also involve extensive data flows between compute resources, storage platforms, applications and external services. Training datasets need to reach processing systems efficiently, while completed models require secure access to applications and users. These requirements create an environment where performance, availability and security must work together across the data center.

Why AI Workloads Need Stronger Data Center Security

AI is changing how businesses operate across knowledge-based industries, and financial markets are reflecting concerns about the scale of that change. An analysis reported a sharp selloff in 2026 across software, data providers and other sectors. Price declines emerged amid growing concerns about AI-driven disruption, including advances in coding and business process agents. 

The market reaction illustrates the uncertainty surrounding AI-driven changes to established business models. Security risks add another layer of concern as organizations increasingly depend on AI workloads to support critical business operations. These workloads concentrate valuable data and intellectual property in highly connected environments, raising security considerations for the data centers that host them. 

• Sensitive data exposure: AI workloads may process proprietary business information, customer records, research, financial information and other sensitive datasets. Strong access controls and encryption help keep this information within authorized environments.
• High-value computing infrastructure: GPUs, accelerators, servers, storage systems and networking equipment represent significant computing resources. Unauthorized access or malicious activity affecting these components could disrupt AI operations and expose valuable systems.
• Expanded attack surface: AI workloads connect models, datasets, APIs, applications, storage platforms, networks and orchestration tools. Each connection introduces security requirements that need to remain aligned across the environment.
• Model and data integrity: Training data and model files directly influence AI outputs. Security controls need to protect these assets from unauthorized changes, manipulation or tampering throughout the AI pipeline.
• Operational and financial impact: Disruptions to AI training and inference can consume valuable computing resources, delay projects and affect applications that depend on AI services. Security measures help maintain reliable access to the infrastructure supporting these workloads.

How to Secure AI Workloads in Data Centers

Securing AI workloads requires several protective layers across the data center environment. Each addresses a different part of the workload life cycle and helps teams maintain reliable, controlled operations. A layered strategy also gives security teams multiple opportunities to detect and contain threats as AI environments grow.

Secure the Physical Infrastructure

The scale of AI infrastructure makes physical protection especially relevant. Meta operates 15 data center campuses across 21 U.S. states, with a combined capacity of 12,690 megawatts. The company is also expanding this infrastructure to support its growing AI operations, illustrating how AI computing can span large data center environments.

As AI infrastructure continues to expand across data center campuses, physical access restrictions become increasingly important. Teams should secure servers, GPU systems, storage equipment and networking hardware through controlled entry points, authentication systems, surveillance and access logging. 

Administrative access requires similar attention. Strong identity controls help ensure that only authorized personnel can manage critical infrastructure.

Security teams also need to maintain the underlying technology. Regular patching for servers, operating systems, firmware, GPUs, and related components closes known vulnerabilities and supports a more resilient environment. Hardened configurations and secure boot mechanisms can provide additional protection where the hardware supports them.

Protect Networks and Control Workload Access

AI workloads depend on fast connections between computing resources, storage, applications and users. Network security must protect these connections while meeting the performance requirements of AI operations.

Network segmentation can separate AI environments from unrelated systems and limit the movement of unauthorized activity. Least-privilege access also helps ensure that users, applications, services and machine identities receive only the permissions required for their roles.

For data moving between systems, encryption helps secure these transfers. Network monitoring also gives security teams visibility into activity across compute nodes, storage platforms, applications and external services. Restricting unnecessary inbound and outbound connections further reduces opportunities for unauthorized access.

Zero-trust principles provide another layer by requiring authentication and authorization for access based on established policies. These controls remain useful as AI workloads span multiple systems, environments and administrative teams.

Secure AI Data, Models and Data Orchestration

AI security also depends on protecting the information that feeds models and the assets that AI systems generate. Training, validation and inference data may pass through several systems before reaching their destination, creating a need for consistent safeguards throughout the pipeline.

Encryption helps protect sensitive information at rest and in transit. Access controls can restrict datasets according to user roles, workload requirements and data sensitivity. Data lineage also gives teams visibility into where information originated and how it moves through AI workflows.

Apart from storage, data orchestration helps manage how information moves through the systems that support AI workloads, including its transformation and preparation. Secure orchestration requires appropriate permissions, reliable logging and controls that help verify data as it moves between systems.

Monitor Workloads and Respond to Threats

AI workloads need continuous monitoring because their activity can change as models, applications, users and data pipelines evolve. Security teams should establish visibility across multiple areas of the system:

• Logging and audit trails provide records that help teams investigate unusual events.
• Monitoring GPU utilization, data transfers, authentication activity and workload behavior reveals patterns that warrant further review.
• Automating alerts helps security personnel identify potential threats quickly. 

Response procedures should define how teams isolate affected systems, investigate incidents, restore services and review the controls involved.

Build Security Into Every Layer of AI Infrastructure

AI workloads are becoming a major driver of data center capacity, creating a growing need for safeguards across the infrastructure that supports them. Organizations can strengthen AI workload security by establishing controls at each layer and maintaining visibility as systems evolve. Protecting the infrastructure, connections, data, models and workload activity helps create a secure foundation for continued AI deployment.

# # #

About the Author

April Miller is a Technology writer with more than 5 years of experience writing on AI applications that help SMBs compete and Enterprises strategize. You can connect with her on LinkedIn.

The post How to Secure AI Workloads in Data Centers appeared first on Data Center POST.

TL;DR

• As enterprises shift to hybrid and multi-cloud architectures, a data center’s connectivity options matter as much as its physical location.
• 365 Data Centers’ expanded partnership with Megaport adds nine new Points-of-Presence, bringing its connectivity footprint to 15 U.S. facilities.
• The move reflects a broader industry shift: colocation providers are increasingly judged on their connectivity ecosystem, not just power and space.

# # #

For years, choosing a colocation facility largely came down to three questions: How much power is available, how reliable is the site, and how close is it to my end users. Today, a fourth question carries just as much weight: How easily can I connect from this facility to the cloud platforms, networks, and digital services my business actually runs on?

That shift is playing out visibly in the market. 365 Data Centers recently expanded its partnership with Megaport, a global Network-as-a-Service (NaaS) provider, adding nine new Points-of-Presence across facilities in Fort Lauderdale, Detroit, Chicago, Commack, Philadelphia, Rancho Cordova, Richardson, Smyrna, and Herndon. The expansion brings Megaport’s footprint across 365’s platform to 15 facilities nationwide – a scale-up worth examining for what it signals about enterprise infrastructure decisions more broadly.

Why On-Demand Connectivity Matters More Than Ever

Enterprises rarely run on a single cloud anymore. A typical mid-market or enterprise IT environment might touch AWS for compute, Azure for productivity tools, and a specialized platform like Oracle Cloud or IBM Cloud for a legacy workload that hasn’t migrated yet. Managing separate physical connections to each of these providers used to mean long procurement cycles and rigid, hard-to-change infrastructure.

Software-defined connectivity platforms change that equation. Instead of provisioning a dedicated line to each cloud provider individually, customers can provision, scale, or retire connections on demand through a single interface. That flexibility matters most when business needs shift quickly, such as for a merger, a new regional office, or a sudden capacity spike shouldn’t require months of lead time just to get connected.

Edge Markets Are Where This Plays Out

It’s worth noting where these new connections are landing. Markets like Commack, New York, and Smyrna, Georgia, aren’t traditional hyperscale hubs; they’re regional edge markets, closer to where mid-sized businesses and regional offices actually operate. As AI workloads and latency-sensitive applications push more compute closer to end users, connectivity in these secondary markets becomes just as important as in major metros.

This is a trend worth watching across the industry: Colocation providers are extending carrier-neutral, multi-cloud connectivity into markets that previously lacked it, narrowing the gap between edge and core infrastructure.

What This Means for IT Decision-Makers

For organizations evaluating colocation partners, a few practical takeaways emerge from this kind of expansion:

• Ask about the connectivity ecosystem, not just the facility specs. A data center with excellent power density but limited network options can still create bottlenecks.
• Favor providers offering on-demand, software-defined connections over fixed, long-term circuits, especially if your cloud strategy is still evolving.
• Consider regional and edge locations carefully. As more colocation providers build out connectivity in secondary markets, businesses gain more flexibility in balancing latency, cost, and resilience.

The Bigger Picture

Partnerships like this one reflect a maturing colocation market, where facilities are increasingly valued as connectivity hubs rather than just physical real estate. As hybrid IT becomes the default rather than the exception, the providers that make multi-cloud connectivity simple – rather than something customers have to engineer themselves – are the ones best positioned to support enterprise growth in the years ahead.

Click here to read the full announcement.

The post How 365 Data Centers and Megaport Are Redefining What “Connected” Means for Colocation appeared first on Data Center POST.

TL;DR

• Assured Communications is supporting the design, engineering and construction of a new cable landing station (CLS) at Digital Realty’s LAX12 data center in Los Angeles.
• The project brings Assured Communications’ subsea infrastructure expertise to a new cable landing station designed to strengthen connectivity between Asia Pacific and North America.
• The CLS will bring subsea landing infrastructure closer to terrestrial networks, cloud providers, carriers and data center infrastructure.
• Currently in development, the new facility is expected to begin supporting subsea cables in 2028.

# # #

Thousands of miles of ocean separate Asia Pacific and North America, but the digital infrastructure connecting the two regions ultimately comes down to a critical transition point: where subsea cables reach land and connect to the terrestrial networks that carry data onward.

In Los Angeles, Assured Communications is helping build a new piece of that global connectivity ecosystem.

The subsea infrastructure specialist is supporting the design, engineering and construction of a new cable landing station at Digital Realty’s LAX12 data center. The project is designed to establish a new landing point for subsea cable systems while bringing subsea infrastructure closer to the terrestrial networks serving the Los Angeles market.

For Assured Communications, the project puts its experience in subsea cable systems and landing infrastructure development to work on infrastructure designed to support connectivity across the Pacific.

“Assured Communications is supporting the design, engineering and construction of the project, drawing on its experience in subsea cable systems and landing infrastructure development. Projects like this are reshaping how global infrastructure is deployed and connected. We’re excited to help bring a new subsea cable landing station to Los Angeles to further strengthen connectivity between Asia Pacific and North America,” said Joel Ogren, Founder and Chief Executive Officer of Assured Communications.

For subsea cable operators, the new CLS is designed to provide a turnkey solution for landing and powering their systems while providing access to Digital Realty’s broader Los Angeles ecosystem. That integration will provide access to terrestrial networks, cloud providers and carriers across the company’s Los Angeles campus.

By bringing these resources together, the CLS will help create closer integration between subsea and terrestrial networks, giving carriers, cloud providers and content platforms additional routes and greater flexibility to connect between Asia Pacific and North America.

“This development reflects our continued focus on expanding Digital Realty’s role in supporting global connectivity,” said Andy Power, President and Chief Executive Officer of Digital Realty. “By bringing together subsea landing infrastructure with our existing interconnection and data center platform, we are strengthening the intercontinental connectivity available across our Los Angeles campus and providing our customers with a more seamless and scalable environment for connecting across regions.”

With the project currently in development, the new CLS is expected to begin supporting cables in 2028.

The Los Angeles project is another example of Assured Communications putting its subsea infrastructure experience to work, supporting the development of the landing infrastructure needed to connect subsea cable systems with terrestrial networks. As the project advances toward supporting cables in 2028, Assured will continue supporting the design, engineering and construction of the new CLS.

Additional details are available in the full announcement: globenewswire.com/news-release/2026/09/22/3366954/0/en/digital-realty-to-develop-new-cable-landing-station-in-los-angeles.html 

The post Building the Next Subsea Gateway Between Asia Pacific and North America appeared first on Data Center POST.

TL;DR

• AlixPartners’ 2027 Data Center Market Outlook finds that 83% of respondents expect AI inference to require silicon architectures different from today’s GPUs, while 86% of operators anticipate moderate-to-major facility upgrades, redesigns or replacements.
• Near-term market strength is accompanied by uncertainty: 70% of respondents report moderate or strong pricing power, but 64% have reliable demand visibility extending only 12 months or less.
• Rising costs, technology changes and financing challenges are expected to drive industry consolidation, with 62% of respondents viewing data center M&A as attractive now or within the next 12 months.

