TL;DR

• The global hyperscale data center market is projected to skyrocket from USD 80.9 billion in 2025 to USD 624.2 billion by 2035, expanding at a compound annual growth rate (CAGR) of 22.2%.
• This explosive growth is heavily driven by AI-focused architectures, high-performance computing (HPC), and public cloud migrations, which require massive deployment of GPU-intensive server environments with higher rack power densities.
• Hardware accounted for 93.9% (USD 76 billion) of the market in 2025. While air-based cooling held a 52.8% share, direct-to-chip (DTC) liquid cooling is fast becoming the preferred thermal management method to handle intense AI workloads without requiring full infrastructure rebuilds.

 # # #

The global hyperscale data center market was valued at USD 80.9 billion in 2025 and is estimated to grow at a CAGR of 22.2% to reach USD 624.2 billion by 2035, according to the recent report by Global Market Insights Inc.

The market expansion is shaped by major changes in computing infrastructure, including the shift toward AI-focused architectures, high-performance computing environments, and increased dependence on cloud-based enterprise platforms. The growing need for advanced processing capacity for artificial intelligence applications is encouraging hyperscale operators to deploy GPU-intensive server environments with significantly higher rack power densities compared with traditional computing systems.

AI model development and large-scale data processing require advanced infrastructure capable of supporting high-performance workloads, creating substantial demand for hyperscale facilities. Enterprise digital transformation and the continued movement of business applications to public cloud platforms are further strengthening market growth. In addition, hyperscale companies are increasing investments in renewable energy procurement to support sustainability objectives and meet rising power requirements. The technology sector represented nearly 40% of corporate renewable power purchase agreements in 2025, highlighting the increasing role of hyperscale operators in clean energy adoption. Large-scale renewable energy agreements signed by major technology companies demonstrate a long-term approach toward securing reliable and sustainable energy sources for future data center expansion.

The hyperscale data center market from the hardware segment accounted for USD 76 billion, representing 93.9% share in 2025. This dominance is attributed to the significant capital requirements associated with hyperscale infrastructure development. Servers, storage systems, networking equipment, power management solutions, and cooling technologies collectively account for most of the investment required to establish and expand hyperscale facilities.

The air-based cooling systems segment held a 52.8% share, valued at USD 42.7 billion in 2025, owing to increasing thermal management requirements from AI-driven GPU infrastructure. Direct-to-chip (DTC) liquid cooling has emerged as the most adopted liquid cooling method, as it enables coolant circulation through cold plates attached directly to processors and memory components. This approach provides effective heat removal while allowing integration with existing data center designs without requiring complete infrastructure replacement.

North America hyperscale data center market held 42.8% share in 2025. The region continues to serve as the foundation of worldwide hyperscale investment, supported by significant infrastructure expansion from leading cloud and technology companies across major data center hubs. Continued capital spending by Amazon Web Services, Microsoft Azure, Google Cloud, Meta Platforms, and Oracle is contributing to the development of advanced hyperscale facilities across key markets, strengthening North America’s position in the global industry.

Key players operating in the global hyperscale data center market include Schneider Electric, Vertiv Group, Dell Technologies, Hewlett Packard Enterprise (HPE), Cisco Systems, Huawei Technologies, Super Micro Computer, Arista Networks, Lenovo, Eaton, Legrand, Hitachi, Gigabyte Technology, MiTAC Holdings, Cormant Technologies, Mortenson Construction, DPR Construction, Vast Data, LiquidStack, and GRC.

Companies operating in the hyperscale data center market are adopting strategies focused on infrastructure expansion, strategic partnerships, technological innovation, and sustainability initiatives to strengthen their market position. Industry participants are investing in advanced cooling solutions, high-density computing infrastructure, renewable energy integration, and next-generation networking technologies to address increasing AI and cloud workload requirements. Companies are also expanding their global data center footprints through new facility construction, acquisitions, and collaborations with energy providers and technology partners. Developing energy-efficient designs, improving operational reliability, and enhancing data center automation capabilities are key priorities for market participants.

The post Hyperscale Data Center Market to Surpass USD 624.2 Billion by 2035 appeared first on Data Center POST.

Data Center POST had the opportunity to connect with Shahar Belkin, Chief Evangelist at ZutaCore. Starting in 2018, he has led the development and product organizations, contributing to the advancement of next-generation two-phase cooling solutions for high-power silicon and data center infrastructure. Since 2024, he has served as the company’s Chief Evangelist, representing ZutaCore’s technology vision across the global AI and high-performance computing ecosystem.

What does ZutaCore do?

ZutaCore^® delivers waterless, two-phase direct-to-chip liquid cooling for data centres. The technology removes heat directly from CPUs and GPUs using a non-conductive dielectric fluid rather than water. Our flagship solution, ZutaCore HyperCool^®, helps AI, cloud and enterprise data centres cool higher-power processors, increase compute density and cut cooling energy use in half.

Proven at scale in demanding data centre environments, HyperCool eliminates water risk at the server level and gives operators a safer way to move beyond air cooling. Its cold plates, dielectric fluid, manifolds, cooling distribution units and control software are engineered together as one thermal architecture rather than sold as separate parts. In AI cooling, performance depends on how effectively the entire system works together, rather than on any single component.

What problems does ZutaCore solve in the market?

AI and high-performance workloads are pushing CPUs and GPUs to power levels that traditional air cooling can struggle to manage efficiently. We help data centre operators address this by removing heat directly at the chip, supporting higher-density compute without bringing water inside the server.

HyperCool enables this through a waterless, closed-loop, two-phase architecture that uses a non-conductive dielectric fluid to absorb and remove heat from processors. As rack densities continue to rise, this helps operators reduce water risk, improve cooling efficiency and scale AI infrastructure more safely.

What are ZutaCore’s core products or services?

Our flagship product is HyperCool, a waterless, two-phase direct-to-chip liquid cooling solution used across AI, HPC, cloud, financial services, enterprise and colocation environments. Every part of HyperCool is engineered as one integrated thermal system rather than optimised in isolation. It follows the same systems-engineering approach a Formula 1 team applies to overall lap time rather than any single part of the car.

HyperCool uses cold plates mounted directly on CPUs and GPUs. Inside each cold plate, our dielectric fluid absorbs heat from the processor, boils into vapour and carries that heat away from the chip. The vapour then condenses back into liquid in the heat rejection unit and recirculates through the closed-loop system.

We have also introduced OmniTherm™, a two-phase cold plate designed for advanced server configurations, including reversed orientation, inverted mounting, back-side power delivery and non-standard layouts. It gives data centre operators and system designers more thermal flexibility for high-density AI, HPC and cloud infrastructure.

What markets do you serve?

We serve the digital infrastructure market. Our focus is on AI data centres, high-performance computing, cloud, enterprise, colocation and edge environments. We also work with performance-sensitive sectors such as financial services and high frequency trading.

In these environments, cooling can affect latency consistency and hardware performance. It can also influence whether firms can deploy dense compute in strategic colocation facilities close to exchanges and liquidity venues.

What challenges does the global digital infrastructure industry face today?

Data centres are trying to support far more compute while using energy, space and water more carefully. AI workloads are pushing chip power and rack densities beyond what legacy air cooling was designed to handle. Facilities that once planned for tens of kilowatts per rack are now planning for much higher densities, and the industry is already discussing megawatt-class racks.

Power and cooling capacity are now also shaping growth, so operators need systems that use less energy, keep water away from IT equipment and reduce deployment risk. Yet much of the industry still discusses cooling in terms of individual parts, whether the cooling distribution unit, cold plate or coolant. The reality is that as power and density climb, cooling has become a systems-engineering challenge, where the design of each component affects the performance of the whole system.

How is ZutaCore adapting to these challenges?

We are scaling HyperCool from rack-level deployments to end-of-row and larger cooling systems capable of supporting higher-density AI infrastructure. By removing heat at the processor through a sealed, waterless two-phase loop, HyperCool helps operators reduce cooling energy and keep water away from IT equipment.

We are also validating multi-megawatt operation and designing systems around the practical requirements of live data centres. At our facility in Israel, we have built a 2 MW end-of-row emulation platform that replicates real-world thermal and facility conditions without using live IT equipment. This allows us to validate performance and stability at multi-megawatt scale ahead of customer deployment. OmniTherm extends the same approach to advanced server designs, including reversed orientation and back-side power delivery, giving operators more flexibility as AI hardware becomes denser.

What can we expect to see/hear from ZutaCore in the future?

We will use our recent $100 million Series C funding round to expand global commercialisation and advance R&D for next-generation AI and HPC chip cooling. The funding will also support work on in-package thermal management, megawatt-class deployments and integration with existing air and single-phase cooling systems.

We have more than 75 deployments worldwide today, and we are focused on scaling our platform as operators look to support higher compute density, reduce water risk and meet rising AI infrastructure demand. Our two-phase platform is designed to support next generation AI and HPC processors exceeding 4,000 watts. OmniTherm now supports full-power operation for the NVIDIA RTX PRO™ 6000 Blackwell Server Edition in a single-slot PCIe form factor. We will continue to advance the platform to meet the evolving requirements of AI-driven data centres.

What upcoming industry events will you be attending?

We will then be a Gold Sponsor at DataCloud USA in Austin, Texas on 2 and 3 September, where we will moderate the panel “Keeping it cool: evaluating air, liquid and hybrid approaches for high-density workloads”.

ZutaCore will also be a Sapphire Sponsor at the OCP Global Summit in San Jose, California from 12 to 15 October. On 15 October at 1:05pm, we will host the panel “Reducing Operational Complexity through Standard Two-Phase Liquid Cooling for AI Infrastructure” with Wiwynn, ASRockRack and Corinitis. We will also attend SuperComputing’26 in Chicago, Illinois from 15 to 20 November.

Do you have any recent news you would like us to highlight?

We recently announced our $100 million Series C funding round, with investment from Mitsubishi Electric, Carrier Ventures, Samsung Electronics through Samsung Ventures and additional investors. The funding will support global commercialisation, further R&D and continued work on cooling technologies for next-generation AI and HPC processors.

We also announced more than 75 deployments worldwide across the Americas, Europe and Asia, along with senior hires across finance, people, research and development and operations as we scale for further global growth. Those hires include Yaniv Reinhold as Chief Financial Officer, Sarah Warshavsky Oberman as Chief People Officer, Yoni Nir as Chief R&D Officer and Sharon Shafran as Chief Operating Officer, joining CEO Erez Freibach, President and CRO Brian Lillie, Chief Product and Technology Officer My Truong and Chief of Staff Susan Mor.

Is there anything else you would like our readers to know about ZutaCore and capabilities?

HyperCool is more than cold plates and fluid. It includes HyperCool Cloud, our software layer for real-time monitoring, control and optimisation across racks, rows and sites. The full system, with its hardware and software working together, is what customers are deploying.

ZutaCore was named an IDC Innovator in 2025, won Best Supercomputing Cooling Innovation at ISC 2025, and is a member of the Sustainable Digital Infrastructure Alliance and the STAC Alliance. Our $100 million Series C, backed by strategic investors including Mitsubishi Electric, Carrier Ventures and Samsung Electronics, reflects strong confidence in this approach as AI infrastructure moves to megawatt-class density.

Where can our readers learn more about ZutaCore?

Readers can visit the ZutaCore website and follow ZutaCore on LinkedIn to learn more about our technology, company news and upcoming events.

# # #

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 ZutaCore is Scaling Global AI Infrastructure appeared first on Data Center POST.

TL;DR

• In 2026, AI is shifting cloud ERP from a retrospective system of record that merely stores data into a system of prediction and autonomous action.
• Businesses achieve the most immediate, measurable returns by focusing on intelligent process automation (such as automated invoice matching), predictive financial forecasting to continuously update baseline models, and predictive inventory management to prevent supply chain stockouts.
• Successful deployment requires organizations to audit their data quality first, target two or three high-impact processes for initial pilots, establish clear boundaries for what AI can decide versus what requires human sign-off, and continuously check for model drift.

# # #

Most finance and operations teams don’t wake up asking for “AI in their ERP.” They wake up frustrated that month-end close still takes eight days, that inventory forecasts are wrong every quarter, and that someone still has to reconcile three systems before a board meeting manually. That’s the real starting point for this topic, not the buzzword.

Cloud ERP software has quietly become the operating backbone of most mid-size and enterprise businesses. What’s changed in 2026 isn’t the ERP itself – it’s what’s running underneath it. AI is moving from a bolt-on reporting feature to something closer to a coworker inside the system: reading documents, flagging anomalies, forecasting cash, and in some cases, taking action without waiting for a person to click “approve.”

This guide covers what that actually looks like: what’s genuinely new, where the real ROI shows up, and what to check before committing budget to an “AI-powered” ERP upgrade.

Key Takeaways

• AI-powered cloud ERP software combines the core system of record (finance, inventory, HR, supply chain) with machine learning, natural-language interfaces, and increasingly, autonomous AI agents.
• The highest-ROI use cases in 2026 are financial forecasting, predictive inventory management, and process automation, not chatbots.
• Agentic ERP workflows, AI that acts rather than just answers, are the biggest structural shift from the last ERP generation.
• The main adoption barrier isn’t the AI itself. It’s data quality, integration debt, and change management inside the business.
•  Judge any “AI-powered ERP” claim against three questions: what decision does it actually improve, what data does it need to work, and what happens when it’s wrong.

What Is Cloud ERP Software?

Cloud ERP software is a hosted platform that unifies core business functions, including finance, accounting, inventory, procurement, human resources, and supply chain, into a single system accessible over the internet rather than installed on local servers.

Instead of maintaining separate spreadsheets or on-premise systems for each department, a cloud ERP software gives every team a shared source of truth: one general ledger, one inventory count, one employee record. Vendors host the infrastructure, push updates automatically, and typically charge on a subscription basis.

Common examples include SAP S/4HANA Cloud, Oracle Fusion Cloud, Microsoft Dynamics 365, Oracle NetSuite, and Workday, all of which have added AI capabilities on top of this same core structure over the past two to three years.

How AI Is Changing Cloud ERP Software

AI changes cloud ERP in three layers: it makes existing processes faster through automation, it makes predictions more accurate through forecasting, and it makes the system usable by people who aren’t ERP specialists through natural-language interfaces.

Before AI, cloud ERP was primarily a system of record that stored data and let people query it. Reports were retrospective, so a business would find out inventory was short after the shelf was already empty. AI shifts ERP toward being a system of prediction and, increasingly, a system of action, flagging the shortage before it happens or reordering automatically within limits a human has set.

Analyst coverage over the past two to three years has consistently pointed the same direction: AI adoption inside ERP is accelerating faster than in most other enterprise software categories, with forecasting, reconciliation, and anomaly detection cited most often as the use cases delivering measurable time savings first.

Top AI-Powered Cloud ERP Platforms in 2026

Treat this table as a starting point, not a verdict. The right platform depends on your existing tech stack, industry, and which processes you actually want AI to touch first.

How to Evaluate an AI-Powered Cloud ERP Platform

Choosing between AI-powered ERP platforms comes down to a handful of criteria that matter more than any feature list, since most vendors now offer some version of automation, forecasting, and a conversational assistant.

• Data security and compliance: Confirm certifications relevant to your industry, such as SOC 2 or ISO 27001, and where the AI models process and store your data, especially if you operate under GDPR, HIPAA, or similar regulations.
•  Integration with your existing stack: Check what pre-built connectors exist for your CRM, e-commerce, and logistics tools, and how much custom development an integration will realistically require.
• AI governance controls: Look for granular settings on what an AI agent can do without approval, a clear audit trail of automated actions, and an easy way to override or roll back a decision.
• Total cost of ownership: Add up the base ERP license, AI module add-ons, implementation, and training, since the AI layer is rarely included in the headline subscription price.
•  Vendor roadmap and update cadence: Ask how often AI features change and whether you can control when updates roll out, especially for a business that can’t tolerate workflow disruption mid-quarter.
• Support for AI-specific issues: Confirm who you actually contact when an automated forecast or reconciliation looks wrong, and what the response-time commitment is.
• A sandbox test with your own data: Request a pilot using a real slice of your transaction data before signing, since AI accuracy claims in a demo rarely reflect how a model performs on your specific data quality.

In short: evaluate the platform on how it handles your actual data and processes, not on how convincing the AI demo looks.

7 Key Ways AI Is Transforming Cloud ERP Software in 2026

1. Intelligent Process Automation

Intelligent process automation means AI-driven bots handle repetitive, rules-based ERP tasks, such as invoice matching, purchase order approvals, and data entry, without a person doing it manually each time.

Unlike older robotic process automation tools that just followed rigid scripts, today’s automation layers use machine learning to handle exceptions. An invoice with a mismatched PO number gets flagged and routed correctly instead of stalling the whole workflow. That’s the practical difference: less babysitting, fewer stuck queues.

2. AI-Powered Financial Forecasting

AI-powered financial forecasting uses historical transaction data, seasonality, and external signals to project cash flow, revenue, and expenses more accurately than manual spreadsheet models.

Finance teams that used to rebuild forecasts every month by hand now get a continuously updated baseline that adjusts as new data comes in. The finance team still owns the assumptions and the final call. The AI just removes the grunt work of rebuilding the model from scratch each cycle.

3. Predictive Inventory and Supply Chain Management

Predictive inventory management uses AI to forecast demand at the SKU level and flag supply chain risks before they cause a stockout or overstock.

This is where ROI tends to show up fastest for product-based businesses: fewer emergency reorders, less capital tied up in dead stock, and earlier warning when a supplier is likely to miss a delivery window based on historical patterns.

4. Conversational ERP and Natural-Language Analytics

Conversational ERP lets non-technical users ask questions in plain language, such as “what were our top five underperforming SKUs last quarter,” and get an answer without writing a report or waiting on the analytics team.

This matters more for adoption than it sounds. A lot of ERP value has historically been locked behind people who know how to build the right report. Natural-language interfaces open that same data up to managers who never learned the reporting tool.

5. AI Agents and Agentic ERP Workflows

Agentic ERP workflows are AI systems that don’t just answer questions or flag issues. They complete multi-step tasks on their own, within boundaries a business sets, such as reconciling a discrepancy, generating a draft purchase order, or routing an approval to the right person.

This is the biggest structural change from the last wave of AI in ERP. Earlier tools mostly assisted a human who was still doing the work. Agentic workflows shift some of that work to the system itself, with a human reviewing outcomes rather than performing every step.

6. Real-Time Business Intelligence

Real-time business intelligence means dashboards and alerts update continuously as transactions happen, instead of refreshing on a nightly or weekly batch cycle.

For a leadership team, that’s the difference between finding out about a cash flow problem in next Monday’s meeting versus getting flagged the moment the pattern starts. AI adds the layer that decides what’s worth surfacing, instead of leaving someone to notice it in a static report.

7. AI-Driven Risk, Security and Compliance

AI-driven risk and compliance tools continuously monitor transactions for fraud patterns, policy violations, and regulatory exposure, rather than relying solely on periodic audits.

This includes anomaly detection on expense reports and payments, automated checks against changing tax and regulatory rules across jurisdictions, and audit trails that are easier to produce because the system has been tracking exceptions all along instead of reconstructing them after the fact.

Cloud ERP Software vs AI-Powered Cloud ERP

The core system, including the general ledger, inventory, and HR records, doesn’t change. What changes is how much of the analysis and routine decision-making the system can do before a person has to step in.

Benefits of AI-Powered Cloud ERP Software

• Faster close cycles: Automated reconciliation and anomaly flagging cut down the manual chase during month-end close.
• More accurate forecasting: Continuously updated models reduce the lag between something changing and the forecast reflecting it.
• Lower operational overhead: Fewer people spend time on data entry, matching, and routine exception handling.
• Earlier risk detection: Fraud, compliance, and supply chain risks get flagged as patterns emerge, not after a quarterly audit.
• Broader data access: Natural-language querying means more people in the business can self-serve answers instead of waiting on a report.

None of these benefits are guaranteed just by turning AI features on. They depend on clean underlying data and processes that are actually being followed, which is exactly where most implementations struggle.

Challenges of Integrating AI With Cloud ERP Software

• Data quality debt: AI forecasts and automations are only as good as the transaction data underneath them. Years of inconsistent entries don’t get fixed by turning on AI.
• Integration complexity: Most businesses run ERP alongside a patchwork of other tools, such as CRM, e-commerce, and logistics platforms, and AI features often need clean data flowing from all of them, not just the ERP itself.
• Change management: Teams that have run a process manually for years don’t automatically trust an AI recommendation, especially if the first few outputs are wrong.
• Governance and oversight: Agentic workflows that can take action need clear boundaries around what the AI is allowed to do without approval, and what always needs a human sign-off.
• Cost and licensing: AI modules are frequently priced as add-ons, and the total cost isn’t always obvious from the initial vendor pitch.

How to Prepare Your Business for AI-Powered Cloud ERP

Step 1: Audit your data quality first

Look at how consistent your chart of accounts, SKU records, and vendor data actually are. This is the single biggest predictor of whether AI features will work.

Step 2: Pick two or three processes, not everything at once

Month-end reconciliation, demand forecasting, or invoice matching are common starting points with measurable ROI.

Step 3: Define what “AI-assisted” versus “AI-automated” means for each process

Decide upfront what the system can do on its own and what still needs a human sign-off.

Step 4: Involve the people who do the work today

The team currently doing manual reconciliation or forecasting should help define what “correct” looks like. They’ll spot AI mistakes faster than anyone else.

Step 5: Run a pilot before a full rollout

Test the AI features on one business unit or process for a full reporting cycle before extending them company-wide.

Step 6: Set a review cadence

AI models drift as business conditions change, so build in a regular accuracy check rather than a one-time setup.

What Is the Future of Cloud ERP Software?

The direction is toward ERP systems that act more like a coordinated team of specialized agents than a single monolithic application: one agent handling procurement exceptions, another managing cash forecasting, another monitoring compliance, all working from the same underlying data and increasingly coordinating with each other.

Expect tighter integration between ERP and adjacent systems, such as CRM, HR, and supply chain planning, as vendors compete on how well their AI layer reasons across all of it, not just within one module. The businesses that get the most value won’t necessarily be the ones with the newest AI features. They’ll be the ones with clean data and clear rules for what the AI is allowed to decide on its own.

Conclusion

AI isn’t replacing cloud ERP. It’s changing what the system is capable of doing without a person driving every step. The businesses getting real value out of this shift aren’t the ones chasing every new AI feature. They’re the ones that fixed their data quality, picked a couple of high-impact processes, and set clear rules for what the AI can decide versus what still needs a human.

If you’re evaluating an AI-powered ERP upgrade, start with the question that actually matters: which specific decision, whether forecasting, reconciliation, inventory, or compliance, do you most need to make faster or more accurately? That answer should drive the platform and features you choose, not the other way around.

Considering an AI-powered ERP upgrade for your business? Talk to our team about which processes are worth automating first, and which aren’t.

Frequently Asked Questions

What is AI-powered cloud ERP software?

AI-powered cloud ERP software is a cloud-based enterprise resource planning system that uses machine learning, natural-language processing, and in some cases autonomous AI agents to automate tasks, generate forecasts, and support decisions across finance, inventory, HR, and supply chain, on top of the same core system of record as traditional ERP.

How much does an AI-powered ERP system cost?

Costs vary widely by vendor, company size, and which modules are deployed, and AI features are frequently priced as an add-on to the base ERP subscription. Get a detailed quote scoped to your specific processes rather than relying on published starting prices, which rarely reflect the full cost.

Is AI replacing ERP staff and finance teams?

No. AI is removing manual, repetitive work like data entry and reconciliation, but forecasting assumptions, exception judgment calls, and financial sign-off still require human oversight, especially as agentic workflows expand.

What’s the difference between AI-assisted and agentic ERP?

AI-assisted ERP surfaces insights and recommendations for a person to act on. Agentic ERP goes a step further and completes multi-step tasks on its own, within boundaries the business defines, with a human reviewing outcomes rather than performing every step.

Which businesses benefit most from AI-powered cloud ERP?

Businesses with high transaction volume, multiple locations or subsidiaries, and inventory-heavy operations tend to see the fastest ROI, since forecasting and reconciliation automation scale with volume.

What are the biggest risks of AI in ERP?

The main risks are acting on inaccurate data, since AI outputs inherit any underlying data quality problems, unclear governance over what AI agents can do without approval, and over-trusting automated outputs without a human review step.

Do I need to replace my existing ERP to get AI features?

Not necessarily. Most major ERP vendors are adding AI capabilities to their existing cloud platforms through upgrades or add-on modules, so a full replacement often isn’t required if you’re already on a modern cloud ERP.

How long does it take to implement AI features in an existing ERP?

Timelines depend on data readiness and process complexity, but a focused pilot on one or two processes typically takes a few months, while a company-wide rollout across all AI modules can take considerably longer.

The post How AI Is Transforming Cloud ERP Software in 2026 appeared first on Data Center POST.

TL;DR

• Leadership appointment: Dipan Patel joins Duos as Chief Operating Officer, reporting to CEO Doug Recker.
• Operational focus: Patel will lead deployment of Duos Edge AI data centers and GPU-as-a-Service platforms nationwide.
• Industry experience: He brings leadership experience across data centers, fiber networks, communications infrastructure, and technology strategy.

Duos Technologies Group, Inc. (“Duos” or the “Company”) (Nasdaq: DUOT) has appointed Dipan Patel as Chief Operating Officer, effective immediately. Patel joins the Company’s executive leadership team as Duos continues to scale its modular edge data center, colocation, and technology infrastructure solutions business.

As COO, Patel will have operational accountability for the deployment of Duos Edge AI data centers and GPU-as-a-Service platforms across the United States. Working with the executive leadership team, he will help strengthen operational performance and build the capabilities needed to support Duos’ growth as a next-generation AI infrastructure platform.

“Dipan brings a unique combination of deep infrastructure, technology, and operational leadership experience that aligns directly with our Company’s next phase of growth,” said Doug Recker, CEO of Duos. “His track record of building and scaling businesses across data centers and communications infrastructure will be instrumental as we expand our Edge AI platform and execute on our growth objectives.”

Prior to joining Duos, Patel served as a Digital Infrastructure Solutions Executive at Telstra InfraCo, where he led product strategy, solution innovation, and commercial monetization across a national portfolio of data centers, fiber networks, exchange facilities, and critical communications infrastructure. He previously served as Executive Vice President of Strategy, Technology and New Business Initiatives at SBA Communications Corporation, where he established the company’s corporate growth division and led investments in data centers, power solutions, and open-access fiber networks.

“AI is driving a once-in-a-generation infrastructure buildout, and Duos already has a proven modular platform and a team that can move fast,” said Dipan Patel. “I look forward to leveraging my background to help scale our operations, execute on our growth strategy, and expand the Duos footprint.”

Patel holds a PhD in Digital Communications and a BEng degree in Information Systems Engineering from Imperial College London. He is a Senior Member of the IEEE and holds 12 granted domestic and international patents.

The appointment reflects Duos’ continued investment in operational leadership as the Company expands its Edge AI-powered digital infrastructure ecosystem.

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

The post Duos Technologies Group, Inc. Names Dipan Patel Chief Operating Officer appeared first on Data Center POST.

TL;DR

• While cloud computing has traditionally been defined by speed, scale, and flexibility, mature environments now present unpredictable costs, complex governance, and rising sovereignty concerns.
• Abstraction at scale has created a disconnect between organizations and the systems they must manage, with pricing hikes and licensing changes forcing businesses to re-evaluate control versus convenience.
• The future of cloud is defined not by how much an organization can scale, but by how effectively it can scale while retaining control, choice, and sovereignty.

# # #

By Juan Aguirre, CCO at Ilkari

Cloud computing has been defined by scale, speed, and flexibility for over a decade. As data volumes grew exponentially, organisations turned to the cloud to support growing workloads, enable rapid expansion, and to keep pace with increasingly dynamic industries.

But as cloud environments continue to mature, a new reality has emerged: flexibility alone is no longer enough. Today, many organisations face a very different data challenge. While traditional cloud systems still deliver in principle, how they work has ultimately changed how organisations operate and make decisions.

Costs are now more difficult to predict, governance is more complex to enforce, and sovereignty concerns are shaping infrastructure decisions in ways we’ve never seen before. What was once a decision based on convenience versus control is now a strategic business decision.

But the issue isn’t that the cloud has failed, it’s that abstraction at scale can create a disconnect from the systems organisations need to manage effectively.

Pricing hikes, changes in licensing, and growing dependence on third-party platforms are driving businesses to re-evaluate the balance between control and convenience

For steady, regulated, and performance-sensitive workloads, the answer is not more abstraction, but more selectivity. Organisations need cloud environments that are flexible where it matters, but bounded where control, predictability, and sovereignty are essential.

That requires designing infrastructure with cost control and governance built in from the start, rather than layered on later as an afterthought. It means recognising that not every workload needs unlimited elasticity. And it means creating cloud models that preserve choice over time, rather than quietly restricting it.

The next phase of cloud will not be defined by how much can be scaled, but by how effectively organisations can scale without losing control.

# # #

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 From Cost Drift to Sovereignty Demands: Why the Future of Cloud Is Selective appeared first on Data Center POST.

TL;DR

• Revenue growth: Second-quarter revenue increased nearly 30% year over year to $6.18 million, driven by the initial ramp in Technology Solutions revenue.
• Commercial progress: Duos signed 55 MW of hosting agreements with Axe Compute, secured a $111 million hyperscaler agreement in Columbus, Georgia, and now has 25 MW contracted for deployment in 2026.
• Outlook reaffirmed: The Company expects to deploy 25 MW this year and generate more than $50 million in 2026 revenue.

# # #

Duos Technologies Group, Inc. (“Duos” or the “Company”) (Nasdaq: DUOT), reported second-quarter 2026 results, reflecting continued progress in its transition to a focused provider of AI and edge data center infrastructure.