# # #

The data center industry is entering a pivotal year as artificial intelligence requirements, infrastructure constraints and regulatory pressures reshape the market. According to AlixPartners’ 2027 Global Data Center Market Outlook, demand continues to outpace supply, but limited visibility and rapid technology changes are creating new risks for operators and investors.

The report draws on responses from more than 480 senior executives across the data center ecosystem in North America, Europe, and the Middle East. Participants represented builders, operators, AI infrastructure companies, investors, and lenders working across hyperscale, colocation, edge, enterprise, and neocloud environments.

One of the report’s key findings is the expected transition from AI training to inference workloads. Eighty-three percent of respondents believe inference will require silicon architectures different from today’s GPUs, while 86% of operators anticipate moderate-to-major upgrades, redesigns or replacements of existing facilities. Higher rack densities, direct liquid cooling, and new network configurations could make this transition especially challenging for operators with older infrastructure or limited capital.

Although 70% of respondents currently report moderate or strong pricing power, confidence in future demand remains limited. Sixty-four percent have reliable demand visibility extending only 12 months or less, while 60% believe a major market disruption is extremely or very likely within the next two years. This combination leaves the industry balancing near-term strength against longer-term uncertainty.

Financial pressure is also expected to increase. Sixty-four percent of respondents anticipate more distressed data center situations during the next 12 to 18 months, driven primarily by rising energy costs, insufficient operating cash flow, technology disruption, and competition. Smaller operators may face the greatest risk due to reduced pricing power, debt-maturity challenges, and limited ability to fund necessary upgrades.

Construction and operational challenges will remain central in 2027. Project coordination and construction delays were each cited by 66% of respondents as leading capital deployment challenges, while regulatory restrictions and local community opposition ranked as the top financing obstacle. The findings suggest that operators capable of modernizing their facilities, improving operational efficiency, and maintaining flexibility will be best positioned to navigate the market ahead. Explore AlixPartners’ 2027 Global Data Center Market Outlook for deeper insights into the trends, risks and opportunities shaping the data center market in the year ahead. 

The post AI Demand, Infrastructure Constraints Set the Stage for a Transformative 2027 Data Center Market appeared first on Data Center POST.

TL;DR

• Queue Totals Reflect Intent, Not Delivery: Alberta’s interconnection queue has reached 19,565 megawatts, measuring speculative demand rather than actual capacity being built
• Grid Allocations Are Strictly Constrained: Initial connection allocations are fully committed, leaving developers without existing approvals unable to obtain credible energization dates while permanent frameworks are finalized
• Priority Favors “Bring-Your-Own-Generation”: Under Alberta’s Data Centre Regulation, queue priority is granted to projects that pair data loads with new dedicated generation or storage
• Behind-the-Meter Power Bypasses Queue Delays: Co-located, off-grid self-supply under AUC Rule 007 avoids interconnection requests entirely, trading unmanageable queue delays for manageable execution risks

# # #

By Jolie Kahn, Chief Executive Officer and Director, FingerMotion, Inc.

Alberta has become the most quoted jurisdiction in North American data center development, and the figure doing most of that quoting is the interconnection queue. As of July 30, 2026, the Alberta Electric System Operator reported approximately 19,565 megawatts of requested data centre load awaiting connection. The number is generally presented as evidence of a boom.

It is worth setting that figure beside another one. Alberta’s all-time peak demand — every home, business and industrial facility in the province, drawing at once — is roughly 12,800 megawatts. The queue is more than one and a half times everything Alberta has ever consumed at a single moment.

A number that large is not a forecast of supply. It is a measure of how many parties want the same scarce thing.

A Queue Measures Demand, Not Delivery

Entering an interconnection queue is inexpensive relative to building anything. A queue position commits no one to construction, no one to an offtake agreement and no one to a capital plan. It is closer to an option than a project.

This matters because queue totals are increasingly cited as though they describe capacity that is on its way. They do not. They describe intent, and intent in a constrained market is abundant. Operators evaluating where to site workloads over the next three years are better served by asking a narrower question: how much capacity in this jurisdiction is permitted, and how much is actually energized.

The First Allocation Is Already Spent

Alberta’s answer to the volume of requests was to limit them. Phase 1 of the AESO’s approach permitted 1,200 megawatts of new data centre connections, and that allocation is now fully committed. Phase 2, the permanent framework, is still being finalized. An interim bridging approach allows limited additional load under strict conditions, capped system-wide at 1,600 megawatts, limited to three years per project, and requiring that data centres reduce demand first when the grid is stressed.

The practical consequence is straightforward. A developer without an existing allocation cannot presently provide a credible grid energization date. Not a pessimistic one — none at all, because the framework that would produce it is not yet final.

The Queue Sorts By Who Brings Power

Alberta’s Data Centre Regulation, effective June 2026, formalized the tiebreaker: projects that bring dedicated new generation or storage are prioritized in the connection queue. The province’s bring-your-own-generation framework asks a developer to pair new generation with the load it intends to draw. Meta’s Sturgeon County campus is the widely cited example, combining a grid connection with a multi-billion-dollar private generating station.

This reframes the planning problem. The binding constraint on AI infrastructure in a jurisdiction like Alberta is not land, and it is not hardware. It is permitted power. Capital and equipment can be procured on commercial timelines. A connection cannot be procured at all when the allocation is closed.

Below The Threshold, A Different Path Exists

There is a second route, and it is less discussed because it does not appear in queue statistics at all.

Alberta Utilities Commission Rule 007 governs facility applications for power plants, substations, transmission lines and industrial system designations. Smaller individual power plants fall into a lighter review than large facilities. An Industrial System Designation permits co-located generation and load to be treated as a single industrial system for self-supply purposes, so that internal wiring is not regulated as transmission.

A facility developed this way generates and consumes power on site. It makes no interconnection request, holds no queue position and requires no transmission service arrangement. Because each application is comparatively small, several sites may be permitted in parallel over a period measured in months rather than in the years an interconnection can require.

That is a meaningful difference in a market where the grid path currently offers no date at all.

The Trade Is Real

Behind-the-meter generation is not a shortcut, and it should not be described as one.

Generation capital that the utility would otherwise carry moves onto the developer’s balance sheet. Fuel supply must be contracted rather than assumed. Reliability becomes an operating responsibility rather than a purchased service, and a site without a grid connection has no system to lean on during an outage. Permitting itself is not guaranteed; municipal processes, landowner agreements and regulatory review all consume time, and timelines slip.

The honest characterization is that this path trades one set of risks for another. It substitutes execution risk, which a capable operator can manage, for queue risk, which no operator can manage at all.

What To Ask Instead

The industry would be better served by reporting Alberta’s numbers in four categories rather than one: requested, allocated, permitted and in service. Those are very different quantities, and at present only the first is widely quoted.

For anyone sitting capacity in a constrained jurisdiction, the more useful question is not how large the queue has grown. It is whether a given project needs the queue at all.

# # #

About the Author

Jolie Kahn is Chief Executive Officer and Director of FingerMotion, Inc. (Nasdaq: FNGR). She has been a public-company CEO, CFO and corporate finance attorney for more than three decades, with over $5 billion in capital raises and experience spanning data centers, power, capital markets, fintech and high-growth technology. She previously served as Principal Outside Counsel and General Counsel to a major Bitcoin mining company, overseeing SEC compliance, governance, financings and regulatory matters. She holds a BA from Cornell University and a J.D. magna cum laude from Benjamin N. Cardozo School of Law.

Jolie Kahn has also written on modular data center economics for the Forbes Business Council.

The post A Queue is Not a Pipeline: What Alberta’s Data Center Numbers Actually Say appeared first on Data Center POST.

TL;DR

• AI introduces an operational scaling challenge rather than a raw storage capacity issue, as continuous data flows across training, inference, backup, and archival overwhelm traditional, isolated storage systems.
• Managing fragmented infrastructure has become the primary bottleneck, with separate tools, security policies, and unpredictable AI workloads creating massive operational overhead.
• Autonomous data infrastructure unifies these workloads into a single namespace, automating data placement, lifecycle optimization, and cyber resilience directly at the storage layer.
• IT professionals can transition from routine maintenance to strategic architecture, focusing on governance, security, and system design while allowing adaptive platforms to handle operational tasks autonomously

# # #

By Erwan Girard, Chief Product Officer at Scality

Enterprise infrastructure has always evolved in response to scale. Virtualization addressed server sprawl. Cloud computing reduced the need to provision physical resources for every new application. Automation helped administrators manage environments that had grown too large to operate manually.

Artificial intelligence presents a different kind of scaling problem.

The discussion around enterprise AI has largely centered on models, GPUs, and inference performance, but those technologies represent only a fraction of what organizations must operate. Every production AI deployment creates a continuous flow of data that must be ingested, protected, moved, analyzed, retained, governed, and eventually archived. Those activities place demands on infrastructure that are very different from the workloads storage systems were originally designed to support. 

This is becoming increasingly apparent as organizations move beyond pilot projects. AI is no longer a single workload running on isolated infrastructure. A single application may include high-speed storage for model training, object storage for inference data, lower-cost capacity for operational datasets, immutable storage for cyber resilience, and long-term archives to satisfy regulatory requirements. Traditionally, those functions have been handled by separate products with separate management tools, security policies, and operational teams. That architecture worked reasonably well when data moved slowly and applications followed predictable lifecycles. AI changes both assumptions. 

Training datasets expand continuously. New models are introduced far more frequently than traditional enterprise applications. Inference workloads fluctuate as demand changes. The same dataset may move repeatedly between active processing, backup, compliance, and archival over its lifetime. Each transition introduces another operational task, another opportunity for inconsistency, and another point where administrators must intervene. Before long, the effort required to manage the infrastructure begins to rival the effort required to build the AI applications themselves.

Complexity Has Become the Real Infrastructure Challenge 

For years, the answer to operational complexity was automation. Administrators automated provisioning, scripted maintenance, and orchestrated repetitive tasks. Those capabilities remain valuable, but they were designed to execute predefined actions under predefined conditions. AI environments are considerably less predictable. Infrastructure must continually adapt to changing workloads, shifting performance requirements, evolving security policies, and rapidly growing data volumes, often without the benefit of stable operating patterns.

That is where autonomous data infrastructure represents something more substantial than another automation framework. Rather than treating storage as a collection of independent systems, it starts with the assumption that the platform itself should continuously optimize how data is managed throughout its lifecycle. Capacity, performance, protection, and cost become policy decisions rather than infrastructure projects. Data moves between performance tiers automatically according to business requirements instead of being exported, migrated, and re-imported into separate platforms. A single namespace spans workloads that historically required multiple storage systems, allowing infrastructure to evolve without repeatedly forcing administrators to redesign the environment. 

That architectural change may ultimately prove more important than the automation itself. Many organizations underestimate how much operational complexity accumulates simply from running multiple storage platforms. Every environment has its own authentication model, monitoring tools, lifecycle policies, upgrade schedules, recovery procedures, and performance characteristics. As AI expands across the enterprise, those management layers multiply alongside the data. Reducing the number of operational boundaries often creates greater long-term value than introducing another layer of orchestration.

The same principle applies to cyber resilience. AI has increased the value of enterprise data far beyond traditional business records. Training datasets, model checkpoints, vector indexes, and inference pipelines have become strategic assets in their own right. Protecting them requires more than backup software. It requires infrastructure that assumes failures and attacks will occur and is designed to recover without depending on manual intervention. 

The conversation also extends beyond security to control. As AI initiatives become more strategic, organizations are under growing pressure to understand where data resides, who can access it, and which legal and regulatory frameworks govern it. That is especially true for enterprises operating across multiple countries or in highly regulated industries, where data residency requirements, digital sovereignty initiatives, and industry-specific compliance obligations increasingly influence infrastructure decisions. Rather than treating these as separate governance exercises, modern infrastructure must make location, retention, and access policies part of the data lifecycle itself, enabling organizations to meet regulatory requirements without introducing additional operational complexity.

One of the more significant design decisions behind autonomous data infrastructure is that immutability exists within the storage engine itself rather than being implemented solely through administrative policy. Instead of modifying existing data in place, new versions are written separately while previous versions remain intact. Combined with distributed self-healing that rebuilds only affected objects instead of entire disks, this creates a fundamentally different operational model for resilience. Recovery becomes part of normal system behavior instead of an exceptional event requiring administrators to coordinate lengthy repair efforts. 