Revenue for the quarter ended June 30, 2026 increased nearly 30% to $6.18 million, compared with $4.77 million in the same period last year. The increase was driven primarily by the Company’s Technology Solutions business, which generated approximately $3.23 million in quarterly revenue through sourcing, integration, and supply chain services that support data center, AI, and enterprise computing deployments.

Duos also reported its first positive operating quarter as a data center infrastructure company, with approximately $0.05 million in operating income. Gross margin increased 94% year over year to $3.45 million, and adjusted EBITDA was positive at $0.5 million.

“Financially, we began to see the early stages of the substantial performance ramp we expect to build over the course of this year, highlighted by a 30% increase in revenue and a material improvement in profitability,” said Doug Recker, CEO of Duos. “We also secured over $100 million in growth capital through two major transactions.”

The Company continued to expand its commercial momentum during the quarter and in the weeks that followed. Duos signed five-year, 55 MW hosting agreements with Axe Compute valued at more than $500 million and secured $111 million in contracted revenue through a five-year agreement with an investment-grade hyperscaler for 10 MW of critical IT-load capacity at its Columbus, Georgia campus.

Duos now has 25 MW contracted, with all 25 MW planned for deployment in 2026. The Company also entered into an exclusive, non-binding term sheet with 0Lat LLC for a proposed structured lease across a 15-site, 225-cabinet edge data center portfolio in Texas and Georgia.

During the quarter, Duos secured more than $100 million in growth capital through a $55 million registered direct offering and $50.4 million in proceeds related to the sale of substantially all assets of New APR Energy, LLC. These proceeds support the Company’s planned Columbus campus expansion, related infrastructure investments, and contracted customer deployments.

Following the sale of its rail technology subsidiary, Duos is now fully focused on scaling its Edge Data Center and AI infrastructure businesses through Duos Edge AI and Duos Technology Solutions. The Company reaffirmed its 2026 expectation for revenue exceeding $50 million and 25 MW of deployed capacity by year end.

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

The post Duos Technologies Group, Inc. Reports Q2 2026 Results appeared first on Data Center POST.

TL;DR

• The rapid expansion of AI computing has significantly increased energy consumption and altered the thermal and power profiles of server racks, forcing operators to rethink traditional facility designs.
• Designers are transitioning from traditional metals and legacy materials to high-performance thermoplastics, especially in environments where metal is prone to electrical interference, corrosion, or flammability risks.
• These advanced materials are best suited for critical architectural and equipment functions, including airflow containment systems, rack doors, plenum partitions, cable management, and power distribution.
• Thermoplastics provide a lightweight, highly available, and cost-effective alternative that features excellent impact and chemical resistance, strict flame-resistance compliance, and high dielectric insulation to prevent short circuits.

# # #

By Dan Mesler, Market Development Manager at PLASKOLITE

The rapid explosion of AI intelligence has altered the infrastructure landscape, escalating requirements for computing power and cloud storage reliability. This allows for global data consumption to drive unprecedented demand for hyperscale facility expansion. However, as advanced AI workloads multiply, operators face severe physical constraints – AI chips consume massive amounts of energy and drastically alter the thermal and power profiles of server racks.

To survive in environments that grow hotter with every complex query, facility designers must rethink their traditional reliance on legacy materials.

A substantial portion of infrastructure investment is shifting toward high-performance thermoplastics. Materials such as polycarbonate, PVC, and CPVC are actively replacing traditional perforated metal and legacy materials. Thermoplastics perform equally well or better than metals while offering significant structural advantages, particularly in facility zones where metal cannot be used due to electrical interference, severe corrosion risks, or strict flame-resistance mandates.

Within modern facility layouts, specialized materials like polycarbonate, acrylic, HDPE, and polypropylene are ideally suited for critical architectural and equipment applications. Key applications where these thermoplastics are best suited are:

• Airflow and containment systems
• Door and panel glazing
• Rack doors and side panels
• Plenum partitions and ceiling tiles
• Cable management and power distribution

The widespread adoption of these materials is fueled by their unique combination of physical properties and holding up well in extreme environments. High-performance plastics are inherently lightweight, cost-effective, and highly available compared to custom metal fabrications. More importantly, they provide excellent impact resistance, stand up well to chemical exposure and high temperatures produced in servers, have critical flame resistance that meets strict UL and FM flammability ratings, and have high dielectric insulation to prevent electrical short circuits.

As the artificial intelligence economy scales, every square foot of real estate is at an absolute premium. Stepping back from the chip itself, it is clear that facility design must adapt alongside server hardware. By utilizing these high-performance plastics, data center operators can successfully mitigate the extreme thermal, chemical, and electrical challenges of the next computing era.

# # #

About the Author

Dan Mesler is the Market Development Manager at PLASKOLITE, where he focuses on expanding the company’s presence in the data center market and identifying new opportunities for thermoplastic solutions within the industry. With approximately 15 years of experience in the plastics industry, he brings a strong combination of product knowledge, market expertise, and customer-focused strategy to his role.

The post How Thermoplastics Support Data Center Workloads appeared first on Data Center POST.

TL;DR

• The rising adoption of modular data center technologies is driving market growth by lowering civil engineering, design, and turnkey contracting expenses by approximately 30% compared to traditional data centers.
• Rising energy demands are pushing operators to adopt renewable energy sources (such as wind and solar) and implement advanced cooling techniques like liquid and AI-driven cooling to optimize efficiency.
• Propelled by technology giants investing in digital infrastructure and expanded cloud computing services, the United States dominated the sector in 2024 with USD 3.6 billion in revenue.

# # #

As per the report by Global Market Insights, Inc. “Data Centre Power Market ’’ was valued at USD 70 billion in 2025, and is projected to grow at a CAGR of 9.1% from 2026 to 2035.

The Global Data Center Power Market, The increasing adoption of modular data centers is a major factor driving this growth, as these facilities offer enhanced efficiency, scalability, and flexibility. They are becoming a preferred solution to meet the rising demand for data processing and storage. Modular power systems integrate generators, power distribution units, and uninterruptible power supplies in prefabricated designs that reduce costs and energy consumption, leading to lower operational expenses and a smaller carbon footprint. As digital infrastructure expands, the need for robust and energy-efficient power solutions continues to grow.

Leading technology firms and internet companies have already embraced modular data center technologies to optimize costs, reduce energy consumption, and accelerate deployment. Compared to traditional data centers, modular facilities lower civil engineering costs and cut design, management, and turnkey contracting expenses by approximately 30%. The push toward low-latency data processing, driven by advancements in edge AI, 5G networks, and cloud computing, is further fueling demand for high-performance power solutions in data centers.

The market is segmented by application into BFSI, colocation, energy, government, healthcare, manufacturing, IT & telecom, and others. In 2024, IT & telecom accounted for 24% of the market, with the colocation segment expected to surpass USD 11 billion in revenue by 2034. Both IT and telecommunications providers are prioritizing power-efficient data centers, focusing on cooling technologies that enhance performance, reduce costs, and minimize environmental impact. As high-performance computing requirements grow, energy-efficient cooling and power systems are becoming critical for managing power-intensive workloads.

The power market for data centers is also classified by component into solutions and services. In 2024, the solutions segment held a 63% market share, while the services sector is anticipated to expand at a CAGR of approximately 8% through 2034. The transition toward renewable energy sources, including wind and solar, is gaining momentum due to cost efficiency and environmental concerns. This shift is transforming power distribution units (PDUs) and uninterruptible power supplies (UPS), which are now incorporating advanced monitoring and management features to optimize power reliability and efficiency in data centers.

By data center type, the market is categorized into hyperscale, colocation, edge, and enterprise facilities. The colocation sector led the market with USD 5.7 billion in 2024, driven by increasing energy costs and the need for energy-efficient operations. Colocation providers are actively investing in renewable energy solutions and adopting advanced cooling methods such as liquid and AI-driven cooling to enhance efficiency and reduce operating expenses. These innovations help lower reliance on conventional air conditioning while improving the cooling management of high-density IT infrastructure.

In North America, the United States dominated the data center power sector in 2024, generating USD 3.6 billion in revenue. Growth in this region is propelled by technology giants investing in efficient power infrastructure to support expanding data center operations. The rising adoption of cloud computing services by enterprises of all sizes further fuels the demand for advanced power solutions. With continuous advancements in digital infrastructure, efficient and sustainable power systems will remain a priority for data center operators worldwide.

The post Data Center Power Market Set to Reach USD 169.1 Billion by 2035 appeared first on Data Center POST.

TL;DR

• 55 MW secured: Five-year hosting agreements cover 55 MW of AI facility capacity.
• $500M+ value: Agreements represent more than $500 million in base payments over five years.
• AI capacity ahead: Initial project readiness is targeted for late 2026 and early 2027, subject to testing and acceptance.

# # #

Duos Technologies Group (Nasdaq: DUOT) has announced that two of its project entities have executed five-year hosting service orders with Axe Compute Inc. (Nasdaq: AGPU), a neocloud AI infrastructure platform providing dedicated enterprise GPU compute capacity.

The agreements reserve an aggregate 55 MW of AI facility capacity across multiple U.S. data center sites and are valued at more than $500 million in aggregate contractual base payments over their initial five-year terms. Electricity and other usage-based charges are excluded from that value, and billing is subject to successful ready-for-service testing and Axe Compute’s written acceptance of each deployment.

“These executed agreements represent an important advancement of our AI infrastructure strategy and demonstrate our ability to translate development opportunities into long-term commercial relationships,” said Doug Recker, Chief Executive Officer of Duos. “Axe Compute brings a clear vision for deploying high-performance AI capacity at scale.”

The projects are being designed around the density, cooling, and availability requirements of next-generation GPU systems. Initial project readiness is targeted to begin in late 2026 and continue into early 2027, subject to construction, commissioning, performance testing, and Axe Compute’s written acceptance.

“It is a significant milestone in the growth of Axe Compute’s AI infrastructure platform,” said Chris Miglino, Chief Executive Officer of Axe Compute. “We look forward to working with Duos to bring this capacity online and support the growing compute requirements of our customers.”

The agreements include renewal options and rights supporting potential future expansion. Duos and Axe Compute have also signed non-binding term sheets that contemplate potential minority investments by Axe Compute in the special-purpose entities associated with the projects, while Duos is expected to retain majority ownership. Any such investment remains subject to definitive agreements, closing conditions, and approval by both companies.

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

The post Duos Technologies Signs 55 MW AI Hosting Agreements with Axe Compute appeared first on Data Center POST.

TL;DR

• AI workloads are rendering standard real estate playbooks obsolete, forcing data centers to be built and operated like high-density industrial plants.
• Evaluating facilities solely on initial speed, capacity, and cost-per-watt overlooks a decades-long lifecycle that must adapt to rapidly shifting GPU platforms, rack densities, and cooling needs.
• Synchronizing design, factory fabrication, testing, and site preparation in parallel reduces expensive field modifications and bridges traditional silos.

# # #

Written by iMiller Public Relations on behalf of Compu Dynamics Modular.

Modern data centers must be designed and evaluated as ongoing industrial systems rather than static real estate assets evaluated solely on day-one metrics like speed, megawatts, and initial cost per watt. Driven by the intense demands of AI workloads, the industry is shifting away from traditional commercial office playbooks to build facilities that function more like high-density industrial plants. This transition requires end-to-end planning that spans every stage from initial design, offsite factory integration, and site preparation, through decades of daily operations and future technology refreshes.

A critical challenge in modern facility planning is bridging the gap between physical infrastructure, which operates for decades, and compute technology, which evolves at a breakneck pace. As GPU platforms advance, rack densities, power needs, and cooling architectures rapidly shift, requiring infrastructure that can adapt incrementally without disrupting live workloads or requiring major reconstruction. Implementing a proactive five-to-10-year lifecycle strategy allows operators to gracefully integrate technologies like liquid cooling and scale electrical distribution capacity as workloads compress.

Achieving this adaptability requires breaking down traditional silos between engineering, offsite manufacturing, and on-site construction to prevent expensive coordination issues in the field. Utilizing modular delivery fundamentally restructures the execution model by coordinating design, assembly, and testing in a controlled factory environment while site preparation happens in parallel. Ultimately, this end-to-end coordination ensures a predictable, repeatable deployment process and establishes a foundation for seamless upgrades throughout the facility’s longest lifecycle phase: operations.

This content originally appeared on Compu Dynamics Modular’s website and has been adapted for syndication on Data Center POST. Read the complete article here.

The post Why End-to-End Thinking Matters: The Real Data Center Lifecycle in the AI Era appeared first on Data Center POST.

TL;DR

• Automation replaces tedious device-by-device configurations with policy-based deployments across hundreds of systems simultaneously. Continuous monitoring allows systems to instantly recognize abnormalities, redirect workloads, and execute automated incident responses.
• Automated systems use sensor tracking and real-time utilization analysis to dynamically scale power modes and adjust cooling based on actual zone demands, drastically lowering operational costs.
• Using historical performance analytics enables predictive maintenance, shifting operations from reactive firefighting to proactive prevention. Automation also enforces security standards by streamlining mass patching and minimizing human error.
• Instead of replacing human professionals, automation shifts their focus away from repetitive tasks and is most successful when paired with reliable human oversight and control measures.

# # #

Data centers continue to play increasingly large roles in providing digital services. Cloud computing services, artificial intelligence, live streaming, banking systems, and enterprise programs rely heavily on uninterrupted computing power. As systems get more complex, the effort required to manually manage them organically increases. Data center automation is becoming a necessity.

A data center can handle a significant volume of transactions every day, while monitoring facilities receive thousands of pieces of data every minute. It becomes harder to manage systems manually at this scale.

Rising Complexity of Infrastructure

Expansion of infrastructure has been one of the reasons for automation. An establishment with 10,000 hosts has many more configuration points than one with 1,000. A single host would need updates for its operating system, security measures, workload positioning, and checking the performance.

As per Market Intelo, the global data center automation market was valued at $12.4 billion in 2025 and may reach $45.3 billion by 2034, thereby showing a CAGR of 14.2% from 2026 to 2034. The growth indicates the necessity for managing the continued expansion of an IT infrastructure effectively and efficiently.

Automated processes are effective in minimizing repetitive administration activities through standards across multiple systems. Instead of configuring the 500 systems one by one, automation takes care of this in the same manner for all the systems at the same time.

Rapid Incident Detection and Reaction

Speed in operations is an important aspect as well. The minute there is an equipment failure, a link gets disconnected, or utilization climbs up, delays can be costly. Manually conducting an investigation requires the teams to analyze logs, dashboards, and equipment configurations.

An automated monitoring system can instantly recognize abnormal situations and produce set responses. A system can restart a failing service, redirect workload to the existing capacity, and generate an incident ticket as soon as a threshold is triggered.

Improved Capacity and Energy Management

The significance of energy efficiency increases with the complexity of computing. Data centers utilize electrical power for servers, cooling systems, storage, networking, and electric supply systems. Keeping equipment running at high capacities unnecessarily is not only energy-inefficient, but it also leads to excessive operational costs.