From Infrastructure Operators to Infrastructure Architects

Perhaps the most interesting implication has little to do with storage technology itself. Infrastructure teams are already responsible for environments that are growing faster than headcount, and AI is accelerating that imbalance. The objective is not to remove people from operations, but to reduce the amount of time highly skilled engineers spend on repetitive maintenance that adds little strategic value. As more routine activities become policy-driven and continuously optimized, infrastructure professionals can devote more attention to architecture, governance, capacity planning, and aligning technology decisions with business priorities.

That evolution mirrors what is happening across software engineering, networking, and cybersecurity. AI is steadily shifting human expertise away from repetitive execution and toward system design, governance, and strategic decision-making. Autonomous data infrastructure reflects the same progression. Rather than asking administrators to manage an ever-growing collection of storage products, it treats the infrastructure as an adaptive system that operates within policies established by the people responsible for it. The most effective approach is not to take humans out of the loop, but to keep them in control of the decisions that shape security, compliance, and business outcomes while allowing the platform to execute routine operational tasks autonomously. 

Viewed from that perspective, autonomous data infrastructure is less about storage than it is about preparing enterprise IT for the next decade. AI has exposed the limitations of architectures built around isolated products, manual coordination, and steadily increasing operational overhead. Organizations will continue investing in faster GPUs and more capable models, but those investments will deliver their greatest value only if the infrastructure beneath them becomes equally capable of managing complexity. The next generation of enterprise infrastructure will not simply store data more efficiently. It will actively participate in operating the environments that modern AI depends upon.

# # #

About the Author

Currently serving as the Chief Product Officer at Scality, Erwan Girard has been instrumental in shaping the company’s product strategy and driving its success. In his tenure, he has built an all-round product management team focused on serving and bringing value to Scality customers and prospects. He has led RING product evolutions and launched Artesca in order to better address the Object storage market resulting in significant revenue growth and market expansion.

The post AI Didn’t Create a Storage Problem. It Created an Operations Problem. appeared first on Data Center POST.

TL;DR

• Redefining the Assessment Boundary: While data centres carry a significant operational and lifecycle footprint, judging them solely at the facility boundary creates an incomplete environmental evaluation that ignores what the digital infrastructure enables, reduces, or replaces.
• Measuring Net System Impact: A complete evaluation requires a three-part ledger accounting for the direct lifecycle burden, enabled environmental benefits (avoided physical journeys, energy savings, extended asset life), and potential rebound demand.
• Testing Structural Displacement: The Structural Displacement Ratio measures whether enabled benefits, after subtracting rebound effects, exceed the direct lifecycle burden, requiring verifiable evidence rather than assumed environmental credit.
• Demanding Independent Assurance: Digital infrastructure is not inherently sustainable; policymakers, investors, and operators must mandate independent assurance and clear evidence quality controls across the entire system boundary

# # #

A complete environmental ledger must measure direct burden, enabled benefit and rebound

The environmental cost of data centres is real, but a facility boundary is not a system boundary. Decision makers need evidence of the physical activity digital infrastructure enables, reduces or replaces, alongside an uncompromising account of its own lifecycle burden.

The data centre has become one of the most visible symbols of the environmental cost of the digital economy. Electricity demand is large and concentrated. Cooling can place pressure on water systems. Construction consumes land, concrete, steel and electrical equipment, while servers rely on complex supply chains and short replacement cycles.

Those impacts must be measured without euphemism. Yet visibility is not the same as causation, and measuring only what enters the facility can produce an incomplete environmental judgement. A data centre does not exist for its own sake. It processes information used by buildings, utilities, transport systems, businesses, public services and consumers. The relevant question is therefore not simply what the facility consumes. It is what the complete digital system consumes, enables, reduces and replaces.

The wrong boundary

A kilowatt hour consumed in a server hall is immediately visible. The effect of a digital service elsewhere in the physical economy is usually dispersed. Energy saved through intelligent building controls may be distributed across thousands of assets. Water preserved through network leakage detection appears through many small interventions. Avoided journeys, more efficient logistics and longer equipment life are rarely attributed back to the digital infrastructure that enabled them.

This creates an asymmetric ledger. The direct burden is counted at the facility boundary, while possible reductions in physical consumption remain outside it. The answer is not to award data centres an assumed environmental credit. It is to widen the system boundary and apply the same evidential discipline to both sides.

Three ledgers and an evidence test

A practical Net System Impact assessment begins with three distinct ledgers. The first records the direct lifecycle burden: site, construction, embodied carbon, operational electricity, cooling, water, backup generation, network infrastructure, hardware manufacture, replacement, electronic waste, decommissioning and reuse.

The second record enabled environmental benefits. This may include physical activity demonstrably reduced through digital services, such as avoided travel, optimised energy use, prevented water loss, reduced material consumption or extended asset life. These benefits must be tested against a credible counterfactual. What would have happened without the digital intervention, and would an equivalent outcome still have been required?

The third records rebound and additional demand. Efficiency can lower cost and increase convenience, causing total consumption to rise. New digital capability can also create activity that did not previously exist. These effects cannot be hidden inside a net figure.

A fourth control sits across all three ledgers: evidence quality. The boundary, baseline, causal relationship, allocation method and assurance status must be visible. Where evidence is weak, the claimed benefit should be discounted rather than promoted as fact.

The structural displacement test

The proposed Structural Displacement Ratio asks whether the enabled benefit, after rebound, exceeds the direct burden. In simple terms, the test is: Structural Displacement Ratio = enabled benefit minus rebound, divided by direct lifecycle burden.

A result above one may indicate that the system eliminates more physical waste than it creates, but only within the environmental category being tested and only when the evidence is credible. A carbon result cannot erase a material failure involving water, land, biodiversity, resilience or community impact. Nor should a favourable calculation become a permanent entitlement. 

Counterfactuals change as grids decarbonise, regulations tighten and alternative technologies improve. The assessment must therefore expire and be retested.

Technology supports proof but does not create it

Internet of Things devices can connect environmental claims to measured physical conditions. Artificial intelligence can interpret complex data and optimize performance. Blockchain can preserve provenance, permissions, attribution and retirement across organisations. None of these technologies creates environmental value independently.

Value arises when trusted information changes physical outcomes. A smart building that gathers more data but consumes the same resources has not demonstrated an environmental benefit. A predictive maintenance system matters when evidence shows that it prevented failure, avoided replacement or extended useful life. A digital logistics platform matters when it reduces real journeys, fuel or wasted capacity against a defensible baseline.

A decision framework for owners and regulators

Data-centre planning and investment decisions should therefore require a complete lifecycle account and a defined statement of the physical systems the proposed workloads are expected to influence. 

Operators and customers should identify measurable outcomes, allocate benefits conservatively and prevent double counting between the data centre, software provider, asset owner and end user.

Independent assurance is also essential. The party benefiting from a sustainability claim should not be the only party defining the counterfactual or verifying the result. Evidence should be capable of review by investors, regulators, customers and affected communities without disclosing commercially sensitive operational data unnecessarily.

The conclusion is deliberately conditional. Digital infrastructure is not inherently sustainable, and scale alone will not make it so. Poorly located or weakly governed facilities can deepen pressure on grids, water systems and communities. Equally, treating the data centre as an isolated consumer can ignore the inefficient physical activity that digital services may displace.

The data centre is not the real white whale. The real target is the accounting boundary that separates what digital infrastructure consumes from what it replaces. Society should neither pursue data centres as convenient symbols nor accept the industry’s benefits without proof. It should demand a complete ledger.

Full paper

The complete TUTUM white paper, Is the Data Centre the Next White Whale?, including the full Net System Impact Assessment and illustrative calculations, is available here. 

# # #

About the Author

Mark Burns is Executive Director of TUTUM Group and a chartered development and infrastructure leader with nearly three decades of experience across sustainable master development, real estate, infrastructure and complex programme delivery. He previously served as Senior Development Director at Masdar City and has delivered six LEED Platinum projects client-side. His work focuses on asset governance, programme assurance, sustainability and the relationship between physical infrastructure and digital capital.

The post The Data Centre Should Be Judged by What It Replaces appeared first on Data Center POST.

TL;DR

• Major local investment: Empire Fiber Internet has launched 100% fiber internet in Berwick, PA, supported by a $1.4 million investment and now available to more than 2,800 homes and businesses.
• New choice for connectivity: The build gives Berwick residents and local employers access to symmetrical fiber speeds, responsive local support and service plans up to 2 Gig.
• Momentum across the region: The Berwick launch builds on Empire Fiber Internet’s continued growth across Pennsylvania and New York, including recent expansions in the Lehigh Valley, Steuben County, Wellsville, Ithaca, and Victor.

# # #

Empire Fiber Internet, a leading fiber optic internet service provider serving communities across New York and Pennsylvania, has launched its 100% fiber network in Berwick, PA, bringing fast, reliable internet service to more than 2,800 homes and businesses.

Enabled by a $1.4 million local investment, the new network gives the Columbia County community another choice for the high-capacity connectivity increasingly essential to daily life and local commerce.

For Berwick, the investment comes at an important time. High-speed broadband has become foundational infrastructure for families, schools, employers and small businesses alike as it enables remote and hybrid work, online learning, telehealth, cloud-based business applications, e-commerce and customer communications. As Berwick continues to support local employers, attract investment and meet the digital needs of residents, the availability of a new fiber network strengthens a critical piece of the community’s infrastructure.

“Reliable, high-speed internet is essential infrastructure for today’s businesses, whether it’s a manufacturer moving large amounts of data, a small business processing transactions, or an employer simply trying to stay connected with customers and employees,” said Chris Berleth, President/CEO of the Columbia Montour Chamber of Commerce. “Empire Fiber Internet’s investment in Berwick brings another broadband option to the community, and that kind of private investment and increased choice helps strengthen the environment in which our local businesses operate and grow.”

This latest launch also reflects Empire Fiber Internet’s broader expansion strategy across Pennsylvania and New York. In 2026 alone, the company announced new or expanded fiber availability in Pennsylvania’s Lehigh Valley and in New York communities including Steuben County, Wellsville, Ithaca, and Victor.

To learn more about Empire Fiber Internet, visit empirefiber.com.

The post Empire Fiber Internet Connects More Than 2,800 Berwick Homes and Businesses appeared first on Data Center POST.

TL;DR

• • Community concerns about data centers often extend beyond infrastructure to questions about AI, resource use, surveillance and Big Tech.
  • Effective community engagement starts with understanding local priorities rather than immediately presenting a project’s benefits.

• Communities need clear information about potential impacts, including power, water, noise, land use and economic development.
• Engagement strategies should reflect the priorities and characteristics of each individual community.
• Sharing community engagement lessons across the industry could help improve how future projects are introduced and developed.

# # #

Data center development is accelerating to support growing demand for cloud computing, artificial intelligence and other digital services. At the same time, projects are facing increased public scrutiny as communities ask more questions about power, water, land use, environmental impacts and what large-scale digital infrastructure development means for them.

A recent Cool Vector podcast episode, “Data Centers Have a Community Engagement Problem,” explored this changing environment with Megan Baker, Founder and Principal of Tuli Public Affairs; Ilissa Miller, Founder and CEO of iMiller Public Relations (iMPR); and Phillip Koblence, Co-Founder of Nomad Futurist and Head of Critical Ventures. Moderated by Cool Vector host David Snow, the conversation examined why community engagement needs to become a more integral part of data center development.

Understanding What Is Driving Community Concerns

For decades, much of the public had little reason to think about the physical infrastructure supporting digital services. Today, data centers are increasingly associated with AI, energy demand, water consumption, Big Tech and other issues receiving significant public attention.

“People fundamentally fear what they are not comfortable with,” said Koblence. That uncertainty can make technical explanations alone insufficient when residents have broader concerns about how development could affect their communities.

Miller emphasized the emotional component behind some of those concerns. “It’s all rooted in fear and whether that fear is justified or not is secondary, but it’s a real emotion,” she said. “And so that’s why the facts don’t matter, because it’s emotional. They’re afraid of AI. They’re afraid of surveillance.”

Addressing those concerns requires understanding the questions behind them. Communities need information about potential impacts, but they also want to understand who is developing the project, why it is coming to their community and how local priorities will be considered.