Automation provides an analysis of utilization and allows changes in capacity depending on workloads. In the lower demand, certain units may go on low-power mode. In the peak timeframe, more capacity may be switched on automatically.

Cooling systems can also make use of this approach. Sensors allow tracking temperature and humidity in various zones. Control systems will be able to adjust cooling needs on the basis of real needs rather than using fixed schedules. A 5% increase in cooling efficiency can make a great difference.

Improved Safety and Conformity

The process of managing data center security involves making countless decisions regarding configuration. There are many things that need to be kept that way, such as permissions, firmware versions, network policies, and encryption configuration. The need for manual processes adds the risk of missing an update or configuration drift.

Automation can assist in ensuring compliance with the configuration specifications and help in detecting deviations that may have taken place. For example, when there is a need for a software update for two thousand servers, it becomes possible to automate this process.

Developments in Automation

The correlation between automation and analytics has become more pronounced in recent times. Because of the analysis of performance history, we can know about trends that could have been overlooked otherwise.

Thus, if we indicate temperature bursts previous to the crash of the server, then it is possible to use the analytics system to solve the problem. The possibility of predictive maintenance enables engineers to solve this problem during preventive maintenance and not wait for a disaster to occur.

This demonstrates the change in operations from reactive to proactive management. One does not need to wait for utilization to reach 95% for a decision to be made; instead, automated systems can identify possible warning signals at 70% or 80% and take preventive measures accordingly.

Minimizing Human Mistakes

Automation provides consistency by performing recognized processes regardless of the situation. Furthermore, it can help with validation checks before introducing changes in work processes. Consistency may be crucial in large-company settings wherein numerous daily operational changes take place.

What Automation Usually Includes

The automation of data centers is not restricted to server provisioning alone. It encompasses various operational tasks, including:

• The capability to provision infrastructures therefore reduces deployment time from hours to minutes in case of standardized workloads.
• Network automation can help apply settings to many routers and switches.
• The ability to orchestrate workloads that allows moving applications from one resource to another based on demand.
• Automated patching enables performing updates on thousands of systems at the same time.
• Asset monitoring is in charge of checking devices’ performance, available capacity and lifecycle.

This operational value is not achieved by the automation of isolated operations but rather done through the interconnection of different operations.

The Future of Data Centers

In the current scenario, data center automation is becoming vital for increasing infrastructure, more dynamic workloads, and the need for operational teams to handle increased complexity of operations. In this context, automation is not going to eliminate the professionals in the industry but will change the way they spend their time.

As organizations start adopting AI workloads, edge computing, hybrid infrastructuring, and software-defined environments, the need for manual processes is going to reduce. The main benefit of automation for operations teams is measurable as a result of less repetitive actions, quicker response, and system visibility. The objective of automation is not to maximize its scope, but to implement moderate automation with the help of reliable controlling measures.

Valid operational models will provide a combination of automation with human oversight and management as well as clear policies, testing, and controlling measures.

# # #

About the Author

Raksha Sharma is a Jr. Team Lead at Market Intelo 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 Why Data Center Automation Is Becoming Essential for Modern Operations? appeared first on Data Center POST.

TL;DR

• AI infrastructure is shifting beyond centralized hyperscale environments as organizations adopt distributed architectures that better align infrastructure with workload requirements.
• NeoCloud and regional colocation work together to support AI workloads by connecting specialized compute with enterprise and cloud environments.
• Reliable, high-performance interconnection is critical to moving data efficiently and supporting AI applications across distributed environments.
• 1547 provides the regional infrastructure and connectivity needed to support scalable, interconnected AI deployments.

# # #

Written by iMiller Public Relations on behalf of fifteenFortySeven.

The landscape of artificial intelligence infrastructure is undergoing a significant shift from centralized hyperscale cloud environments to distributed architectures. While early AI initiatives relied heavily on centralized GPU clusters to train large models, modern enterprise deployments require planning that goes beyond raw compute capacity. As AI initiatives move into production, organizations must balance technical requirements such as performance, latency, cost, and data location. This has led to the adoption of distributed models that combine specialized computing, regional infrastructure, and high-performance connectivity.

Different AI workloads perform best in tailored environments rather than a single, one-size-fits-all platform. Large-scale training continues to depend on specialized GPU clusters, which can be accessed flexibly through NeoCloud providers to avoid the high costs of building dedicated hardware. Meanwhile, inference workloads are increasingly deployed closer to users in regional colocation facilities to minimize latency, while sensitive data is kept within private infrastructure to meet compliance and governance needs. For these environments to deliver value, specialized compute must operate as a highly connected ecosystem rather than in isolation.

In this distributed ecosystem, carrier-neutral connectivity and regional colocation have become strategic advantages rather than mere supporting services. Reliable interconnection allows data to flow efficiently between NeoCloud platforms, enterprise systems, public clouds, and edge deployments, reducing complexity and increasing flexibility. Industry trends reflect growing investments in regional markets that provide the necessary power density, cooling capacity, and geographic diversity to complement traditional hyperscale setups. Ultimately, the future of AI will be built on an interconnected infrastructure that places every workload in the environment where it performs best.

This content originally appeared on fifteenFortySeven’s website and has been adapted for syndication on Data Center POST. Read the complete article here.

The post Distributed AI Infrastructure: Why AI Won’t Live in One Data Center appeared first on Data Center POST.

TL;DR:

• AI raises the standard for human effort rather than making jobs obsolete. As routine tasks are automated, uniquely human skills like judgment, creativity, leadership, and deep technical knowledge will become increasingly valuable.
• Much like the iPhone established the platform for mobile businesses like Uber and Airbnb, current AI infrastructure will serve as a launchpad for entirely new products, experiences, and technical challenges.
• While public concerns over data center energy use, water consumption, and noise are valid, the industry is actively developing engineering solutions, such as more efficient cooling systems and cleaner power sources, to reduce environmental and community impacts.

# # #

By Ishaan Gupta

Today, many fear that AI will make their jobs obsolete, while businesses and consumers often use the technology without understanding the engineering and infrastructure that enable it. Those fears overlook a larger reality: AI is raising the standard for human effort, not replacing it.

As artificial intelligence automates more routine tasks, workers will increasingly need to develop higher-level technical, creative, and leadership skills. Understanding that transition also requires understanding the infrastructure behind AI and the engineers building it.

The Infrastructure Behind the AI Revolution

As AI infrastructure expands, I believe the conversation has become too focused on fear and not enough on how the technology is actually being built. AI can raise the standard for human work rather than eliminate the need for human workers. Data centers, while resource-intensive today, can also become more efficient as the technology matures.

As AI workloads grow more complex, they often require dozens of machines working together in a single data center. Keeping those systems communicating efficiently and securely is one of the major engineering challenges behind modern artificial intelligence.

Embedded software plays an important role in that process. It is the foundational code that helps power the hardware behind AI systems and large language models. Making that infrastructure faster, more reliable, and more cost-effective is essential as AI workloads continue to operate at increasingly large scales.

Building AI Infrastructure at Scale

Building software at scale requires infrastructure that can operate continuously and reliably, including systems that can receive software updates without interrupting services customers are actively using.

As AI infrastructure grows more complex, developing the skills of the next generation of engineers becomes equally important. Technology is advancing rapidly, and the technical expertise required to build and maintain it must keep pace.

AI as a Platform, Not a Replacement

I believe that AI should be viewed as a platform rather than merely as replacement technology.

Consider what happened when the iPhone was introduced. The device did more than change mobile phones. It created an infrastructure for innovation in mobile computing, and businesses such as Airbnb, Uber, and Square were eventually built on top of that platform.

AI may be entering a similar stage.

The infrastructure being built today can support products and experiences that have yet to be developed. As routine tasks become automated, human judgment, creativity, technical expertise, and leadership become more valuable, not less.

That does not mean the growth of AI infrastructure comes without challenges.

Public concerns surrounding data centers are valid. Energy use, water consumption and noise are real issues, but they are also engineering problems the industry is actively working to solve. More efficient cooling systems, cleaner power sources, and improved infrastructure can reduce the environmental and community impacts of data centers as technology develops.

Even if today’s AI infrastructure boom plateaus over the next three to five years, I expect that infrastructure to create room for another wave of advancements and refinements. We have seen this progression with other technologies, from early computers to modern smartphones. Technology does not simply stop developing. It adapts.

Looking Beyond Today’s AI Boom

The next stage of AI will increasingly focus on the products and experiences built on top of the infrastructure being developed today.

That shift will create new technical challenges and new opportunities for engineers, businesses, and workers. It will also require people to adapt. Some routine work will undoubtedly become automated, but that places a greater premium on the skills machines cannot simply replace: judgment, creativity, leadership and deep technical knowledge.

For engineers, that also means sharing knowledge and mentoring those entering the industry. Investing in the development of younger engineers benefits both the individuals building their careers and the companies relying on their contributions.

The AI revolution is still developing, and so is the infrastructure supporting it. The challenges surrounding data centers, sustainability, and automation should not be dismissed. They should be solved.

The next era of AI will depend not simply on what machines can automate, but on what people build on top of them.

The post AI Is Raising the Standard for Human Work, Not Replacing It appeared first on Data Center POST.

TL;DR

• Backup generator updates are driven by rising data center power demands, escalating maintenance costs, or tightening emissions regulations.
• Regulatory compliance introduces significant costs and complexity, as modern upgrades often require advanced emissions-control systems like diesel particulate filters.
• System replacements are highly complex, impacting fuel storage, controls, pumping, and monitoring infrastructure beyond just the generator itself.
• Early expert collaboration with integration specialists helps prevent design pitfalls, manage budgets, and ensure seamless system integration.

# # #

Originally posted on Bisnow.

Data center backup generators can run reliably for years with proper maintenance, but eventually require retrofits or replacements when power needs outgrow original capacity, repairs become too expensive, or emissions regulations tighten. In many older facilities built with low power density, the existing backup infrastructure no longer aligns with current or future operational demands. However, resolving these issues is rarely as straightforward as swapping out an old unit for a newer, more powerful model.

Upgrading or replacing generators involves complex environmental compliance, as federal, state, and local regulations often require advanced emissions-control technologies like diesel particulate filters and selective catalytic reduction. These regulatory demands can significantly increase both the capital and operating cost of a retrofit project. To navigate these hurdles, facilities can operate a mix of older permitted units and newer compliant generators, provided the overall site remains within legal emissions limits. This approach requires carefully balancing compliance, budget constraints, and ongoing operational resiliency.

A successful generator transition affects multiple interconnected systems, including fuel storage, pumping, controls, and monitoring systems. Integrating these complex elements typically requires collaborating with specialized fuel system and generator integration experts early in the design phase to identify potential pitfalls and prevent costly mistakes. Proactive planning, combined with structured preventative maintenance, ultimately extends equipment life and controls long-term operating costs.

To read the original article, please click here.

The post Time For A Change? What Data Centers Must Consider To Retrofit A Generator appeared first on Data Center POST.

TL;DR

• Growing public opposition and NDAs have created an information vacuum, leading to the ousting of local officials in April 2026 and nearly 150 proposed or enacted data center moratoriums across the United States.
• Data center developers should adopt this transparent model by addressing community complaints early and head-on with clear, factual parameters.
• Developers can successfully build trust without compromising proprietary secrets, as the public is primarily concerned with local environmental resources like water, heat, and power usage.

# # #

Written by iMiller Public Relations on behalf of The Tech Capital.

Data center developers are facing rapidly intensifying public opposition and political backlash due to concerns over environmental impacts and a widespread lack of transparency. Strict non-disclosure agreements and developer silence have created an information vacuum, fueling public resentment and misinformation. This has led to dramatic political consequences, such as voters ousting pro-data center politicians in Missouri’s April 2026 elections and nearly 150 municipalities proposing or enacting moratoriums.

To overcome this hostility, developers can learn from the public engagement strategy of the U.S. Nuclear Navy led by Admiral Hyman Rickover. Despite severe post-WWII public anxiety surrounding nuclear energy, the Navy built lasting public trust through policies of cradle-to-grave responsibility, voluntary disclosures, and direct community outreach by engineers. This battle-tested approach proves that proactively addressing difficult questions and treating the community as a respected partner is highly effective for winning public support.

Applying this strategy requires developers to engage with local communities early in the scoping process and address criticisms head-on. This “Intensive Honesty” does not require exposing proprietary business secrets, as the public’s primary concerns are operating parameters like water, heat, and electricity usage. By offering upfront, fact-based details and clear explanations, developers can establish community buy-in and replace fear with contextualized understanding.

This content originally appeared on The Tech Capital’s website and has been adapted for syndication on Data Center POST. Read the complete article here.

The post What Data Centre Developers Can Learn From The Nuclear Navy appeared first on Data Center POST.

TL;DR

• While external disruptions trigger delays, the underlying reasons 90% of projects run late are internal issues like optimism bias, weak governance, and fragmented accountability.
• Modern hyperscale data centers are too complex for traditional, spreadsheet-based tracking, which often obscures critical delays until recovery options are limited.
• Organizations need to govern time as strictly as capital, introducing centralized leadership and rigorous schedule controls to protect project timelines.

# # #

Written by iMiller Public Relations on behalf of Foresight.

Approximately 90% of data center projects suffer from schedule overruns, with average delays approaching 20% to 30%. While external bottlenecks like power availability, permitting, and supply chain disruptions are commonly blamed, research indicates that the true root causes are internal, driven by optimism bias, weak governance, fragmented accountability, and ineffective risk management. As hyperscale data center builds grow increasingly complex with thousands of interconnected activities, traditional project management tools like spreadsheets fail to surface schedule slippage until it is too late to easily recover.

To combat these systemic delays, organizations must treat time with the same strict governance and discipline typically reserved for financial capital. Implementing a “Chief Time Officer” model can bring the necessary executive accountability, controls, and forecasting rigor to schedule performance. By establishing a single source of truth, focusing reviews on active decision-making rather than data reconciliation, and using early warning indicators, leadership can significantly reduce the gap between a problem emerging and the organization’s response.

For instance, one global data center portfolio that replaced fragmented reporting with unified schedule intelligence successfully slashed reporting overhead by 90% and estimated project delays by 20%, saving over $100 million in potentially lost revenue.

This content originally appeared on the Foresight website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post Why 90% of Data Center Projects Still Run Late appeared first on Data Center POST.

TL;DR

• AI’s rapid infrastructure expansion is generating a massive, overlooked volume of single-use packaging waste used to protect servers and chips during transport.
• Hyperscale data centers often discard packaging immediately because tight installation timelines make immediate disposal easier than managing recovery on site.
• The future of sustainable AI must look beyond energy and cooling efficiency to responsibly manage the heavy physical supply chains moving materials every day.

# # #

Artificial intelligence is reshaping nearly every industry, from healthcare and finance to manufacturing and logistics. But behind the rapid growth of AI lies a physical infrastructure of immense scale. Data centers are expanding rapidly to support increased computing demand, and with that growth comes a massive environmental footprint that extends far beyond energy consumption alone.