Start With Listening

Rather than beginning with tax revenue, investment, or other anticipated project benefits, the discussion emphasized listening to communities before developing an engagement strategy. Understanding local priorities can help shape how developers communicate and identify issues that may need to be addressed earlier in the process.

“They have to slow down, listen,” said Miller. She later summarized the approach simply: “Go to these communities with curiosity.” This community-first approach puts greater emphasis on understanding what residents and local stakeholders want, need, and value before determining how a project should be communicated.

No two communities are the same, which makes a one-size-fits-all approach to community engagement ineffective. Some communities may prioritize economic development, while others may be more focused on natural resources, preserving community character or managing growth.

“We need to recognize too that each community is so different, right?” said Baker. “It can’t be a blanket procedure of going into a community and just deploying the same tools or the same process over and over again. You need to spend the time to understand, what does this community value?”

Community research and stakeholder engagement conducted early in the development process can help answer those questions before a project becomes the subject of public debate. It can also provide local leaders with better information when residents begin asking questions about power, water, noise, jobs, taxes and other potential impacts.

Sharing Lessons Across the Industry

Community engagement in one market can also influence perceptions elsewhere. Residents may view controversial projects as representative of the broader data center industry rather than the actions of a single developer, making community relationships an industry-wide concern.

Baker argued that companies have an opportunity to learn from one another rather than treating engagement strategies as proprietary. “The community engagement and sustainability tactics of these projects aren’t proprietary and they shouldn’t be. And we should be more collaborative about sharing the things that are going well and the things that aren’t going well so that we can all be moving forward with less friction,” she said.

The podcast also highlighted the Digital Infrastructure Framework being developed through OIX, which is intended to provide communities with guidance for understanding and evaluating digital infrastructure development. Efforts such as these could give municipalities additional tools for asking questions while helping developers better understand what communities need from them.

As data center development continues, community engagement is becoming another important consideration alongside power, land, connectivity and permitting. Starting earlier, listening before communicating project benefits and approaching each community individually can help developers build stronger relationships with the people and places that will ultimately host digital infrastructure.

To hear the full conversation with Megan Baker, Ilissa Miller and Phillip Koblence, listen to Cool Vector’s “Data Centers Have a Community Engagement Problem” on Substack here.

The post Data Centers Have a Community Engagement Problem: Why Listening Must Come First appeared first on Data Center POST.

TL;DR

• Physical constraints dictate deployment viability: Rack power density, liquid cooling capabilities, and grid interconnection timelines now determine AI deployment success rather than traditional floor space or total facility square footage.
• Avoidable sequencing and architecture mistakes: Enterprise buyers frequently purchase high-powered accelerators before verifying per-rack power or cooling constraints, or misallocate budgets by buying training-class clusters for inference workloads.
• Low utilization is the dominant cost leak: Unscheduled idle capacity, overprovisioned inference clusters, and unmonitored data egress fees multiply effective costs far more than hardware unit prices.
• Strategic preparation for 2026: Organizations must audit legacy colocation contracts, price data exit costs, and instrument actual utilization metrics before committing to new hardware procurement

# # #

The AI data center infrastructure conversation has been dominated by the supply side: megawatts secured, land acquired, cooling retrofitted, grid queues navigated. Less attention has gone to the demand side, where enterprise buyers are making commitments they do not fully understand.

That gap is becoming expensive. Organizations are signing capacity agreements, buying accelerators, and choosing platforms based on assumptions that stopped being true two hardware generations ago. Operators and providers see the results: customers arriving with hardware their contracted space cannot power, or with clusters that run at a fraction of their capability.

This piece looks at what has actually changed at the facility level, and where buyers on the other side of the table are getting it wrong.

What Has Actually Changed in AI Data Center Infrastructure?

The changes that matter are physical rather than digital. Rack power density, heat removal, and grid interconnection timelines now determine what is possible, and none of them can be resolved through procurement or software.

Three shifts sit underneath everything else.

Power density outgrew existing facility design. Halls provisioned for general enterprise computers were built around assumptions about per-rack draw that AI accelerator deployments break. Total facility capacity has become a less useful figure than what a single rack position can deliver.

Air cooling reached a practical ceiling. Beyond a certain density, the airflow and volume required make air cooling uneconomical before it becomes technically inadequate. Liquid has moved from specialist option to default assumption for dense AI deployments.

Power availability became the site selection criterion. In multiple markets, interconnection timelines now exceed construction timelines. That has pushed site selection toward available electricity and driven interest in on-site generation and long-term power purchase agreements.

For the industry, none of this is new. For the enterprise buyer signing a three-year agreement, most of it still is.

Why Rack Density Has Become the Deciding Specification

Rack-level power availability is now the specification that determines whether a deployment is viable. Facility totals and floor area say little about whether specific AI hardware can be installed and operated in a given position.

This is where the most common buyer failure occurs. A customer evaluates space on square footage and total capacity, signs, and then discovers that the contracted kilowatts per rack falls short of what their accelerator configuration draws. The hardware is purchased. The space is contracted. Neither can be used as planned.

Providers increasingly see this arrival as an escalation rather than a question. The underlying problem is that legacy procurement habits carried into a workload category they were never designed for.

The practical guidance for buyers is narrow and specific: treat power per rack as a contractual figure, confirm it in writing, and establish whether the position supports the density today or only after a future phase of work.

How Cooling Moved from Utility to Design Constraint

Cooling has shifted from a background consideration to a primary constraint on deployment. Most new high-density AI capacity assumes some form of liquid cooling, and retrofit programs are widespread across existing estate.

The point buyers most often miss is that liquid cooling is a facility decision, not an IT decision. It involves floor loading, plumbing routes, leak detection, maintenance access, and, in leased environments, the landlord. Enterprise teams that select accelerators before establishing cooling capability create a sequencing problem that takes months to unwind.

Water use has also moved up the agenda. Cooling designs vary considerably in consumption, and in water-stressed regions this has become a permitting and community-relations matter. Water usage effectiveness (WUE) now appears alongside PUE in enterprise supplier questionnaires with increasing frequency.

Why Networking and Storage Are Part of the Compute Decision

AI training clusters generate heavy east-west traffic between nodes rather than the north-south pattern typical of enterprise applications. The interconnect stops being a supporting system and becomes part of the compute platform itself.

In distributed training, nodes synchronize constantly. A slow fabric leaves expensive accelerators waiting, which is why these environments use InfiniBand or high-speed RDMA over Converged Ethernet rather than standard enterprise networking.

Storage follows the same logic. Training reads large volumes continuously, and a tier that performs adequately for business applications will starve a cluster. Bandwidth matters more than capacity, which is why NVMe tiers and parallel file systems sit in front of colder object storage.

There is a third factor that shapes buyer decisions more than either, and it is rarely priced during evaluation. Data gravity constrains architecture over time. Computers are comparatively portable; large datasets are not. Once training data settles in a particular provider or region, moving it becomes slow and expensive, which quietly narrows every future platform option.

Buyers evaluate computer pricing carefully. Very few price the exit.

Training or Inference: The Distinction Most Buyers Miss

Most enterprise organizations need inference capacity, not training capacity. Training builds or fine-tunes models and demands large, tightly coupled clusters. Inference runs finished models and is latency-sensitive, distributed, and far less demanding per request. Conflating the two drives a substantial share of misdirected spending.

This distinction has commercial consequences for providers as well as buyers. An enterprise customer asking about training-class capacity when their actual requirement is regional inference is a customer who will either overbuy and churn, or discover the mismatch during deployment.

The qualifying question is simple and rarely asked early enough: is this organization building models, or deploying them?

Build, Collaborate, or Rent?

For most enterprise buyers, renting remains the correct default. Ownership makes sense at sustained, predictable utilization. The deciding variable is workload stability, not unit pricing.

Ownership converts a variable cost into a fixed one. That is an advantage when demand is stable and a liability when it is not and enterprise AI demand in 2026 is still, for most organizations, unstable.

Where AI Infrastructure Budgets Actually Leak

The largest avoidable cost in AI infrastructure is low utilization, not unit price. Idle accelerators cost what busy ones cost. A cluster running well below capability multiplies effective cost per useful hour, and no negotiated rate compensates for it.

The recurring leaks are consistent across organizations:

• Idle capacity between workloads, where provisioning matches peak demand but scheduling does not exist
• Overprovisioned inference, sized permanently for spiky traffic
• Data movement, where egress and cross-region transfer are underestimated during evaluation
• Duplicate environments, with development and experimentation clusters running continuously
• Model inefficiency, where a larger model than the task requires becomes an infrastructure cost

The first meaningful saving in most organizations comes from scheduling and right-sizing rather than from buying differently. That is an uncomfortable message for a procurement cycle, but it is generally the accurate one.

A Preparation Checklist for 2026

Preparation is mostly a sequencing problem. The technical decisions are less difficult than the order in which they are made.

1. Classify workloads before selecting a platform. Separate training, fine-tuning, and inference, with volume and latency requirements for each.
2. Audit existing contracts. Check power per rack, cooling capability, and expansion rights in agreements signed before AI hardware became a requirement.
3. Treat power and equipment lead times as project constraints, not procurement details.
4. Price the exit. Model the cost of moving data and workloads to another provider before committing.
5. Instrument utilization before expanding. Accelerator utilization, queue depth, and cost per inference should be measured, not estimated.
6. Plan for shorter refresh cycles. Accelerator generations are moving faster than traditional depreciation models assume.
7. Establish the compliance position early. Where data is processed and moved affects obligations under GDPR, sector frameworks such as HIPAA, and emerging AI-specific regulation.
8. Assess capability alongside capacity. Platform engineering, MLOps, and data engineering gaps account for more failed deployments than hardware decisions do.

Frequently Asked Questions

How much power does an AI server rack need?

Substantially more than a traditional enterprise rack, with the gap widening across accelerator generations. Rather than relying on a general figure, buyers should obtain the specific draw for their configuration from the hardware vendor and confirm in writing that the facility can deliver it per rack position.

Can existing data centers be retrofitted for AI workloads?

Often yes, but the work is structural. Retrofits typically involve electrical distribution upgrades, cooling infrastructure, floor loading assessment, and plumbing. Timelines are driven by equipment lead times and, where applicable, grid connection. A retrofit should be scheduled as a construction project, not an IT refresh.

Is liquid cooling required for all AI workloads?

No. Lower-density AI deployments can still run on air cooling or rear-door heat exchangers. The determining factor is the power density of the specific hardware, which establishes whether air can remove the heat economically. Dense training and high-throughput inference deployments generally require liquid.

Is it cheaper to run AI workloads on-premise or in the cloud?

Neither is universally cheaper. Cloud trends to cost less for variable demand; colocation or ownership tends to cost less at high, sustained utilization. Comparisons that exclude data transfer, staffing, refresh cycles, and exit costs typically overstate the case for ownership.

What is the difference between PUE and WUE?

Power usage effectiveness measures how much of a facility’s total energy reaches IT equipment rather than overhead such as cooling. Water usage effectiveness measures water consumed per unit of IT energy. Both now appear in enterprise supplier assessments, with WUE gaining prominence as cooling water use attracts regulatory attention.

How long does it take to deploy AI infrastructure?

It depends on the route. Cloud capacity can be immediate. Colocation depends on hardware lead times and facility readiness. Building or substantially retrofitting is measured in quarters or years, and is frequently gated by electrical equipment availability and grid connection rather than construction.

How can organizations reduce AI infrastructure costs without reducing capacity?

By addressing utilization before procurement. Scheduling and queueing across teams, right-sizing inference deployments, shutting down idle non-production environments, and selecting appropriately sized models typically recover more budget than renegotiating unit rates.

Key Takeaways

AI has changed data center infrastructure at the physical layer, and the industry has largely absorbed that. The lag is on the demand side, where enterprise buyers continue to apply procurement habits formed around a different class of workload.

The failures are consistent and avoidable. Capacity evaluated on floor area rather than rack density. Hardware selected before cooling capability is established. Training-class infrastructure purchased for what is fundamentally an inference requirement. Utilization treated as an operational detail rather than the dominant cost driver.

For providers, these are recurring conversations worth having earlier in the sales cycle. For enterprise buyers, the most useful preparation for 2026 is not deciding what to buy. It is establishing, with measurement rather than assumption, what the organization actually needs.