Most discussions about AI’s environmental impact focus on electricity use, cooling systems and water consumption. While these are critical concerns, the enormous volume of packaging waste generated by AI infrastructure expansion is another challenge quietly taking shape in parallel with these issues. Servers, networking equipment, chips and supporting hardware all move through complex global supply chains protected by crates, corrugated materials, foam systems and specialized transport packaging. In many cases, that packaging is used once and discarded immediately.

As AI deployment accelerates, this issue is intensifying. Hyperscale data center environments operate under strict timelines where speed is the priority. Equipment is installed quickly to keep pace with demand, which leaves little time for packaging recovery or reuse decisions on site. Once servers are unpacked, materials are often treated as waste simply because disposal is easier than recovery.

The scale of this opportunity is significant. According to the U.S. Environmental Protection Agency, containers and packaging account for more than 28 percent of municipal solid waste in the United States. At the same time, global packaging waste continues to rise alongside growing e-commerce, manufacturing and technology expansion. AI infrastructure growth now adds another layer to that challenge.

Unlike some sustainability issues tied to data centers, packaging presents a rare opportunity for measurable environmental improvement without requiring major changes to computing infrastructure itself. Packaging systems can often be redesigned, recovered, reused or recycled with little to no disruption to business operations.

This is where circular packaging systems become paramount. Instead of treating packaging as disposable, organizations can treat it as a recoverable operational asset. Durable transport crates can be reused across multiple shipping cycles. Foam components can be separated, reprocessed and reintroduced into future packaging applications. Corrugated materials can move back into recycling streams and return as molded pulp or other protective packaging formats.

Closing the loop in packaging systems not only reduces waste but also improves material efficiency and minimizes reliance on raw materials to ensure greater supply chain resilience amid ongoing global volatility. Leading recovery programs divert up to 98% of waste from landfills by scaling material recapture, refurbishment and recycling initiatives.

Circular systems are most successful when they fit seamlessly into fast-moving environments. In AI and data center operations, sustainability initiatives must be designed for day-to-day operations without creating delays and complexity.

In practice, this means creating simple recovery pathways directly within existing logistics processes wherein packaging can be collected immediately after installation, separated into reuse and recycling streams, inspected for refurbishment potential and warehoused for future deployment. Materials that cannot be reused can still be processed into future packaging inputs rather than entering landfill streams.

The companies that will see the greatest return on these initiatives are those that recognize packaging as part of their broader resource management strategy, rather than viewing it as a low-value shipping necessity. Making this shift requires greater collaboration between sustainability teams, logistics leaders and operations managers who historically have not always shared ownership of packaging decisions.

With the ongoing rapid expansion of AI infrastructure, environmental responsibility will increasingly be measured not only by how efficiently data centers operate, but by how responsibly the surrounding supply chains function. Packaging may not be the largest sustainability challenge facing the AI sector, but it is one of the most practical areas where organizations can achieve immediate progress.

The future of sustainable AI infrastructure will extend beyond renewable energy and efficient cooling systems to how effectively industries close the loop around the materials already moving through their supply chains every day.

# # #

About the Author

Paul Budsworth is the chief executive officer of Specialized Packaging Group (SPG). Since 2022, he has led the company’s growth into one of North America’s largest independent protective packaging providers. For more information, visit www.spgpackaging.com.

The post AI’s Rapid Expansion and How It Impacts the Environment appeared first on Data Center POST.

TL;DR

• Modern IT strategy has shifted from choosing between “cloud or on-premises” to identifying where specific workloads live, leading to a deliberate blend of public cloud, colocation, private capacity, and edge environments.
• The return of private infrastructure is fueled by the data-intensive nature of AI inference, strict data sovereignty regulations, and cloud cost management.
• Modern private infrastructure is not a massive, capital-heavy data center or an informal server room; it is a compact, properly engineered environment run to standardized operational frameworks.
• Infrastructure projects fail when treated solely as hardware purchases; long-term reliability depends heavily on the physical engineering environment and documented operational procedures to minimize human error.

# # #

AI, cloud cost discipline, data sovereignty and edge workloads are pushing businesses to rethink what belongs in the public cloud and what needs a controlled data center environment.

For most of the last decade, the default infrastructure move was to push more workloads into the public cloud. For a lot of companies that was the right call. Cloud gave them speed, elastic capacity, managed services and a way to avoid heavy capital spending on physical infrastructure. None of that has stopped being true; the public cloud remains essential.

What has changed is the question companies are asking. It used to be “cloud or on-premises?” Increasingly it is “which workloads belong where?” In the AI era that question carries real strategic weight. Workloads are getting more data-intensive, more latency-sensitive and more power-hungry, while cost control, data sovereignty, resilience and operational predictability have all climbed the priority list. For most companies the answer is not to build a large data center of their own. It is that private infrastructure has re-entered the conversation, for reasons quite different from the ones that drove it a decade ago.

Private infrastructure no longer means a massive data center

When business leaders hear “our own data center,” many still picture a large capital project: a building, hundreds of racks, heavy engineering, multi-year construction and a dedicated operations team. That model still exists, but it is no longer the only one.

Private infrastructure today can mean a small server room built to a real standard, a modular or micro data center, a local edge node, or a hybrid setup where only specific workloads stay under direct control. Size is not the point. What matters is whether the environment is properly designed, powered, cooled, secured and monitored, and whether it is run to a recognized operational standard or framework such as EN 50600, ANSI/TIA-942 or Uptime Institute’s operational reliability guidance, rather than improvised.

In my own work on infrastructure projects across emerging digital markets, I keep running into the same misconception: that “private infrastructure” means going back to the old world of inefficient server rooms. It does not. I have seen organizations start out treating private infrastructure as if it simply meant putting a few servers in an office room and calling it a data center. In practice the modern alternative looks nothing like that: a compact but properly engineered environment, built and run to a real standard, with redundant power, controlled cooling, fire protection, access control, monitoring, documented maintenance and clear operational ownership. In that model “private” does not mean informal or outdated infrastructure. It means a controlled, standardized, purpose-built environment for the workloads that genuinely need to sit closer to the business. The real question is whether the business needs tighter control over a specific set of workloads.

AI Changes the Economics of Placement

AI is a large part of why placement now matters. Training big models stays concentrated in specialized, high-density environments, but inference is spreading out toward applications, users, factories, branches, hospitals, banks and public-sector systems. These workloads do not all belong in the same place. Some need the scale and managed services of hyperscale cloud. Others need low latency, predictable performance, local data processing, or tight control over sensitive datasets, and for those, private or edge infrastructure becomes part of the architecture rather than an exception to it.

The power math reinforces this. The IEA’s 2026 update projects electricity demand from data centers roughly doubling from about 485 TWh in 2025 to around 950 TWh by 2030, with AI as the main driver. That pressure does not land only on hyperscalers; it changes how every enterprise has to think about power availability, workload placement and long-term capacity planning. As compute demand climbs, the cheapest or fastest deployment path stops being automatically the right one, and the strongest architectures increasingly blend public cloud, colocation, private capacity and edge.

Cloud Cost Control is Also Changing the Conversation

Cost is the second driver. Cloud is powerful, but it is not automatically efficient, and as estates grow, governing the spend gets harder than most teams expect. Flexera’s 2026 State of the Cloud report found that estimated wasted cloud spend rose to 29%, reversing a five-year decline, and attributed the uptick directly to AI workloads and the complexity of new IaaS and PaaS pricing. That is not an argument against cloud; it is an argument for putting each workload in the right home. Elastic, unpredictable workloads belong in the cloud. Stable, long-running, data-heavy ones often get better cost visibility and operational control in a private or hybrid model, especially where the business needs predictable performance over time. Cost depends less on the label and more on workload behavior. The real question is whether the architecture fits the workload profile.

Sovereignty and Control Matter More Than Before

Data sovereignty is the third reason private infrastructure is back on the table. Across many regions, companies and public institutions need to know exactly where data sits, who can reach it, and which jurisdiction governs it. That matters most for financial institutions, healthcare, telecom operators, public-sector systems and critical infrastructure. Some organizations can meet those requirements through cloud regions, contractual controls and encryption. Others, particularly in markets where local cloud regions are thin or regulation is still evolving, need a controlled private or hybrid environment.

This is an operational question as much as a legal one. Beyond compliance, businesses want to know who actually controls the infrastructure, how incidents are handled, how backups are managed, and how fast systems can be recovered.

In a number of infrastructure discussions in Central Asia, technical capability was only one part of the decision. Organizations also had to show where sensitive data was actually stored, who administered the environment, how access was audited, and how recovery would work in an incident. For banks, telecom operators and government systems, those questions often become decisive. A private or hybrid architecture can give an organization a clearer operational boundary: critical data and systems stay under local control, while less sensitive or more elastic workloads can still use public cloud services.

The Weak Point is Often Not the Server but the Environment

Private infrastructure projects fail when companies treat them as an IT hardware purchase instead of a data center decision. Servers are only one part of the system. The environment around them is what makes the difference: reliable power, proper cooling, physical security, fire protection, monitoring, access control, maintenance procedures and clear ownership. Strip those out and a “private cloud” is just another fragile server room with a better label. Many teams underestimate exactly this: they focus on compute, storage and networking and treat the engineering and operational layer as secondary. That is a costly mistake.

Uptime Institute’s outage research keeps pointing to the same conclusion: human error and weak procedures remain central to data center reliability. For smaller private and edge sites, that lesson is especially important. They still need documented procedures, monitoring, maintenance schedules, escalation paths and trained people. Frameworks like ANSI/TIA-942 and Uptime Institute’s operational reliability guidance exist precisely to make that discipline repeatable. The smaller the site, the less room there is for improvisation.

Emerging Markets Have a Specific Opportunity

This plays out vividly in emerging digital markets such as Central Asia, where organizations are often building digital services, adopting AI tools, modernizing enterprise systems and meeting new data-localization expectations all at once. Public cloud is part of the answer there, but rarely the whole answer. Designed correctly, private infrastructure can carry local workloads, reduce dependence on distant facilities, tighten control over sensitive data and lay a foundation for hybrid AI and edge deployments.

One pattern I see often in the region is that companies start their infrastructure modernization from the application or hardware side and underestimate the engineering environment around it. They invest in servers, storage or AI-ready equipment while power redundancy, cooling capacity, monitoring, access control and operational procedures stay underdeveloped. The better approach is to treat even a small private site as a data center environment from day one: standardized, monitored, documented and built for future hybrid integration, rather than as an isolated server room.

Seen this way, private infrastructure is not a step backward for these markets. It is part of a more mature hybrid architecture.

The Decision Should be Workload-Driven

The future is not “everything in the cloud” or “everything on-premises.” It is workload-driven. Public cloud stays the best option for rapid development, elastic demand, global services, managed platforms and speed of innovation. Colocation keeps its place where companies want professional facilities without owning the full stack. Private infrastructure earns its place where control, latency, sovereignty, predictable cost or local resilience matter most.

The data already shows companies acting on this rather than just talking about it. Hybrid is now the dominant pattern: Flexera puts 73% of organizations on hybrid estates, yet only 23% of cloud-based workloads and 23% of cloud-based data have been repatriated, suggesting that migration and repatriation are increasingly happening side by side as teams place each workload more deliberately. That is the real shift. Companies are becoming more deliberate about matching each workload to the environment that fits it.

The companies that get this right will not be the ones chasing a slogan. They will be the ones that classify their workloads honestly and design infrastructure around real requirements: performance, control, cost and resilience. Private infrastructure is back because AI, edge, sovereignty, cost discipline and resilience are forcing a more careful answer to a single question: where should this particular workload actually live?

# # #

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 hybrid infrastructure, with over 20 years of experience in IT infrastructure projects across Central Asia and Eastern Europe.

The post Private Infrastructure Is Back, but Not for the Old Reasons appeared first on Data Center POST.

TL;DR

• AI and cloud demand are expanding Latin America’s digital infrastructure opportunity while increasing requirements for power, connectivity and scalable capacity.
• Debt and equity markets are adapting to larger data center developments with greater attention to execution risk, tenant demand, and long-term viability.
• Power availability and grid access are increasingly influencing where projects move forward and where capital is deployed.
• Brazil, Mexico, Chile, and Argentina offer distinct opportunities but policy, incentives, and infrastructure readiness will shape the pace of development.

Latin America’s digital infrastructure market is entering a new phase as AI demand, larger data center projects, and growing international investment reshape how and where capital is deployed. These themes took center stage at the LATAM Finance Forum 2026, held July 30 in São Paulo, Brazil where leaders from investment, technology, digital infrastructure, energy, and public policy examined the factors influencing the region’s next phase of growth. 

Across the Forum, discussions reflected a market with significant potential but increasingly complex requirements for turning demand into delivered capacity. Capital remains interested in the region, but power availability, government policy, infrastructure readiness, and execution are becoming more important as projects increase in size and complexity.

AI Is Reshaping LATAM’s Infrastructure Opportunity

AI is adding a new dimension to Latin America’s digital infrastructure growth. Global technology companies evaluating the region increasingly require more than available land and conventional capacity with energy, connectivity, high-density infrastructure, and operational resilience becoming critical considerations.

During a discussion examining what global technology leaders need to scale across the region, Victor Arnaud, President, Brazil at Equinix, joined industry leaders to explore how AI workloads are changing data center design, operations, and supply chains. The conversation also addressed the investments in energy, connectivity, and compute infrastructure that will be necessary for Latin America to remain competitive as global AI infrastructure strategies evolve.

The scale of the opportunity was also examined through the economics of AI mega-campuses where significantly larger power and infrastructure requirements introduce new questions around financing and risk. Jeff Ferry, Partner & Head of Digital Infrastructure at Safanad, participated in a discussion exploring what makes these projects bankable and how responsibility for power, connectivity, and other enabling infrastructure can be shared across operators, investors, lenders, technology companies, utilities, and governments.

Providing broader market context, Hadassa Lutz, Senior Consulting Analyst LATAM at Structure Research, delivered a data-led keynote examining the evolution of Latin America’s data center market. Her presentation addressed demand trends, supply dynamics, market maturity, and the factors influencing investment decisions as the region prepares for its next stage of development.

Capital and Risk Evolve With Project Scale

As data center developments grow larger, financing strategies are evolving alongside them. Debt discussions at the Forum analyzed whether domestic capital markets can support the scale of future projects or whether international banks, development finance institutions, and private credit providers will assume a greater role.

Rafael Bomeny, Co-Founder & CFO of ODATA – An Aligned Data Centers Company, joined finance leaders to examine how operators and lenders are approaching larger projects and increasing power requirements. The discussion addressed emerging financing structures and how lenders are assessing factors including power availability, tenant demand, construction risk and regulatory exposure.

Equity investors are also evaluating opportunities across platforms, individual projects, and the broader data center value chain. Discussions around M&A, co-investment, and the monetization of stabilized assets demonstrated that capital remains available, but investors are increasingly focused on credible demand, infrastructure readiness, and a clear path to project execution.