# # #

About the Author

Bijal Soni, Co-Founder of Briskstar, is a visionary leader shaping the future of software innovation. With a proven track record of building and leading impactful solutions across web, mobile, and cloud platforms, he blends technical depth with strategic foresight.

The post AI Data Center Infrastructure in 2026: What Enterprise Buyers Need to Prepare For appeared first on Data Center POST.

TL;DR

• Easing AI & Data Center Power Bottlenecks: The technology directly targets growing energy and cooling constraints in AI data centers and mission-critical facilities, allowing operators to expand capacity without adding costly electrical infrastructure.
• Built-In Harmonic Mitigation: By embedding Active Front End harmonic reduction directly into the motor system, Infinitum reduces electrical noise below IEEE 519 standards, lowering thermal load, saving floor space, and boosting reliability.
• Product Momentum & Event Presence: The draft highlights recent product expansions in integrated EC fans and previews Infinitum’s participation at upcoming industry conferences including Yotta, OCP Global Summit, and DCD Connect.

# # #

Data Center POST had the opportunity to connect with Ben Schuler, CEO & Founder of Infinitum. Prior to founding Infinitum, Ben spent the previous 15 years leading the design and deployment of complex industrial automation equipment across Taiwan, Japan and the United States. He was involved in the production ramp of Corning’s Gorilla Glass, Bloom Energy’s solid oxide fuel cell and Solyndra’s thin film solar module.

Infinitum is the direct beneficiary of Ben’s expertise, which helps him lead the mission of disrupting the way industrial machinery and electric motors consume and create electricity. As one of the original creators of Infinitum’s air-core motor system, Ben holds 36 patents related to electric machines and drives.

What does Infinitum do?

Infinitum develops next-generation motor and fan systems that improve power quality and maximize electrical capacity by reducing harmonics at the source, enabling power-constrained facilities to support more critical loads without expanding electrical infrastructure.

What problems does Infinitum solve in the market?

Traditional motors are often heavy, inefficient at variable loads and difficult to service at the end of their useful life. Infinitum is advancing the century-old motor industry with a new design that prioritizes efficiency, reliability, serviceability and responsible material use. This approach delivers meaningful improvements in energy consumption, operational performance and lifecycle impact.

What are Infinitum’s core products or services?

Infinitum’s core products are our Aircore EC+, Aircore EC+ Fans, and Aircore EC+ Pumps. You can learn more about these products on our website, goinfinitum.com/aircoreecplus.

What markets do you serve?

We serve a variety of markets, including data centers and mission-critical facilities, commercial buildings and HVAC, industrial facilities, healthcare, universities and schools, and agriculture.

What challenges does the global digital infrastructure industry face today?

Infinitum’s roots are in sustainability, and its technology is increasingly addressing one of the most urgent challenges facing digital infrastructure: meeting growing power and cooling demands as AI adoption accelerates and available electrical capacity becomes a business constraint.

Organizations are under intense pressure to balance growth, uptime and energy efficiency without adding costly electrical infrastructure or increasing strain on the grid. To help solve this challenge, Infinitum developed the Aircore EC+ motor system, enabling customers to do more with the power they already have.

How is Infinitum adapting to these challenges?

Rather than relying on conventional approaches that use bulky, heat-generating harmonic filters, Infinitum’s engineering team designed unique embedded harmonic mitigation directly into the motor system. This innovation improves power quality while recovering valuable floor space, reducing thermal load and lowering energy losses. By maximizing usable electrical capacity and maintaining high efficiency across the load curve, Aircore EC+ helps facilities increase performance while consuming less energy and requiring less infrastructure.

Equally challenging was proving that sustainability and operational performance can improve together. Infinitum conducted extensive testing, validation and real-world deployments to demonstrate that customers could reduce energy consumption, optimize available power and maintain reliability simultaneously.

The company successfully deployed this technology in some of the world’s most demanding environments, including AI data centers, hospitals and other mission-critical facilities, where performance standards are exceptionally high and downtime is not an option.

What are your company’s key differentiators?

Traditional variable frequency drives (VFDs) are essential for improving motor efficiency, but they produce harmonics that can degrade electrical performance, reduce usable power and threaten uptime. As power demands from AI and electrification rise, harmonics can have significant operational and financial consequences in demanding environments where energy efficiency, uptime and electrical performance are critical.

Aircore EC+ enables customers to do more with less by maximizing available power, electrical performance and equipment life while reducing energy use, footprint and system complexity.

Infinitum’s Aircore EC+ is uniquely designed with integrated Active Front End technology within the motor’s VFD to reduce harmonics at the source and below IEEE 519 specifications, delivering cleaner power without the added cost, footprint, and power losses of conventional passive and active filters. Analysts have commented:

”The Infinitum design delivers a near-sinusoidal waveform with harmonic distortion reduced to well below the IEEE 519 standard, far outperforming traditional passive filtering systems that typically only reduce harmonics to 10%,“ said Giselle Canahuati, Senior Consultant, Frost & Sullivan.

Available and application-specific features include:

• Standard IP65 Enclosure: Heavy-duty exterior protects against dust, dirt and water, allowing motors to operate reliably in harsh outdoor or wet environments.
• Standard High Resistance Ground: Delivers normal operation during an electrical ground fault and reduces the chance of equipment failure. Power-loss ride-through capability enables systems to continue operating during standby power.
• Standard Hybrid Ceramic Bearings: As compared to standard bearings, hybrid ceramic bearings do not conduct electricity and resist corrosion, enhancing durability, reliability and efficiency in high-performance applications.

What can we expect to see/hear from Infinitum in the future?

As customers face power budget constraints related to AI, electrification and higher electricity costs, our product development will continue to focus on helping customers do more with less. That translates into products that require fewer motors, fewer kilowatts and less electrical infrastructure. We’ve continued to innovate and expand our product portfolio for specific applications that help the most energy-intense sectors, including heavy industry, commercial buildings and data centers, to optimize how power is used and conserved. Most recently we recognized customer challenges in the EC fan market, and unveiled a high-efficiency, integrated (fan, motor, variable frequency drive) solution for HVAC, industrial and mission critical applications. We are expanding our ability to deliver more fully integrated systems. With Aircore EC+ Fans, Infinitum optimizes the fan, motor and variable frequency drive together to address real-world power, performance and integration challenges.

What upcoming industry events will you be attending?

Infinitum will be attending a variety of data center industry events, including:

• Yotta
• OCP Global Summit
• DCD Connect Virginia
• 7×24 Exchange Fall Conference

For those interested in connecting during the conferences to learn more about Infinitum’s motor and fan systems for mission critical facilities, please reach out to Anthony Lou at alou@goinfinitum.com or John Garza at jgarza@goinfinitum.com.

Do you have any recent news you would like us to highlight?

• Infinitum Introduces Aircore EC+ with Premium Features to Maximize Available Power Capacity
• Infinitum Unveils New EC Fan Systems
• Infinitum’s Aircore Motor System Earns 2026 E+E Leader Award

Is there anything else you would like our readers to know about Infinitum and capabilities?

AI-driven data center growth, electrification and an aging grid are making power availability a defining constraint. Because motors and fans are embedded throughout cooling infrastructure, improving their efficiency and electrical performance can have a meaningful impact on available power, operating costs and facility capacity. Infinitum helps customers do more with less by delivering more cooling and performance with less energy, material, space and infrastructure.

Where can our readers learn more about Infinitum?

Readers can learn more about us on our website, goinfinitum.com. You can also contact us with any questions using the contact form on our website, found here.

# # #

About Data Center POST

Data Center POST provides a comprehensive view of the digital infrastructure landscape, delivering industry insights into the global data center ecosystem. As the industry’s only peer-contributed and online publication, we offer relevant information from developers, managers, providers, investors, and trendsetters worldwide.

Data Center POST works hard to get the most current information and thought-provoking ideas most apt to add relevance to the success of the data center industry. Stay informed, visit www.datacenterpost.com.

If you are interested in contributing to Data Center POST, contact us at contributions@datacenterpost.com or submit your article here.

Want more digital infrastructure news? Stay in the know and subscribe to Data Center POST today!

The post Company Profile: How Infinitum is Overcoming Digital Infrastructure Power Constraints appeared first on Data Center POST.

TL; DR

• The standard model of grid-dependent power, diesel backup, and brief UPS support is increasingly inadequate for today’s volatile, AI-driven workloads.
• DCD Energy & Power Week 2026 will host a panel on maximizing battery asset utilization across data center facilities.
• The Oct. 1 session covers storage placement, layered storage strategies from grid-scale to rack-level, and chemistry and lifecycle tradeoffs.
• Panelists will also explore the strategic role of BESS in decarbonization through renewables capture and grid intermittency mitigation.

# # #

As AI-driven workloads push data centers toward higher densities and more volatile power demands, the standard model of grid-dependent power, diesel-fueled backup, and brief UPS support is increasingly inadequate for managing today’s environmental mandates and operational pressures. On Oct. 1, on its Energy & Sustainability Channel, DCD will host a broadcast titled “The data center battery revolution” from 9-10 a.m. EDT on its Energy & Sustainability Channel. The conversation will explore how Battery Energy Storage Systems (BESS) are transitioning from simple emergency reserves into core strategic assets that support both operational resilience and grid flexibility.

The streaming session brings together leaders from pioneering battery technology providers to deconstruct how operators can maximize battery asset utilization within data center facilities today. Topics include redefining the role and physical placement of energy storage within the modern mission-critical power train; the commercial use cases and strategic trade-offs for layered storage, from grid-scale assets down to rack-level storage; the density, lifecycle, supply chain, and chemistries of current storage options in the market; and the strategic utility of BESS in advancing decarbonization through the capture of surplus renewables and mitigation of grid intermittency.

Who:

• John Gagge, Vice President of Business Development, Network and Infrastructure Solutions, EnerSys
• Liz Wallace, Senior Product Manager, ZincFive
• Eric Hill, General Manager for Product and Marketing, CSB Energy Technology

What: “The data center battery revolution” panel discussion

When: Thursday, Oct. 1, 9-10 a.m. EDT

For operators, the core issue is practical: As workloads grow more volatile and environmental mandates become more stringent, how should BESS be positioned – physically and strategically – within the power train to deliver both resilience and flexibility? That’s where battery architecture becomes more than a backup decision and starts to shape overall infrastructure and decarbonization strategy.

Whether you’re designing new data center infrastructure, upgrading existing facilities, or evaluating future power strategies, this session offers actionable insight into one of the industry’s fastest-evolving challenges.

To register to stream this discussion, please visit: www.datacenterdynamics.com/en/dcd-broadcasts/energy–power-week-2026/the-data-center-battery-revolution-overcoming-myths-to-secure-future-growth

The post The Data Center Battery Revolution appeared first on Data Center POST.

TL;DR

• Extended Project Scope: A data center technology refresh does not end when new hardware goes live, as retired IT assets and facility infrastructure still require clear disposition plans. Leaving retired equipment in storage leads to lost resale value, wasted space, and tracking challenges.
• Inventory & Security Controls: Retired hardware must be cataloged by serial number and configuration, with data-bearing devices kept under a strict chain of custody for sanitization aligned with standards like NIST SP 800-88.
• Flexible Value Recovery: Organizations can recover significant operational and financial value by assessing assets for internal redeployment, secondary market resale, OEM trade-ins, or parts harvesting. Internal redeployment alone can save up to 85% compared to purchasing new equipment.

# # #

A data center technology refresh does not end when new servers, storage systems, or networking equipment goes live.

The retired equipment still needs a clear next step. Some assets may still have useful life and resale value. Others may contain sensitive data, reusable components, or materials that need to be routed for reuse, resale, recycling, or disposal

If these assets sit in storage, companies can lose resale value, miss internal reuse opportunities, take up valuable space, and make asset tracking more difficult.

In this blog, you will learn what happens to data center assets after a technology refresh, how retired equipment is evaluated, where it can go next, and why planning disposition early can help recover more value.

What Assets Are Retired During a Data Center Technology Refresh?

The assets retired during a data center technology refresh depend on the scope of the project. A server upgrade may remove only a limited set of IT equipment, while a larger refresh can affect storage, networking, power, cooling, and other supporting infrastructure.

The retired assets usually fall into two broad groups.