Power and Policy Will Determine Where Growth Happens

Power availability is rapidly becoming one of the most important competitive differentiators across Latin America. As campus sizes move from tens of megawatts toward 100MW, 1GW, and beyond, access to generation and transmission infrastructure is increasingly influencing site selection, project timelines, and capital allocation.

Marcelo Mendes, Head of LATAM at Ada Infrastructure, joined energy and infrastructure leaders to examine how the region can convert its energy resources into digital infrastructure investment. The discussion explored renewable energy procurement, grid access, transmission capacity, and the role of utilities, generators, and policymakers in supporting large-scale data center growth.

Policy will be equally important because the region’s markets offer different combinations of opportunity and execution risk. Brazil’s ReData initiative and AI regulatory environment, Mexico’s grid and permitting challenges, Chile’s sustainability requirements, and Argentina’s efforts to attract AI infrastructure investment demonstrate why regional demand alone cannot determine where capital ultimately flows.

These market differences reinforce the importance of evaluating Latin America country by country rather than as a single digital infrastructure market. Investors and operators must consider how energy, regulation, incentives, and infrastructure readiness align with customer demand before committing capital at scale.

Converting Demand Into Delivered Capacity

LATAM Finance Forum 2026 demonstrated how closely connected AI, capital, energy, and policy have become as Latin America’s digital infrastructure market expands. The opportunity is significant, but converting growing demand into operating capacity will require coordination across the financial, technology, energy, and public sectors.

As AI increases both the size and urgency of future infrastructure requirements, the markets best positioned for growth will be those that can align capital with reliable power, supportive policy, and effective project delivery. How successfully the region addresses those challenges will help determine Latin America’s position within the next wave of global digital infrastructure investment.

To learn more about The Tech Capital and future events, visit the official Tech Capital website:https://events.thetechcapital.com/.

The post Capital, Power, and AI Reshape Latin America’s Digital Infrastructure Market appeared first on Data Center POST.

TL;DR

• Lightpath has entered the Greater Atlanta market with high-capacity Managed Bandwidth services.
• Wavelength and IP Transit services are now available across 30 Atlanta-area data center campuses.
• Customers can access capacity from 10 Gbps to 800 Gbps per wavelength, with aggregate system capacity scaling into multi-terabit configurations.
• Lightpath’s Atlanta long-haul network hub connects to New York, Philadelphia, Ashburn, Chicago, Dallas and Miami.

Atlanta has emerged as one of the fastest-growing data center markets in the country, with enterprises and hyperscalers requiring greater capacity to support AI workloads, cloud interconnection and large-scale data movement.

Lightpath is expanding its network footprint to meet that demand, entering the Greater Atlanta market with high-capacity Managed Bandwidth services across 30 data center campuses.

The expansion brings Lightpath’s Wavelength and IP Transit services to Atlanta-area customers, with capacity ranging from 10 Gbps to 800 Gbps per wavelength. Aggregate system capacity can scale into multi-terabit configurations, providing the high-capacity optical transport required by increasingly bandwidth-intensive workloads.

The new market also extends the reach of Lightpath’s national backbone. Lightpath operates a long-haul network hub in Atlanta with connections to New York, Philadelphia, Ashburn, Chicago, Dallas and Miami. High-capacity managed connectivity is delivered over 100% underground fiber infrastructure to the 30 data center campuses served in the Atlanta market.

“Atlanta is one of the fastest-growing data center markets in the country, and demand for high-capacity managed bandwidth continues to accelerate,” said Chris Morley, CEO of Lightpath. “Entering this market connects this growing ecosystem to our national backbone, giving customers a direct path to the intensive capacity that AI and cloud workloads require.”

The Atlanta expansion is part of Lightpath’s broader metro expansion strategy, focused on entering and densifying major U.S. economic markets. It also strengthens the company’s presence in a market that is becoming increasingly important to AI and cloud growth.

For organizations operating within Atlanta’s growing data center ecosystem, the combination of local data center connectivity and access to Lightpath’s long-haul backbone provides a way to scale optical transport as workload requirements increase.

“Our customers need optical transport that scales with their workloads,” said Tim Haverkate, Chief Commercial Officer at Lightpath. “By pairing our long-haul backbone with high-capacity managed Wavelength and IP Transit services, we give Atlanta-area customers a single, trusted path to the bandwidth their businesses demand.”

With its entry into Atlanta, Lightpath continues to expand the reach of its AI-grade, all-fiber network, which now spans 12,100 route miles across major U.S. metro markets and supports high-capacity connectivity for cloud, hyperscale, data center and mission-critical enterprise workloads.

Read more in the press release here. 

The post Lightpath Expands into Atlanta with High-Capacity Managed Bandwidth Services appeared first on Data Center POST.

TL;DR

• Earnings release: Duos will report second quarter 2026 results after market close on Monday, August 17. 
• Investor call: Company management will host a conference call on Monday, August 17, at 4:30 p.m. Eastern Time.
• Recent progress: Duos recently announced a five-year, 10 MW colocation agreement at its Columbus, Georgia campus.

# # #

Duos Technologies Group, Inc. (“Duos” or the “Company”) (Nasdaq: DUOT), a provider of modular colocation, AI infrastructure, and communications hub solutions, has scheduled its second quarter 2026 earnings call for Monday, August 17, 2026, at 4:30 p.m. Eastern Time.

Based in Jacksonville, Florida, Duos is focused on scaling its digital infrastructure platform through its Duos Edge AI and Duos Technology Solutions subsidiaries. The Company supports high-power AI and enterprise computing deployments across Tier 3 and Tier 4 markets.

During the call, Duos management will discuss financial results for the quarter ended June 30, 2026, followed by a question-and-answer session. The Company will release its financial results after market close on Monday, August 17, through the Investor Relations section of its website.

The call follows Duos’ recent five-year, 10 MW colocation agreement at its Columbus, Georgia campus, valued at more than $111 million in contracted revenue. The agreement supports the Company’s continued focus on scaling owned AI infrastructure and long-term recurring revenue.

Investors joining by phone can use the U.S. dial-in number +1 877-407-3088 and confirmation number 13761911. International participants can access the call through the Company’s dial-in matrix, and a replay will be available in the Investor Relations section of the Company’s website following the event.

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

The post Duos Technologies Group, Inc. Shares Q2 2026 Results Amid AI Infrastructure Momentum appeared first on Data Center POST.

Modular construction is having a moment. As AI infrastructure demand pushes developers to build data centers faster than ever, “modular” has become one of the most-used words in the industry. It shows up in press releases, RFPs, and investor decks, often applied to very different delivery approaches.

That’s a problem. When a term gets used to describe everything from a preassembled electrical skid to a fully manufactured building, it stops being useful. Owners comparing “modular” options may be comparing approaches that differ enormously in schedule impact, quality control, and risk, without realizing it.

Understanding these differences matters more now than it did five years ago. AI-driven data center demand has compressed timelines to the point where the construction method itself has become a competitive variable. Getting the terminology right isn’t academic. It shapes how quickly a facility can come online and how predictable that timeline will be.

Two Different Things Called “Modular”

Most off-site construction falls into one of two categories, and the distinction is bigger than it sounds.

Traditional prefabrication manufactures individual components or assemblies away from the site and ships them in for installation. Electrical skids, mechanical skids, pipe racks, generator packages, and preassembled electrical rooms are common examples. This work reduces labor on site, but the project is still primarily built in the field. Prefabrication delivers the pieces; the project team still assembles them.

Volumetric modular construction starts from a fundamentally different premise. Instead of manufacturing individual parts, entire building sections, complete with structural systems, electrical infrastructure, mechanical systems, fire protection, controls, and interior finishes, are built in a factory. Modules arrive on site substantially finished and are set into place to form the completed facility. The focus shifts from building to installation and integration.

Both methods qualify as “off-site construction.” Only one fundamentally changes how a project is built.

Why the Percentage of Work Moved Off-Site Is the Real Metric

If there’s one number that separates these approaches, it’s how much of the total construction labor actually leaves the project site.

Traditional prefabrication typically moves 10% to 30% of construction labor off site. The majority of the work, including sequencing, coordination, and finishing, still happens in the field.

Volumetric modular construction can move 60% to 90% of that labor into a factory setting. By the time modules reach the site, most of the building is already complete.

That gap matters because every hour of work performed in a controlled manufacturing environment is an hour removed from weather delays, labor shortages, site congestion, rework, and the schedule uncertainty that comes with coordinating dozens of trades in the field at once. A facility built with 20% off-site labor and one built with 80% off-site labor may both be called “modular,” but they carry very different risk profiles, schedules, and levels of certainty.

The Schedule Advantage Isn’t Manufacturing. It’s Overlap.

It’s tempting to assume the main benefit of modular construction is that factories build faster than field crews. That’s part of it, but the bigger driver is parallel execution.

Traditional construction is sequential by nature: site preparation, foundations, structure, building enclosure, MEP rough-in, equipment installation, and commissioning, largely in that order. Each phase depends on the one before it. Traditional prefabrication changes some of the inputs, but the sequence itself remains mostly intact.

Volumetric modular construction breaks that dependency. Site development and factory manufacturing happen at the same time. While a site team is pouring foundations, the structure can already be built, electrical systems installed, cooling systems integrated, and equipment tested, all in a factory hundreds of miles away. Months of work that would otherwise happen one after another instead happen side by side.

That overlap, not just the fact that manufacturing exists, is what allows project schedules to compress substantially.  It doesn’t just reduce time. It creates a more predictable path to deployment.

Quality Control and Workforce Effects Are Often Overlooked

Two other differences deserve more attention than they usually get.

The first is quality control. Field construction relies on inspections performed as work progresses, often under variable weather and site conditions. Manufacturing environments introduce a different model: repeatable production processes, factory testing, standardized installation methods, and far less exposure to weather. The result tends to be greater consistency from one module to the next, fewer variables during construction, and fewer surprises during commissioning.

The second is workforce impact. The construction industry is dealing with labor shortages across nearly every trade, and large field-based projects require sizable crews working simultaneously on a single site. Manufacturing shifts a meaningful share of that labor into a fixed facility, where productivity tends to be higher, tools and conditions stay consistent, and training can be standardized across projects rather than rebuilt at every new site. That’s a more scalable labor model at a time when skilled trades are stretched across the industry.

Neither of these effects gets as much attention as speed, but both influence whether a facility performs as expected once it’s operating. Speed may get the headlines, but quality and predictability determine long-term success.

What This Means for Anyone Evaluating “Modular”

The practical takeaway is simple: When comparing off-site construction options, ask what percentage of the work actually happens away from the project site, and ask whether that work happens in parallel with site development or simply arrives as preassembled pieces waiting to be installed. Those two questions reveal more about a project’s real schedule and risk profile than the word “modular” ever will.

The question isn’t whether a project uses modular construction. The question is how much of the building still has to be built in the field.

At Mod42, this distinction drives how we approach data center delivery. We build volumetric modular facilities because moving as much of the building as possible into a controlled manufacturing environment is what actually changes the construction timeline, not just the components within it. We don’t start by asking how to prefabricate a traditional data center. We start by asking how much of the building can be completed before it ever reaches the site. Our aim isn’t to prefabricate pieces of a data center. It’s to manufacture the data center itself.

Traditional prefabrication and volumetric modular construction both improve on conventional building methods, just at very different scales. Prefabrication moves components off site. Volumetric modular moves the building itself off site. As AI infrastructure demand continues to compress timelines, understanding that difference will matter more, not less, to anyone trying to evaluate how a data center actually gets built.

Ready to see how volumetric modular construction is different? 

Connect with the Mod42 team to learn how factory-built volumetric modular data centers enable parallel execution, improve schedule certainty, and accelerate deployment. 

Get started today: https://www.mod42.com/quote. 

The post Not All Modular Is Created Equal appeared first on Data Center POST.

TL;DR

• How Dr. Ansar’s 15-year research at Oxford led to partnering with Igor Schifrin to co-found Foresight.works and introduce the concept of “time accounting” to match the rigor of financial CFOs.
• A step-by-step breakdown of how a six-month delay on a typical 100MW data center ($1 billion physical build) leads to a $100 million (10%) cost increase in lost revenue, labor, and SLA penalties.
• The platform’s ability to map design geometry to historical project databases, integrate with tools like Procore and Autodesk Construction Cloud, and utilize computer vision to verify progress and surface “bad news” early.
• Dr. Ansar’s outlook on de-biasing project communication, forging professional standards, and implementing AI and robotics to deliver infrastructure worldwide.

# # #

In the latest episode of NEDAS Live!, Episode 69 features a unique perspective on the data center industry with Dr. Atif Ansar, Executive Chairman and Co-Founder of Foresight. NEDAS Live! host, and CEO of iMiller Public Relations, Ilissa Miller explores how Dr. Ansar is redefining how we think about digital infrastructure, project management, and the crucial value of time in the AI-driven world.

Introducing the Chief Time Officer

Dr. Ansar’s journey began at Oxford University, where he spent over 15 years as an academic exploring why megaprojects consistently run late and exceed their budgets. Applying behavioral economics and analyzing large datasets, he uncovered a major blind spot in how organizations deliver large-scale infrastructure.

His research eventually led to a partnership with Igor Schifrin, a developer who had built nearly a million square meters of property around the world and had personally experienced the exact project delivery pain Dr. Ansar had been researching. Igor was attending one of Dr. Ansar’s Oxford executive education courses in 2018 when the two realized they needed to start a business to solve these challenges. Together, they founded Foresight as a way to help CEOs, boards, and organizations achieve greater certainty in budget and schedule delivery.

While organizations bring immense rigor to capital and cost control, time governance is often neglected. Dr. Ansar introduces the concept of the Chief Time Officer (CTO), a role dedicated to establishing standard operating procedures, strategic governance, and valuation metrics for time. Just as CFOs govern money, a Chief Time Officer brings equivalent rigor, diligence, and accountability to a project’s schedule.

Why Time is More Scarce Than Capital

This rebalancing of focus from money to time is particularly urgent for digital infrastructure. Historically, capital was the primary constraint, but today we are in a massive digital race.

Currently, there is over a trillion dollars worth of data centers under construction globally, representing 10% of all global construction, which exceeds both office buildings and retail development. When including the cost of computing chips, the total investment is three times larger. In this high-stakes environment, demand is blistering: data center vacancy rates have compressed to under 1%.

In a world where capital is abundant but time is extremely scarce due to rapid technological obsolescence, speed is the ultimate differentiator. There is no silver medal; you are either first or you are last.

Redefining Strategic Time Governance

Foresight acts as a digital Chief Time Officer, helping developers transition from emotional forecasting to data-driven delivery. The platform ingests data center designs, converts the physical geometry into a structured timeline, and enriches it with the industry’s largest database of past completed data center projects to right-size schedule durations.

Key features of the platform include:

• End-to-End Scheduling: Moving beyond physical construction tasks to account for design, permitting, power, community engagement, commissioning, and other dependencies where delays often emerge.
• Real-Time Progress Verification: Integrating with leading software like Procore and Autodesk Construction Cloud, and using computer vision and cameras to collect objective, on-site evidence.
• Early Warning Insights: Generating real-time insights to highlight risks and surface “bad news” early so developers can proactively address them.

Rather than viewing power and permitting delays as external “surprises,” Foresight helps developers anticipate, plan for, and eliminate these time risks entirely.

Looking Ahead: The Five-Year Vision for World-Class Delivery

When asked what “great delivery” looks like five years from now, Dr. Ansar outlines three key pillars that the industry must work toward:

1. De-Biased, Truthful Conversations: Overcoming the cognitive and political biases that lead people to hide delays out of fear. By utilizing data to establish transparency, teams can have honest, constructive discussions.
2. Professional Standards: Establishing industry-wide professional standards and codes of conduct for world-class delivery, drawing inspiration from how the medical profession established standards in the 19th century.
3. Technological Amplification: Embracing both AI and physical intelligence (robotics) to streamline construction, not to replace jobs, but to make infrastructure delivery faster and more affordable worldwide.

To continue the conversation, listen to the full podcast episode here.

The post The Digital Race and the Economics of Delay: A Conversation with Dr. Atif Ansar appeared first on Data Center POST.

TL;DR

• Transaction complete: Duos Technologies Group has completed the sale of its rail technology subsidiary, Duos Technologies, Inc., to Sandbank Acosta, LLC.     
• New chapter for DuosTI: The business will operate independently under the DuosTI brand, with Javier Acosta appointed President.
• Strategic focus: The sale enables Duos to concentrate its resources on AI infrastructure and communications hub platforms. 

Duos Technologies Group (Nasdaq: DUOT), through its former subsidiary Duos Technologies, Inc. (DTI), has completed the sale of the rail technology business to Sandbank Acosta, LLC. Following the transaction, the company will operate independently under the DuosTI brand, with Javier Acosta serving as President.

“I had the privilege of leading the commercialization and field deployment of the Railcar Inspection Portal across the United States, Canada and Mexico, and I am honored to return to lead this exceptional team into its next chapter,” said Javier Acosta, President of DTI. “As DuosTI, we return to our roots as a focused railroad technology company. With the largest installed base of inspection portals in North America, a deep patent portfolio and an unmatched image database, we are well positioned to expand the reach of our inspection services for our railroad customers and the broader industry.”

Acosta succeeds Adrian Goldfarb, who stepped down as President of DTI after overseeing the business and divestiture process since March 2026. The leadership transition supports DuosTI’s move into its next chapter as an independent company focused on the rail technology market.

“DTI built the technology that put Duos on the map, and its Railcar Inspection Portals remain the standard for AI-driven train inspection in North America,” said Doug Recker, Chief Executive Officer of the Company. “This transaction completes the strategic repositioning we announced earlier this year, placing the rail business with owners who are singularly focused on its growth while allowing Duos to dedicate its full capital and management attention to scaling our Edge Data Center and AI infrastructure platforms. We wish Javier and the DuosTI team every success and look forward to supporting a smooth transition.

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

The post Duos Technologies Group Completes Sale of Duos Technologies, Inc. to Sandbank Acosta, LLC appeared first on Data Center POST.

TL;DR

• To bypass slow grid interconnection queues and keep up with AI buildout timelines, hyperscale data center developers are moving to behind-the-meter gas generation, turning power procurement directly into fuel sourcing.
• Large 1 GW facilities can consume roughly 140 MMcf/d of natural gas, putting data centers in direct competition with utilities, industrial plants, and LNG projects for pipeline space and firm supply.
• Proximity to major digital loads is creating a new strategic premium for midstream pipeline assets while pushing regulators like FERC to update traditional planning structures.
• As fuel security becomes critical, developers are increasingly turning to energy recovery technologies to squeeze more useful power out of existing gas infrastructure without requiring additional fuel input.

# # #

As more data centers bypass the grid with gas-backed generation, they are no longer just competing for power. They are competing for fuel, pipeline access, and the infrastructure that makes power possible.

That changes the problem data center developers are trying to solve. Buying electricity from the grid means buying a delivered product. The utility or independent power producer handles fuel procurement, dispatch, balancing, and most of the infrastructure risk before power reaches the site. Behind-the-meter gas generation changes that model. Once a developer chooses to generate power onsite, power sourcing becomes fuel sourcing. Pipeline access, firm transportation, compression, storage, and backup supply move from the background to the center of the project.

The numbers are large enough to make this trend a market-changer. Energy technology company Enverus estimates that a 1 GW data center can consume roughly 140 MMcf/d of natural gas. At that scale, a hyperscale campus does not behave like a conventional electricity customer. It starts to resemble a major industrial fuel buyer. That pushes data centers into competition with utilities, industrial facilities, LNG projects, and other large gas users for pipeline space, firm supply, and strategically located infrastructure.

The growth trajectory of AI infrastructure is makes that competition consequential. Data center electricity demand is rising quickly, and developers are looking for ways to move faster than utility planning cycles and interconnection queues allow. Gas remains the most commercially available path to dispatchable generation at scale. For many projects, it is the only firm option that can be deployed fast enough to keep pace with AI buildout. That does not make gas-backed power simple or cheap. It does mean that data centers looking for speed-to-power are increasingly moving upstream into fuel strategy.

Recent projects show what that looks like in practice. Entergy’s agreement to support Meta’s Louisiana buildout includes seven new natural gas-fueled combined-cycle plants totaling more than 5,200 MW, along with major transmission and supporting infrastructure. In Ohio, Williams’ Socrates Power Solution Facilities are being developed to serve the growing data center market with 400 MW of natural gas-fired generation. They are examples of fuel access, generation, and digital infrastructure being planned together from the start.

That planning pressure is now reaching regulators. FERC has launched a targeted effort to speed large-load integration because traditional planning and tariff structures are not keeping pace with hyperscale demand. The issue regulators must address is how to manage a market in which some of the largest new customers are no longer waiting for delivered utility power and are instead building projects around direct fuel access and onsite generation. As data centers move upstream, regulators are being pushed to respond to a very different kind of load growth than the system was built to handle.

The same shift is affecting how gas infrastructure is valued. For decades, pipelines were valued primarily for moving molecules from one point to another. That is still their core role, but AI demand is adding a new layer of strategic value. The importance of a pipeline system now depends more heavily on where it connects, what kinds of loads sit nearby, and whether it can support behind-the-meter generation for energy-intensive campuses. A pipeline segment near a major data center project is no longer just transport infrastructure. It becomes part of that campus’s ability to operate, expand, and secure power on schedule.

That has implications for midstream economics. If large digital loads begin to shape where gas infrastructure gets expanded, how firm transport is contracted, and which assets command premium value, then AI is no longer just influencing electricity markets. It is changing the economics of fuel delivery itself. The companies best positioned near major data center corridors or capable of supporting new onsite generation models may benefit from a different demand profile than the midstream sector has historically served.

Once fuel becomes strategic, efficiency becomes strategic too. Developers and infrastructure owners will not only ask how to secure more gas. They will ask how to get more useful power out of the gas infrastructure already in place. That creates a stronger case for energy recovery technologies that convert pressure drops into electricity, improve onsite efficiency, and make existing gas systems more productive without requiring additional fuel input. In a market where both time-to-power and fuel security matter, getting more value from existing gas infrastructure becomes part of the competitive equation.

While many data center projects will continue to rely on conventional utility service, some of the largest digital loads in the economy are moving into a different category. They are no longer only buying electricity; they are competing for the fuel, contracts, and infrastructure that make electricity possible.

For data centers, that changes how power is sourced, how sites are selected, how regulators respond, and how gas infrastructure is valued. For years, the central question was where the next data center could find enough power. Increasingly, the more important question may be where it can secure enough fuel.

# # #

About the Author

Freddie is the Chief Executive Officer of Sapphire Technologies. Freddie has extensive experience in international business development and operations.

He joined Calnetix in 2018, heading up business development in the industrial, energy and aviation sectors. Before joining Calnetix, he was the Vice President of the Americas for Praxair Surface Technologies, responsible for the company’s business in North and South America. Prior to that, Freddie held the position of Vice President of Customer Service at Capstone Turbine Corporation.

He began his career at Rolls-Royce as a member of the Trent industrial gas turbine development team. Freddie holds a B.S. in Mechanical Engineering from McGill University in Canada.

The post What Happens When Data Centers Start Competing for Fuel? appeared first on Data Center POST.

TL;DR

• Fiber infrastructure development is shifting from broad geographic buildouts to focused investments driven directly by AI adoption, cloud platforms, and data center growth.
• Planning prioritizes high-utilization routes connecting enterprise campuses, cloud environments, and data center clusters to enhance path diversity, resilience, and scalability.
• Constructing networks in stages aligns capital resources with real-world customer adoption while preserving flexibility to adapt to evolving technologies and traffic patterns.
• Harbor Link is applying this disciplined, phased expansion strategy across Maryland, Washington, D.C., and Northern Virginia to support growing digital economy needs.

# # #

Written by iMiller Public Relations on behalf of Harbor Link.

Fiber network development is undergoing a fundamental transformation, moving away from broad geographic coverage toward targeted investments tied directly to measurable demand from AI adoption, cloud platforms, and data center expansion. Organizations are generating and consuming data at unprecedented rates, requiring infrastructure providers to evaluate expansion based on actual traffic growth and critical interconnection needs. Rather than deploying capital across wide geographic areas, modern planning concentrates resources on high-density corridors that tie together enterprise campuses, cloud ecosystems, and major data center hubs.

These high-utilization connection pathways serve as vital arteries for modern digital infrastructure, carrying massive traffic volumes and requiring enhanced path diversity, operational resilience, and long-term scalability. By prioritizing strategic routes that reflect emerging market requirements, infrastructure companies can optimize network performance while delivering essential flexibility for evolving enterprise requirements. This approach ensures that capital is deployed where it creates the highest operational and strategic impact.

To maintain flexibility amid rapid technological change and shifting market conditions, providers are increasingly adopting a phased construction model that aligns infrastructure rollouts with customer adoption. Initial segments establish a functional network foundation, while subsequent stages expand reach, boost capacity, and enhance resilience across key locations. Harbor Link demonstrates this disciplined strategy in practice by expanding fiber infrastructure across Maryland, Washington, D.C., and Northern Virginia to meet the Mid-Atlantic’s surging demand for robust digital connectivity.

This content originally appeared on the Harbor Link website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post How Demand Is Reshaping Network Development: Insights from Harbor Link appeared first on Data Center POST.

TL;DR:

• Sustainable infrastructure and operational reliability are not competing priorities; true resiliency means scaling to meet future needs without depleting resources.
• Bad fuel and poor emergency logistics are leading causes of backup generator failures during grid outages.
• ESI’s Uptime IQ software uses sensor data to simulate worst-case scenarios, helping operators optimize fuel logistics, predict needs, and justify operational expenditures.

# # #

As data centers expand at breakneck speeds to support AI, cloud, and edge workloads, the industry is facing a renewed focus on community impact and environmental responsibility. Yet, amid the rush to build the facilities of tomorrow, one critical, foundational element of operational transparency is often overlooked: backup power and fuel systems.

In the latest episode of NEDAS Live!, host Ilissa Miller sat down with Brittani Clarke Clayman, Director of Marketing and Sustainability at ESI Total Fuel Management, to discuss how operators can make smarter, more proactive decisions regarding their digital infrastructure deployments.

With over 30 years of experience serving the data center ecosystem, ESI takes a holistic approach to mission-critical infrastructure, proving that true reliability and sustainability are fundamentally intertwined.

Reliability and Sustainability: Two Sides of the Same Coin

A major misconception in the data center industry is that implementing sustainable solutions requires trading away reliability. ESI, which stands for Environmental Solutions International, was founded with the belief that resiliency is only truly reliable if it can endure for generations to come.

As Clayman explained, ensuring that infrastructure can scale and support future needs without depleting resources is the true definition of sustainability. Rather than competing priorities, reliability and sustainability act as two sides of the same coin, reinforcing one another to protect long-term uptime.

The Hidden Threats to Uptime: Bad Fuel and Logistics

When a grid outage occurs, operators immediately rely on their UPS systems and backup generators to maintain their service-level agreements. However, one of the leading causes of generator failure during these critical moments is bad fuel. Microbial growth or water accumulation inside fuel tanks can clog systems, preventing backup generators from starting when they are needed most.

To mitigate this risk, ESI encourages operators to utilize regular fuel polishing services and proactive testing to ensure their multi-million dollar generators will fire up at a moment’s notice.

Beyond the quality of the fuel itself, Clayman highlighted the massive logistical challenges of fuel management. Simply having fuel contracts with multiple providers is not enough. In an emergency, such as a major snowstorm, operators must know if supply routes are passable and if their facility will be prioritized over local hospitals, for instance. On top of that, managing the actual cadence of fuel trucks arriving at a site requires intense coordination to offload efficiently and safely without blocking roads for other emergency vehicles.

The HVO Revolution and Regulatory Hurdles

To meet net-zero goals, many operators are looking to transition to cleaner alternatives like Hydro-treated Vegetable Oil (HVO). Sourced from feedstocks like soybean or recycled cooking oil (and strictly avoiding palm oil to protect rainforests), HVO burns cleaner and offers a staggering 90% reduction in lifecycle greenhouse gas emissions. Notably, it also eliminates the typical diesel smell and reduces visible smoke, drastically improving environmental goals.

Even better, HVO can actually increase engine performance and resiliency due to its higher energy power.

However, regulatory frameworks have not caught up to digital infrastructure needs. Currently, the EPA’s renewable fuel standard considers the use of HVO in backup generators a “non-compliant use,” effectively penalizing data centers with credit fees for attempting to use a greener fuel. While exceptions exist for marine transportation, ESI is actively working with federal policy influencers to secure similar exceptions for the data center industry.

Data-Driven Resiliency: Uptime IQ™

To further optimize fuel operations, ESI developed a software program called Uptime IQ. This platform uses sensors to capture crucial data points, such as fuel temperature, volume expansion, and generator load speeds, to help operators fully understand their assets.

Uptime IQ allows operators to simulate worst-case scenarios, such as predicting exactly how much fuel is required and how many trucks must arrive per hour to maintain 96 hours of uptime during a simulated hurricane. This data not only secures operational resiliency but also equips facility managers with the concrete evidence needed to justify OpEx spending to higher-ups.

Plan for the Worst Case Scenario

As the industry scales, Clayman’s ultimate advice to operators is to proactively plan for the worst, even going so far as to imagine a “sharknado” hitting their facility. By securing a specialized partner to map out contingency plans before disaster strikes, operators can relax knowing their backup systems are ready, allowing them to focus on running the core of their business.

To continue the conversation, listen to Episode 68 of the NEDAS Live! podcast here.

The post How ESI Is Redefining Mission-Critical Resiliency and Fuel Logistics appeared first on Data Center POST.