     1. Data-Bearing and IT Equipment

This group includes equipment that stores, processes, or transmits data, such as:

• Servers
• Storage arrays
• Hard drives and SSDs
• Network switches
• Routers
• Firewalls
• Backup appliances
• Other computing hardware
  

      2. Supporting Data Center Infrastructure

A refresh may also leave behind equipment that does not primarily store data but still has reuse, resale, or material value, including:

• Server racks and cabinets
• Power distribution units
• UPS equipment
• Batteries
• Cabling
• Cooling equipment
• Power and electrical components

What Happens to Data Center Assets After They Are Retired?

Before retired data center equipment is reused, sold, or recycled, it first needs to be identified, secured, and assessed. This creates a clear record of what was removed and helps determine the next step for each asset.

     1. Assets Are Identified and Tracked

Each asset should be recorded using details such as its asset tag, serial number, manufacturer, model, configuration, age, condition, location, data-bearing status, and ownership records.

A complete retirement inventory helps teams confirm what was removed, where it is located, and who is responsible for it. It also supports value recovery by showing which assets may still be suitable for reuse, resale, parts recovery, or another disposition route.

     2. Data-Bearing Assets Are Separated

Servers, hard drives, SSDs, storage systems, and other devices that may contain company data require controlled handling after removal.

These assets should remain under a documented chain of custody and be placed in secure storage until the required media sanitization or destruction process is completed. Teams should also keep records of data destruction and verify that sanitization was completed before the equipment moves to its next destination.

The National Institute of Standards and Technology (NIST) SP 800-88 Rev. 2 provides guidance for media sanitization, including how organizations should select sanitization methods based on the sensitivity of the information stored on the media.

     3. Assets Are Evaluated for Remaining Value

Retirement does not always mean an asset has reached the end of its useful life.

Teams should assess its working condition, remaining useful life, hardware generation, specifications, secondary-market demand, repair needs, testing costs, and removal or transportation costs.

This assessment helps determine if the asset should be redeployed, resold, used for parts, recycled, or disposed of.

Where Do Retired Data Center Assets Go?

Retired data center assets can follow different paths depending on their condition, remaining useful life, data security requirements, and market value.

Below are the main routes these assets can take after retirement.

     1. Internal Redeployment

Usable equipment does not always need to leave the organization. Servers, storage systems, networking equipment, and other hardware may be moved to another data center, a backup environment, a development or testing environment, another facility, or a lower-priority workload.

Internal redeployment can extend the useful life of equipment that no longer meets the requirements of its original environment. It can also reduce the need to purchase replacement hardware elsewhere in the business.

“Companies retire equipment from one data center even when another location still needs the same type of asset. When teams have visibility into available equipment across the business, internal redeployment can save more than 85% compared with buying new,” said Luke Crihfield, Senior Director of Revenue at Amplio, which manages surplus and retired equipment for enterprises through its surplus asset management services. “The key is identifying that internal demand early, before usable equipment moves into resale, recycling, or another disposition route.”

     2. Resale and Remarketing

Assets that no longer serve an internal need may still have value in the secondary market.

Servers, storage equipment, network hardware, memory, CPUs, racks, and power equipment may attract buyers based on their condition, age, hardware generation, and specifications.

Timing also matters. As newer generations become available, demand for older hardware can decline, which can reduce resale value.

     3. OEM Return or Trade-In

Some retired equipment may qualify for an original equipment manufacturer (OEM) takeback, trade-in credit, lease return, or vendor buyback option.

These routes can provide a structured way to remove equipment and recover some value. Companies should compare the expected return with internal reuse and secondary-market resale before choosing this route.

     5. Parts Harvesting

An asset with limited value as a complete unit may still contain components that can be reused or sold.

CPUs, RAM, power supplies, controllers, fans, network cards, and storage components may retain value even when the full server or system has little market demand. These parts can support existing equipment, serve as replacement inventory, or be sold through secondary markets.

Parts harvesting can recover value that would otherwise be lost if the entire asset were sent directly to recycling.

     6. Recycling and Final Disposal

Equipment with no practical reuse, resale, or parts recovery option can move into electronics recycling.

Recyclers can recover materials such as copper, aluminum, steel, plastics, and circuit boards from retired hardware and supporting infrastructure. This route captures material value from equipment that has reached the end of its practical service life.

Final disposal should come only after reuse, resale, parts recovery, and recycling options have been assessed.

Why ITAD Does Not Cover the Entire Refresh

IT asset disposition, or ITAD, typically focuses on electronic and data-bearing equipment such as servers, storage devices, drives, and networking hardware.

A broader data center refresh can leave behind other assets, including racks, UPS systems, batteries, cooling equipment, power infrastructure, electrical components, and other facility equipment.

Many of these assets sit outside the core scope of traditional ITAD, but they still need a clear disposition route. Depending on their condition and remaining value, they may need to be evaluated for reuse, resale, recycling, or disposal.

This is why a full data center refresh often requires a broader asset disposition and recovery process alongside ITAD.

Conclusion

A data center refresh does not end when the new technology goes live. Retired equipment still needs to be secured, assessed, and assigned the right next step.

The right disposition route depends on the asset’s remaining operational, financial, and material value. Reuse, redeployment, resale, and parts recovery should be considered before equipment moves to recycling or final disposal.

A planned disposition process helps protect data, recover more value, and ensure retired data center assets move to the most appropriate next use.

# # #

About the Author

Luke Crihfield is a senior director of revenue at Amplio, helping manufacturers turn surplus into opportunity through AI-driven growth.

The post What Happens to Data Center Assets After a Technology Refresh? appeared first on Data Center POST.

TL;DR

• Multiple fuel vendors do not necessarily create resiliency. Without one partner accountable for the full path from fuel source to generator, handoffs between vendors can create additional gaps, delays, and points of failure.
• Having fuel onsite is not the same as being fuel-ready. True readiness depends on fuel quality, inventory, storage and transfer equipment, emergency supply, and the ability to reliably deliver usable fuel to the generator when needed.
• End-to-end fuel lifecycle management strengthens accountability and operational readiness. A single partner coordinating fuel quality, monitoring, maintenance, delivery, and emergency response can reduce complexity and help ensure the entire system performs as intended.

# # #

When a facility’s backup power depends on diesel fuel, the biggest risk isn’t always a generator failure; it’s the gaps between vendors. Fuel procurement, storage, filtration, testing, emergency delivery, and equipment servicing often sit under separate contracts, each with its own priorities and response times. In a crisis, that fragmentation can mean delayed deliveries, unclear accountability, and ultimately, downtime.

Just as importantly, having fuel onsite is not the same as being fuel ready. A full tank does not guarantee that the fuel is usable, that the transfer system will perform as intended, or that additional supply can be delivered when an outage extends beyond initial runtime assumptions. True readiness depends on the entire fuel pathway working together.

The Hidden Costs of Vendor Fragmentation

Many organizations assume that using multiple specialized fuel vendors creates redundancy and reduces risk. In practice, multiple vendors do not necessarily equal resiliency. A fuel supplier may be responsible for delivery but not stored-fuel quality. A filtration provider may maintain fuel condition but have no visibility into inventory or transfer-system performance. A testing lab may identify a problem without having responsibility for corrective action or emergency logistics.

Without one party accountable for the complete path from fuel source to generator, each handoff can become an additional point of failure. The issue is not simply how many vendors are involved, but whether any one is responsible for ensuring that the entire system performs as it should.

When something goes wrong – say, water contamination is found during routine testing – each vendor can point to another part of the chain. The result is slower resolution, higher costs from emergency orders, and increased operational risk. Industry guides on supply chain resilience emphasize that clear accountability and coordinated procedures are essential for maintaining fuel readiness, especially during disruptions.

The alternative is a model where one partner manages  the entire fuel lifecycle, from terminal to engine.

What End-to-End Management Looks Like

End-to-end fuel lifecycle management means one provides accountability across the entire journey of fuel: from selection at the terminal, through storage and quality maintenance, to delivery and engine readiness. This model aligns incentives: The same team that designs the filtration system also monitors inventory, conducts testing, and stands ready to deliver emergency fuel.

Key elements include:

• Unified accountability: A single point of contact for all fuel-related issues, reducing finger-pointing and speeding up decisions.
• Integrated monitoring: Continuous tracking of fuel quality, inventory levels, and equipment performance to catch issues before they escalate.
• Coordinated emergency response: Pre-planned logistics that ensure fuel and equipment support arrive together when needed most.

Why It Matters Now

Recent supply chain disruptions have highlighted how fragile fragmented systems can be. Rising fuel costs, transportation bottlenecks, and stricter environmental regulations add pressure to get fuel management right. ESI Total Fuel Management approaches fuel as a complete operational system, connecting supply, stored-fuel quality, inventory, infrastructure, maintenance, emergency logistics, and generator readiness under one accountable model.

For mission-critical facilities like data centers, hospitals, and emergency operations centers, the stakes are especially high. A generator that fails to start due to fuel contamination or delayed delivery can have cascading effects far beyond the facility itself.

Practical Takeaways for Facility Leaders

• Map your fuel lifecycle: Industry resilience guides, including FEMA’s supply chain guidance, recommend mapping every step from procurement to engine readiness to identify accountability gaps.
• Consolidate where possible: Consider whether a single partner can manage more of the lifecycle to reduce coordination overhead.
• Test for readiness, not just compliance: Routine testing should validate that fuel will perform under real emergency conditions, not just meet baseline standards.
• Plan for the worst-case scenario: Ensure your fuel partner can deliver both fuel and equipment support simultaneously during an outage.

A Clearer Path to Fuel Readiness

The goal of fuel management isn’t just to keep tanks full; it’s to ensure that backup power systems start reliably when needed most. End-to-end lifecycle management  reduces complexity, clarifies accountability, and aligns every part of the fuel system toward that goal. For organizations that depend on uninterrupted power, having one partner responsible for the entire fuel lifecycle isn’t just convenient – it’s a strategic advantage.

To discuss what a dedicated fuel partner can mean to your operation, contact ESI Total Fuel Management today.

The post End-to-End Fuel Lifecycle Management: Why One Partner Beats Multiple Vendors appeared first on Data Center POST.

TL;DR

• The global data center immersion cooling market is projected to expand from USD 1.7 billion in 2025 to USD 10.9 billion by 2035 at a 19.8% CAGR, propelled by accelerating artificial intelligence and machine learning workloads, rising rack power densities, and mounting energy efficiency pressures.
• Immersion cooling delivers significantly higher compute density per square foot compared to conventional air-cooled systems, allowing colocation and urban data center operators to maximize existing infrastructure without major capital expansion.
•  The solution segment led the component market with a 72% share (USD 1.2 billion) in 2025, while hyperscale operators held a 29.8% share by end user as they prioritize total cost of ownership, operational efficiency, and long-term sustainability.

# # #

The global data center immersion cooling market was valued at USD 1.7 billion in 2025 and is estimated to grow at a CAGR of 19.8% to reach USD 10.9 billion by 2035, according to the recent report by Global Market Insights Inc.

Market growth is entering a high-growth phase, driven by accelerating digital transformation and next-generation computing requirements. Market expansion is fueled by the rapid surge in artificial intelligence and machine learning workloads, increasing rack power densities, and mounting pressure to improve energy efficiency across facilities. As computing intensity continues to escalate, traditional cooling approaches are becoming less viable, accelerating the shift toward immersion-based systems. Rising electricity costs and heightened environmental accountability are further reinforcing the economic case for more efficient thermal management technologies.

Limited floor space in colocation environments and urban data facilities is also encouraging operators to adopt high-density solutions. Immersion cooling enables significantly higher compute density per square foot compared to conventional air-cooled systems, allowing facility operators to maximize existing infrastructure without major capital expansion. In single-phase immersion cooling architectures, dielectric fluid remains in a liquid state throughout operation, while pumps and external heat exchangers effectively dissipate accumulated thermal loads.

By component, the data center immersion cooling market from solution segment held a 72% share, generating USD 1.2 billion in 2025. This segment encompasses integrated immersion cooling systems, including immersion tanks, coolant management frameworks, thermal exchange units, monitoring platforms, and supporting integration hardware. Vendors are increasingly delivering comprehensive, ready-to-deploy systems designed for seamless compatibility with established data center environments. Modern solutions incorporate intelligent temperature tracking, automated coolant regulation, and connectivity with facility management systems to enhance operational performance and overall energy optimization.

In terms of end users, the hyperscale segment held 29.8% share in 2025 and is forecast to reach USD 3.5 billion by 2035. Large-scale data center operators are heavily investing in advanced computing infrastructure to support high-performance processing demands. These facilities operate extensive server environments within purpose-built spaces, prioritizing efficiency, scalability, and long-term infrastructure resilience. Hyperscale operators place strong emphasis on total cost of ownership, carefully balancing capital expenditures, operating efficiency, and long-term sustainability objectives when selecting cooling technologies.

U.S. Data Center Immersion Cooling Market generated USD 454.2 million in 2025 and is anticipated to grow at a CAGR of 16.9% from 2026 to 2035. Increasing demand for advanced cooling technologies is driven by the expansion of large cloud service providers and major technology enterprises headquartered in the country. As digital infrastructure footprints expand, organizations are prioritizing cost-effective and energy-efficient thermal management systems to maintain operational competitiveness. Strong regional commitments to sustainability and improved power utilization effectiveness are accelerating the adoption of immersion cooling solutions across the U.S. market.

Key participants operating in the global data center immersion cooling market include Vertiv, Dell Technologies, Fujitsu, Submer, Green Revolution Cooling, LiquidCool Solutions, Asperitas, Bitfury, DCX Liquid Cooling Company, and Midas Immersion Cooling. Companies in the market are reinforcing their competitive positioning through continuous product innovation, strategic collaborations, and geographic expansion. Industry players are investing in research and development to enhance system reliability, thermal efficiency, and scalability to meet evolving high-density computing requirements. Many firms are introducing modular and turnkey solutions to simplify deployment and accelerate time to market. Strategic alliances with data center operators, colocation providers, and hyperscale customers are strengthening long-term supply agreements and expanding global reach. Businesses are also focusing on interoperability with existing infrastructure to reduce adoption barriers.

The post Data Center Immersion Cooling Market to hit USD 10.9 billion by 2035 appeared first on Data Center POST.

TL;DR

• As high-density GPU clusters and liquid cooling reduce traditional operating margins, procedural discipline and human error are becoming increasingly important factors in AI infrastructure resilience.
• Operational Complexity of Liquid Cooling: Direct-to-chip and immersion liquid cooling solve thermal barriers but introduce complex operational demands, such as coolant chemistry, leak detection, and CDU coordination, where minor procedural mistakes carry immediate, expensive consequences.
• Focus on Long-Term Operational Maturity: AI infrastructure, especially in emerging markets like Central Asia, will ultimately be judged by long-term operational uptime and disciplined management rather than initial construction speed

# # #

As GPU clusters get denser and liquid cooling becomes standard for high-end AI, the decisive variable in AI infrastructure is shifting from how a facility is built to how it is run. Operational standards, not just megawatts, are becoming a critical measure of readiness.

Every conversation about AI infrastructure opens with the same numbers: megawatts secured, racks pushing past 100 kW, GPU counts, gigawatt campuses, interconnect queues. Those numbers matter without power and somewhere to put the heat, nothing scales. But they describe what a facility can theoretically hold, not whether it can carry that load safely at 3 a.m., when a coolant distribution unit alarms and the on-call technician has minutes to decide whether to fail over, shed load, or ride it out. That decision isn’t made by a GPU or a chiller. It’s made by a person following or failing to follow a procedure. And that is where AI data centers are now most exposed.

AI changes the operating model, not just the hardware

Traditional facilities were forgiving: predictable workloads, mature air cooling, settled procedures. AI breaks that. OCP describes AI-driven infrastructure as an order-of-magnitude increase in rack power density, with roadmaps pointing towards 1 MW racks in the next few years. Power volatility adds another challenge: NVIDIA has described rack-level AI workloads swinging from around 30% to 100% utilization and back again in milliseconds, reflecting the synchronized compute and communication phases documented in large-scale AI training research. Loss of coolant flow can quickly translate into thermal throttling, instability or protective shutdowns. At these densities, the coupling between IT load, power and cooling becomes tight and unforgiving. The hard problem is no longer only construction; it’s operational maturity, and the gap between the two is where expensive clusters get lost.

Liquid cooling solves a thermal problem and creates an operational one

Nowhere is this clearer than with liquid. Direct-to-chip liquid cooling is becoming increasingly standard for high-density AI, while immersion remains one of several approaches, but moving water or dielectric fluid through a live data hall introduces operational disciplines that may be unfamiliar to teams accustomed to air-cooled environments: coolant chemistry and conductivity, biological growth and corrosion in the cooling loop, filtration, leak detection and containment, and constant coordination between the facility water system and the rack-level CDU. The OCP Project Deschutes CDU specification targets roughly 2 MW of cooling capacity at 500 GPM with 80 to 90 psi of available pressure, with defined coolant and system requirements for high-density liquid-cooled deployments. At this scale, liquid cooling becomes an operational system in its own right, with consequences for maintenance, monitoring and response procedures. It does not remove the operational problem. It raises the stakes on it.

Human error is now an infrastructure risk

Uptime Institute’s 2025 outage analysis found that 85% of major human-error outages involved either staff failing to follow procedures or flaws in the procedures themselves, and that nearly 40% of organizations had suffered a major outage caused by human error over the previous three years. Its 2026 analysis shows that failure to follow established procedures remains the leading driver of human-error-related outages. Power is still the leading cause of impactful outages overall, so the point is not that procedures replace infrastructure risk. It is that procedures remain one of the clearest controllable weaknesses once the facility is live: whether the runbook exists, whether it is correct, whether anyone follows it, and whether change management catches the conflict before it reaches the floor. In lower-density environments, operational mistakes often had more thermal and electrical margin around them. At 100 kW per rack and above, particularly in liquid-cooled environments, those margins can shrink sharply. The same procedural error can therefore have much faster and more expensive consequences.

What “standards-based operations” actually means

The phrase has to stop being a slogan. It means a facility runs on documented, auditable practice rather than tribal knowledge and real frameworks exist for this. The operations layer has its own independent assessment framework: Uptime Institute’s Management & Operations (M&O) Stamp of Approval evaluates staffing, organization, training, maintenance, operating conditions, planning and coordination independently of the Tier Classification System. A highly resilient design can still be undermined by weak operations. Around it sits the management-system scaffolding the best operators already use: ISO/IEC TS 22237-7:2018, which specifically addresses data centre management and operational processes, ANSI/TIA-942-C for physical infrastructure, and ITIL-style change and incident management that turns change control from a form into a real risk gate. In practice that means clear methods of procedure and emergency operating procedures; change management that accounts for how one modification now ripples across load, cooling, power, and network at once; staff trained for liquid and high density rather than retrofitted from air; telemetry that correlates IT load with thermal and power behavior in real time; and risk management that prices in operational error.

The industry is already moving this way

The clearest signal is OCP’s Open Data Center for AI initiative, launched in October 2025 alongside an industry call for collaboration initiated by Google, Meta and Microsoft and expanded through new workstreams and specifications in 2026. It is explicitly about standardizing not only power and cooling hardware but management telemetry and operational technology and it names the operational problem directly: AI facilities now involve several parties at once (facility owner, hardware owner, workload owner), making service-level agreements and coordination far more complex than the clean colocation SLAs of the air-cooled era. When the largest operators on the planet start writing shared operating specifications, operations not just construction has become a competitive frontier.

Emerging markets: the sharpest version of the problem

This is where I watch it play out most acutely across Central Asia, where AI capacity, power supply, an engineering labor market, and an operating culture are all being built at once. The risk here isn’t only a shortage of megawatts or GPUs; it’s a maturity gap in procedures, people, and change management that never appears on a capex spreadsheet and surfaces only after the first avoidable incident. But the asymmetry cuts the other way too: a greenfield operator carrying no legacy habits can adopt M&O-grade procedures, real change management, and proper telemetry from day one, instead of bolting them on after a decade of “this is how we’ve always done it.” Framed correctly, standards-based operations aren’t bureaucracy that slows an emerging market down. They are the fastest route to closing the reliability gap with mature regions and to skipping mistakes the industry has already paid for.

The real test comes after the ribbon-cutting

AI data centers are becoming national digital infrastructure, and they will be judged not on how fast they broke ground but on whether they run. Power makes deployment possible, cooling keeps density stable, GPUs do the compute but the facility that earns trust over years is the one with correct procedures, trained people, disciplined telemetry, real change control, and risk management that treats human error as seriously as equipment failure. The strongest AI data centers won’t be the ones built first. They’ll be the ones still running predictably while everyone else is firefighting.

# # #

About the Author

Maksim Gavriliuk is an independent digital infrastructure and data center expert, IEEE Senior Member, and DCOS 2026 Review Committee Member. He specializes in data center operations and AI-ready infrastructure, with over 20 years of experience in IT infrastructure projects across Central Asia and Eastern Europe.

The post AI Data Centers Don’t Fail at the Rack. They Fail in the Runbook. appeared first on Data Center POST.

TL;DR

• Floating data center site selection is a system design decision rather than a simple real estate exercise, requiring location, engineering, and regulation to be managed in parallel to solve land, power, and cooling constraints.
• Developers must evaluate six core factors: energy availability and scalability, marine environment metocean conditions, operational logistics, digital connectivity redundancy, regulatory permitting pathways, and physical/cyber security.
• Long-term viability requires managing key operational trade-offs over multi-decade lifespans, balancing near-shore accessibility against offshore expansion potential and grid reliance versus dedicated offshore generation.
• Early integration of marine risk assessment alongside classification societies like ABS avoids costly late-stage redesigns and provides clarity across complex regulatory jurisdictions

# # #

When it comes to floating data centers, it really is about ‘location, location, location’. Site selection influences engineering, regulation, operations and long-term resilience, which is why it pays to get this right from the outset, writes Frank Genco, ABS Vice President of Strategic Account Program 

When it comes to floating data centers (FDC), the question isn’t why offshore. It is increasingly where. The whys are clear: the rapid growth of data centers is already testing grid capacity, water resources and the tolerance of local communities living alongside these facilities. By siting data centers offshore, however, many of those constraints can be resolved, with developers able to access new energy sources, abundant seawater for cooling and locations unconstrained by land availability.

But where? That’s a bigger question, and the answer will influence the regulatory pathway, engineering design, reliability, cost, maintenance, security and uptime for the next 30 years.

More than a real estate decision

This isn’t a site selection checkbox exercise, however. It’s a system design decision, because location, engineering and regulation are closely interconnected. A change to one often affects the other two. A site that appears attractive from an operational perspective may introduce regulatory complexity. Engineering innovations may suddenly make previously unsuitable locations viable. These three strands (location, regulation and engineering design) need to be managed in parallel because it is at their intersection where the right solution in the right location will be found.

 ABS brings decades of experience supporting complex marine and offshore assets, from floating LNG and ultra-deepwater oil production units to floating wind. That experience can help engineers understand how location choices influence requirements and regulatory pathways long before detailed design begins.

Every answer creates another question

Site selection involves identifying and testing a wide range of competing and interdependent issues all at the same time. An apparently straightforward question about available infrastructure, such as natural gas pipeline proximity, may lead to broader engineering considerations. Does a location provide sufficient access to electrical power today for a standalone FDC, and can it support future expansion or augmentation? Would dedicated offshore generation offer greater resilience and public approval than relying solely on the grid? Could nearby renewable energy or future small modular nuclear reactors reshape the long-term energy strategy?

It’s the same for the cooling requirements of the computing and power generation facilities. Engineers need to understand seasonal temperature variation, water movement, corrosion rates and whether engineering trade-offs could expand the range of viable operating conditions. Improving cooling system design may significantly increase the number of potential locations but how would this innovation impact timelines?

Regulation presents another layer of complexity. The applicable framework depends on numerous factors, including proximity to shore, whether the facility is permanently moored or mobile, the pipeline requirements, the presence of subsea power or communications infrastructure and which authorities have jurisdiction. Depending on the location, developers may need to engage with federal, state, local and port authorities, alongside maritime regulators and environmental agencies.

Understanding these interactions early can avoid costly redesigns later in the project.

The major considerations

Every project will have unique requirements, but several key factors will consistently shape site selection. These include:

1. Energy. Developers must determine where power will come from, whether existing grid infrastructure is sufficient, or whether dedicated offshore generation offers greater resilience and flexibility. Equally important is understanding how future demand growth can be accommodated without major redesign.
2. Marine environment. Long-term metocean conditions, extreme weather exposure, wave climate, currents, corrosion potential and marine growth all influence structural design, maintenance schedules and operating costs.
3. Operational logistics. How easily can personnel access the facility? Which nearby ports can support maintenance activities? How resilient are local supply chains? These questions directly affect availability and lifetime operating costs.
4. Digital connectivity. Fiber availability, cable redundancy and network latency all influence performance. Building resilience into communications infrastructure from the outset can significantly reduce operational risk over the facility’s lifetime.
5. Regulatory landscape. Understanding permitting pathways, environmental approval processes, certification requirements and jurisdictional responsibilities early in the project can help avoid delays and provide greater certainty during development. ABS works with developers at the regulatory interface, helping to map the applicable frameworks and identify the right agencies before those conversations become constraints on engineering or schedule.
6. Security. Physical protection, cybersecurity, vessel traffic management, supply chain resilience and geopolitical risk all become increasingly important as critical digital infrastructure moves offshore.

Thinking beyond first deployment

A floating data center will likely have an operational lifespan measured in decades, making resilience and future-proofing central to site selection. The maritime and offshore industries are used to thinking long-term and building offshore structures designed to withstand rare and extreme events, but these floating data factories will be packed with cutting-edge computing technology and serving markets that are evolving at speed, which may require additional forethought. Developers should consider whether today’s optimal location will still be viable if computing demand doubles, cooling technologies evolve or additional modules are needed.

Managing trade-offs

There is rarely a perfect location. Sheltered near-shore waters may simplify operations initially but limit future expansion. Offshore sites may provide greater access to renewable energy but introduce more demanding environmental conditions that require additional engineering and impact costs. Grid-connected facilities may reduce upfront investment but dedicated offshore generation could offer greater resilience and scalability.

These trade-offs are not unique to floating data centers. Offshore developers are used to navigating competing priorities and understand that the key is to evaluate these trade-offs systematically and simultaneously so that technical, commercial and regulatory implications are considered together rather than in isolation. In this way, a holistic, multi-disciplinary and multi-agency approach can be taken that minimizes the risks of unwelcome surprises further down the road.

Involving marine risk assessment from the beginning

Ultimately, site selection is a risk management exercise. The maritime and offshore industries have spent decades developing sophisticated approaches to evaluating operational, environmental and engineering risk. Many of the challenges floating data centers will face have already been addressed across offshore energy, shipping and marine infrastructure. Applying these proven methodologies during the earliest stages of development enables better-informed decisions before major investments are committed. ABS works across all three of these dimensions simultaneously, combining marine risk assessment, regulatory engagement and engineering insight to help developers match the right asset to the right location.

As floating data centers move from concept to commercial reality, the most successful will be those that view site selection as integral to engineering and commercial outcomes, not simply as a location for the “real” work to take place. By working with ABS to integrate engineering, regulation and marine risk assessment from the outset, developers can identify locations that work today and are also resilient, scalable and commercially viable for decades to come.

 # # #

About the Author

Frank Genco is the Vice President of Strategic Accounts at the American Bureau of Shipping (ABS). He created and heads the Strategic Account Management (SAM) Program. His division is tasked with aligning ABS’s global enterprise resources across maritime, energy, digital infrastructure, cyber security, and risk management to service their highest-value global clients. Frank specializes in critical infrastructure, strategic corporate partnerships, and connecting top-tier technology companies with maritime and digital risk expertise. He is currently building out specialized teams, including expanding executive relationships with Silicon Valley technology firms to bridge the gap between AI/digital infrastructure and maritime compliance.

The post Site Selection: The Most Important Decision You’ll Make Before Construction Begins appeared first on Data Center POST.

TL;DR

• Hose selection must be driven by complete system operating conditions (including specific coolant chemistry, thermal cycling, working pressure, permeation, and cleanliness requirements) rather than choosing a material name directly from a catalog page.
• Hose material performance impacts long-term reliability, influencing key operational factors beyond flexibility, such as pressure drops, gradual fluid permeation, and loop contamination.
• Common hose materials involve distinct design trade-offs, with options like EPDM, silicone, PTFE, PFA, and metal offering different balances of chemical compatibility, flexibility, mechanical strength, cost, and permeation resistance.
• Hoses and fittings should be integrated early into overall rack, manifold, and cooling system architecture rather than left as an afterthought during final installation

# # #

By Xiang Wu, IT & Automation Director, CJan Fluid Technology Co., Ltd.

As liquid cooling moves from specialized deployments into mainstream AI and high-performance computing infrastructure, more attention is being paid to pumps, cold plates, cooling distribution units and control systems.

The hose is easier to overlook.

Yet in a direct-to-chip cooling loop, the hose is part of the fluid path that operates continuously while carrying coolant between components that may need to be serviced independently. Its material affects more than flexibility. It can influence pressure capability, chemical compatibility, permeation, cleanliness, bending behavior and service life.

For that reason, choosing a liquid cooling hose should start with the operating conditions of the cooling loop, not with a material name on a catalog page.

Start With the Coolant, Not the Hose

The first question in hose selection should be simple:

What fluid will actually be inside the hose?

Water, water-glycol mixtures and specially formulated cooling fluids do not place exactly the same demands on hose materials. Even when two coolants appear similar, differences in additives, inhibitors, concentration and operating temperature can affect material performance.

This is particularly important for systems that are expected to operate continuously for years.

A material that appears compatible during a short laboratory test may behave differently after thousands of hours of exposure. Swelling, hardening, extraction of low-molecular-weight components and changes in mechanical properties may develop gradually.

For this reason, coolant compatibility should be evaluated using the actual fluid formulation whenever possible.

Material selection should consider:

• Coolant chemistry
• Glycol concentration, where applicable
• Corrosion inhibitors
• Biocides and other additives
• Operating temperature
• Expected service life
• Cleanliness requirements

The hose should be qualified as part of the actual cooling system rather than treated as an isolated component.

Temperature Is More Than a Maximum Rating

Hose specifications often list a maximum temperature. That number is useful, but it does not tell the whole story.

A cooling loop may experience a relatively stable operating temperature most of the time while still going through repeated thermal cycles. Startup, shutdown, maintenance and changes in computing load can all introduce temperature changes.

Repeated thermal cycling can affect:

• Elastomer flexibility
• Reinforcement
• Fitting connections
• Sealing performance
• Long-term material stability

A hose material therefore needs to be evaluated against the complete temperature profile rather than a single maximum value.

For example, EPDM is widely used in water-based cooling applications because of its resistance to water and glycol-based fluids. Silicone offers excellent flexibility and broad temperature capability in many applications. Fluoropolymer materials such as PTFE and PFA can be considered when chemical resistance, cleanliness or low extractables are particularly important.

The point is not that one material is universally better than another.

The right material depends on what the cooling loop requires.

Pressure and Flow Still Matter

Liquid cooling systems are designed around controlled flow, but that does not mean hose pressure requirements can be ignored.

The hose should be evaluated against:

• Normal operating pressure
• Maximum working pressure
• Pressure fluctuations
• Pump characteristics
• Startup and shutdown conditions
• Potential pressure spikes

Internal diameter also matters.

A hose that is too small can increase pressure drop and require additional pump head. A hose that is unnecessarily large may create packaging and routing problems.

The engineering objective is therefore to balance flow requirements with pressure capability, flexibility and available space.

The hose should also be evaluated as a complete assembly. The pressure rating of the tube alone does not necessarily represent the pressure capability of the hose, fitting and connection together.

Flexibility Has a Practical Value

One reason flexible hose continues to have a role in liquid-cooled data centers is serviceability.

Servers, cold plates, manifolds and rack-level cooling components may need to be removed or repositioned during maintenance. A rigid connection can make this difficult.

But flexibility should not be confused with unlimited bending.

Every hose has a minimum bend radius. Repeatedly forcing a hose tighter than its recommended radius can place unnecessary stress on the tube, reinforcement and fittings.

Good routing should avoid:

• Sharp bends near fittings
• Twisting along the hose axis
• Continuous tensile loading
• Contact with sharp edges
• Uncontrolled movement caused by vibration

A slightly longer hose with proper routing can sometimes be more reliable than a shorter hose installed under constant mechanical stress.

This is one reason hose routing should be considered during rack and manifold design rather than left to the final installation stage.

Permeation Deserves More Attention

Leakage is an obvious concern in a liquid cooling system. Permeation is less visible.

Some hose materials allow molecules to gradually pass through the hose wall. The rate depends on the material, fluid, temperature, pressure and wall construction.

For a short-duration application, this may have little practical significance. For a cooling loop expected to operate continuously over a long period, it can become part of the system reliability calculation.

Potential consequences include:

• Gradual coolant loss
• Changes in coolant concentration
• Moisture or gas transfer
• Changes in fluid condition
• Increased maintenance requirements

This does not mean that a low-permeation material is automatically the correct choice.

It means that permeation should be considered whenever coolant stability and long service intervals are important design objectives.

Cleanliness Is a System Requirement

Data center liquid cooling is not simply a plumbing application.

The coolant passes through components such as cold plates, heat exchangers, pumps and manifolds. Contamination introduced by one component can potentially travel throughout the loop.

For systems with tighter cleanliness requirements, engineers may need to consider:

• Extractables
• Particles
• Ionic contamination
• Internal surface condition
• Manufacturing cleanliness
• Cleaning and flushing procedures

This is one area where material selection and manufacturing quality are closely connected.

A material with appropriate chemical resistance may still be unsuitable if the finished hose assembly introduces unacceptable contaminants into the cooling loop.

The Fitting Is Part of the Decision

Selecting the hose and selecting the fitting should not be treated as two independent decisions.

The connection needs to accommodate the hose construction, pressure, temperature, installation method and expected movement.

Particular attention should be paid to the area immediately behind the fitting.

This is often where bending, vibration and installation forces concentrate. If the hose is forced into a sharp bend immediately after the connection, the fitting may effectively become the mechanical anchor point for the entire assembly.

A good installation distributes mechanical loads instead of transferring them directly to the connection.

A Practical Material Selection Approach

There is no single hose material that should be specified for every liquid cooling system.

A practical selection process can begin with six questions:

1. What is the coolant?
   Identify the actual fluid and additives.

2. What are the temperature conditions?
   Consider both steady-state temperature and thermal cycling.

3. What pressure and flow are required?
   Check working pressure, pressure fluctuations, internal diameter and pressure drop.

4. How much movement is expected?
   Determine whether the hose is static, occasionally moved or continuously flexed.

5. What cleanliness and permeation requirements apply?
   These factors may influence the choice between elastomeric and fluoropolymer constructions.

6. How will the hose be installed and maintained?
   Consider bend radius, fitting accessibility, replacement and service clearance.

Only after these questions are answered should the material be selected.

A Simple Comparison

This table should be treated as a starting point rather than a specification guide. Actual material performance depends on the formulation, hose construction, manufacturing process and operating environment.

The Best Hose Is the One That Fits the Whole System

Liquid cooling projects sometimes approach hose selection from opposite directions.

One team may prioritize flexibility.

Another may prioritize pressure.

A third may focus on chemical compatibility or cleanliness.

All of these considerations are legitimate. The challenge is that optimizing one characteristic without considering the others can create a different reliability problem.

A highly flexible hose may not provide the desired pressure or permeation performance. A chemically resistant material may require a different reinforcement or connection method. A mechanically robust assembly may be harder to route in a dense rack environment.

The selection process therefore needs to look at the complete operating envelope.

For high-density AI infrastructure, this becomes increasingly important as liquid cooling systems become more complex and maintenance access becomes more valuable.

Design the Hose Into the Cooling System

The hose should not be the last component selected after the rest of the cooling architecture has already been fixed.

It should be considered alongside the cold plate, manifold, CDU, pump, quick-disconnects and rack layout.

When material, construction, routing and connections are evaluated together, engineers can make better trade-offs between reliability, serviceability and cost.

The objective is not simply to find a hose that can carry coolant.

The objective is to build a fluid path that remains predictable throughout the operating life of the data center.

As direct-to-chip cooling continues to expand, that distinction will become increasingly important.

# # #