TL;DR

• Rapid GPU compute density and terabit-bandwidth demands are pushing traditional short-reach copper cables to physical limits in reach, routing thickness, power consumption, and thermal management.
• While optical fiber addresses reach, using it for dense, short-reach GPU-to-GPU connections introduces exponential reliability issues while consuming roughly 10% of total compute cost.
• Point2 Technology offers an e-Tube solution using RF over polymer waveguide to transmit millimeter-wave signals through thin, flexible polymer tubes for 5-10 meter reaches.

# # #

By Sean Park, Founder, President and CEO at Point2 Technology 

GPU and XPU compute density is increasing dramatically, and interconnect bandwidth is rising faster than traditional data center wiring can support. As AI systems pack more devices into each rack and push link speeds into multi‑terabit territory, the physical infrastructure that connects these components is becoming a critical bottleneck. The copper cables that once served short‑reach interconnects reliably are now hitting fundamental limits in reach, thickness, system power consumption, and routability. These are constraints that directly cap how far rack‑level compute can scale.

The Copper Cliff: A Physical Limit on GPU Expansion

Copper twin‑ax has been the default for short‑reach connectivity because it is inexpensive, predictable, and operationally simple. But copper’s physics does not scale with AI’s bandwidth trajectory. As signaling rates push past 200G/lane and into terabit‑class aggregates, copper links must become shorter, thicker, heavier, and more power‑hungry on the system level to maintain acceptable BER and signal integrity.

The NVIDIA NVL72 illustrates this cliff with uncomfortable clarity:

• Over 5,000 copper twin‑ax cables inside one system
• 5 kilometers of copper routed within a single rack‑scale design

As hyperscalers and neo-cloud service providers attempt to expand compute density, adding more GPUs per rack, copper becomes the limiting factor. You cannot increase GPU count if you cannot physically route or install the copper required to connect them.

This is why Point2 Technology (Point2) focuses on the short‑reach domain (5–10 meters). Inside a rack or between adjacent racks, copper is not viable at the bandwidths and densities demanded by future AI clusters.

Why Not Just Use Fiber Optics? The Reliability Curve Breaks at Scale

Optical fiber solves distance and bandwidth but introduces scaling behavior that becomes problematic in dense AI systems.

At modest scale, fiber reliability issues, including link flap, contamination sensitivity, and connector alignment tolerances, are manageable. But as interconnect density increases, especially in AI clusters where scale‑up (more GPUs per rack) is 10× those of scale‑out use cases, optical reliability issues grow exponentially, not linearly.

Engineering teams see:

• Link flap events multiplying with port count
• Contamination sensitivity increases with connector density
• Rising operational overhead for servicing, monitoring, and remediation
• Higher failure rates in high‑vibration, high‑thermal‑flux environments

Any of these interruptions is cascaded downstream and causes delays for multi-million-dollar training runs. Additionally, optical interconnects can represent ~10% of total compute cost in large AI clusters. Fiber is excellent for long‑reach, but for short‑reach GPU‑to‑GPU traffic, its cost, fragility, and scaling behavior are mismatched to the environment.

Point2 e‑Tube: RF Over Waveguide as the Best Copper Replacement

Point2’s e‑Tube technology introduces a fundamentally different short‑reach interconnect: RF over polymer waveguide. Instead of electrons through copper or photons through glass, e‑Tube transmits modulated millimeter‑wave signals through a thin, flexible polymer tube.

Each e‑Tube cable contains eight waveguide cores and delivers:

• 5–10 meter reach with minimal signal loss
• ~1/3 the power of optical solutions
• ~1/3 the cost of comparable fiber links

Critically, e‑Tube preserves the operational characteristics engineers value in copper:

• Copper‑class user experience
• Copper‑class reliability (stable, no reliability issues that comes with optical components)
• Copper‑class economics (built in the same facilities as copper with similar manufacturing techniques)

This combination (scalable bandwidth with copper‑class behavior) makes e‑Tube the strongest candidate to replace copper in the short‑reach domain.

Why Short‑Reach Is the New Battleground

AI racks are becoming thermally and physically constrained. Liquid cooling is now standard, and copper’s bulk and routing complexity intensify those constraints. Short‑reach interconnects determine:

• How many GPUs can fit in a rack
• How cooling loops can be arranged
• How power distribution must be engineered
• How flexible GPU placement can be

If GPUs can be distributed more flexibly, connected by longer, lighter, more efficient RF waveguide cables, hyperscalers can:

• Scale rack‑level GPU counts and compute density
• Avoid copper’s physical limitations (routing congestion, weight, and reach)
• Scale with reliability and economics of copper

Short‑reach is where the copper cliff hits first, and where Point2’s e‑Tube provides immediate relief.

The Road Ahead: From Pluggables to Near-Packaged and Co‑Packaged Form Factors

e-Tube is extending copper experience at different stages, each unlocking new architectural freedom.

1. Pluggable Active RF Cables (ARC) – The first step is already shipping: Point2’s 1.6T OSFP Active RF Cable (ARC). These pluggable cables plug directly into existing OSFP cages in switches and servers, delivering terabit‑class bandwidth with copper‑class reliability and economics. ARC provides an immediate, low‑friction path to break through the copper cliff today.
2. Near‑Packaged Waveguide Routing – The next phase brings e‑Tube RF IOs closer to the GPU modules themselves. Near‑packaged designs reduce connector count, lower power overhead, simplify routing in rack-scale backplane connectivity to increase GPU count in data center racks. This decouples GPU count from copper’s constraints.
3. Co‑Packaged RF Waveguide Interfaces – The long‑term destination is co‑packaged RF IO interfaces, where e‑Tube connects directly at the GPU package level. This eliminates external connectors, minimizes latency, and dramatically improves power efficiency at each hop while enabling new rack‑scale architectures to maximize GPU count and compute inside the same rack

Beyond the 72‑GPU Ceiling

Today’s best‑in‑class systems, such as the NVL72, top out at 72 GPUs per rack, which is a limit imposed not by compute, but by copper. As e‑Tube progresses from pluggable ARC to near‑packaged and co‑packaged designs, it unlocks the ability to scale GPU count beyond 72, enabling larger, more flexible, and more power‑efficient rack‑scale AI systems.

Conclusion

The copper cliff is not theoretical. It is already here in modern AI systems. Fiber solves some problems but introduces others, especially as interconnect density scales. Point2’s e-Tube RF‑over‑waveguide approach offers a balanced, engineering‑sound path forward: scalable performance, copper‑class reliability, and copper‑class economics in the short‑reach domain where AI systems are most constrained.

For data center architects and hardware engineers designing next‑generation AI clusters, the message is clear: Short‑reach interconnects must evolve, and e‑Tube is the technology that unlocks the next wave of GPU expansion.

# # #

About the Author

Sean Park is a seasoned executive with over 25 years of experience in the semiconductors, wireless, and networking market. Throughout his career, Sean has held several leadership positions at prominent technology companies, including IDT, TeraSquare, and Marvell Semiconductor. As the CEO, CTO, and Founder at Point2 Technology, Sean was responsible for leading the company’s strategic direction and overseeing its day-to-day operations. He also served as a Director at Marvell, where he provided invaluable guidance and expertise to help the company achieve its goals. He holds a Ph.D. in Electrical Engineering from the University of Washington and also attended Seoul National University.

The post The Copper Cliff: Why AI Data Centers Need a New Kind of Cable appeared first on Data Center POST.

TL;DR

• The choice between edge and cloud isn’t a philosophical preference, but a practical constraint: does your AI feature need a response in under 100 milliseconds, or can it tolerate a network round trip?
• Cloud computing provides virtually unlimited compute and mature tooling, making it the perfect fit for training models and hosting large models (like LLMs). Edge computing trades raw compute for ultra-low latency and offline reliability.
• Edge is the mandatory choice in three scenarios: when latency must be under 20 milliseconds, when network connectivity is unreliable, or when transmitting raw data off-site is too expensive or risky due to security and bandwidth costs.
• Real-world AI applications rarely choose just one; mature deployments typically use a hybrid pattern where models are trained centrally in the cloud, compressed, and then deployed to the edge for real-time local inference.

# # #

Almost nobody actually needs to choose “edge or cloud” as an abstract architectural preference. What they need to answer is a much narrower question: does this specific AI feature need an answer in under 100 milliseconds, or can it tolerate a round trip to a data center? Everything else in this decision cost, complexity, where your team’s expertise sits, is downstream of that one number.

We learned this while helping a manufacturing client decide where to run a computer vision model checking parts on an assembly line for defects. The instinct on the call was to default to cloud, because that’s where the rest of their infrastructure lived. Then someone asked how fast the line moved. The answer was fast enough that a network round-trip to a cloud region would have let three more defective parts pass before the system could flag the first one. That single number ended the debate in about four minutes. Most of these decisions resolve just as cleanly once the actual constraint is on the table. The hard part is that teams often skip straight to a platform preference before anyone asks the question that actually matters.

What Each Term Actually Means for an AI Workload

Cloud computing:- runs your AI model on centralized, provider-managed servers, AWS, Google Cloud, and Azure, over the internet. You get effectively unlimited compute, mature tooling, and someone else managing the hardware. The tradeoff is a network round-trip for every inference request, and dependency on connectivity that isn’t always in your control.

Edge computing:- runs the model physically close to where the data is generated on a device itself, or on local hardware in the same building or vehicle. You trade raw compute power and easy scaling for a dramatic cut in latency and independence from network conditions.

Neither is “better” in general. The honest framing: cloud optimizes for compute capacity and ease of scaling; edge optimizes for response time and operating without a reliable connection. AI applications increasingly need both properties at once, which is exactly why this decision has gotten harder rather than easier as AI adoption has grown.

Why AI Made This Decision Harder, Not Easier

Before AI workloads entered the picture, this was mostly an IoT and telecom conversation sensor data, video feeds, and industrial monitoring. The decision criteria were relatively stable: how much data, how often, how sensitive to delay.

AI changed the math in a specific way that a lot of architecture guides don’t say directly: inference and training have almost opposite infrastructure needs, and most real AI products need both. Training a model, the process of teaching it from data, benefits from massive centralized compute and is not latency-sensitive at all; a training job can take hours in the cloud without anyone noticing. Inference, the model actually predicting new data, is often latency-sensitive in a way training never is, especially for anything touching a live camera feed, a voice interface, or a real-time control system.

That split is why so many “should we use edge or cloud for AI” conversations go in circles: people are usually asking the wrong half of the question, applying one answer to a workload that actually needs two different answers for two different phases of the same system.

Latency Is the Real Variable, Not a Buzzword

Every article on this topic mentions latency. Few explain what latency threshold actually forces the decision, so here’s the practical breakdown we use with clients, based on what the AI feature is actually doing:

• Under 20 milliseconds, real-time control systems, robotics, and autonomous vehicle perception. A cloud round-trip is physically incompatible with this requirement, regardless of how fast your internet connection is; the physics of distance and network hops don’t allow it. Edge is not a preference here; it’s the only option.
• 20–200 milliseconds interactive experiences a person is directly waiting on: live video analysis, voice assistants, AR overlays. Cloud can work if your network is reliable and your provider region is close, but the margin for error is thin, and the edge is often the safer default.
• 200 milliseconds to a few seconds, most “smart” features that feel real-time to a user but aren’t controlling physical machinery: recommendation engines, fraud flagging, chat-based assistants. Cloud handles this comfortably.
• Seconds to minutes, acceptable batch analytics, reporting, model retraining, anything reviewed by a person after the fact rather than acted on instantly. Cloud is the obvious and usually cheaper choice.

Notice that this framework never mentions “which cloud provider” or “which edge hardware” because those choices don’t matter until you know which latency band your feature actually lives in.

Where Cloud Still Wins, Decisively

For all the edge computing enthusiasm in AI conversations right now, cloud remains the right default for a specific and common category of workload: anything where the AI model is large, the inference doesn’t need to happen instantly, and the data isn’t prohibitively expensive or slow to transmit.

Large language models, complex recommendation systems, and any workload requiring a model too large to run on local hardware belong in the cloud, full stop. Edge devices simply don’t have the memory or compute to host them. Batch processing overnight fraud analysis, monthly demand forecasting, and retraining pipelines also belong firmly in the cloud, because none of it needs to happen the moment data arrives.

Machine learning in healthcare is a useful example of this split in practice: a hospital’s population-health risk model, crunching years of patient history to flag which patients need proactive outreach, has zero latency pressure and enormous compute needs, an obvious cloud workload. A bedside monitor flagging an irregular heartbeat in real time is the opposite case entirely, and belongs on the edge, or close to it.

Where Edge Wins, Decisively

Edge computing earns its place in three situations, and we’ve found teams overcomplicate this by adding vague “efficiency” justifications when really only these three hold up:

1. Latency requirements physics can’t solve any other way, robotics, autonomous systems, industrial safety monitoring, live video analysis for security, or quality control.
2. Unreliable or absent connectivity offshore equipment, rural agricultural sensors, in-vehicle systems, disaster response tools that need to function when networks go down, not just when they’re available.
3. Data that’s too sensitive, too large, or too regulated to transmit off-site, a factory floor with terabytes of video footage, where sending everything to the cloud is both expensive and unnecessary, since only flagged anomalies actually need to leave the building.

That third point deserves more attention than it usually gets. Edge computing isn’t just a latency play; it’s often a bandwidth and data governance play. Running inference locally and transmitting only the results (a flag, a classification, a summary) rather than the raw data itself can cut data transfer costs by an enormous margin while also sidestepping regulatory questions about where sensitive raw data is allowed to reside.

The Hybrid Pattern Almost Everyone Ends Up At

Here’s what most “edge vs. cloud” content doesn’t say plainly enough: the answer for a real AI product is rarely one or the other. It’s a specific division of labor that’s become common enough to be a default pattern rather than an exception:

• Train in the cloud, where compute is abundant and time pressure is low.
• Deploy a compressed or distilled version of the model to the edge for real-time inference, a smaller, faster version of the same model, optimized to run on limited local hardware.
• Send only the edge model’s outputs, not raw data, back to the cloud for monitoring, retraining, and improving the next version of the model.

This is effectively what’s happening inside modern security cameras with on-device object detection, in-vehicle driver-assistance systems, and industrial defect-detection cameras: heavy lifting happens centrally, ahead of time; a lighter, faster model runs locally, in the moment; results flow back centrally to make the next version smarter. The architecture question usually isn’t “edge or cloud,” it’s “where exactly does the line sit between the two for this specific model.”

The Cost Comparison Nobody Frames Correctly

Cost comparisons between edge and cloud usually get reduced to “cloud has ongoing fees, edge has upfront hardware costs,” true, but incomplete in a way that misleads more often than it helps.

The comparison that actually matters is data transfer cost at scale, and it’s the one most teams don’t calculate until the bill arrives. A single camera feed running continuous cloud-based video analysis can generate a genuinely large monthly data transfer bill. Once you’re running dozens or hundreds of feeds, the compute cost for the AI model itself is often the smaller line item. Run the same analysis at the edge, transmitting only flagged events, and the data transfer cost drops sharply, sometimes enough to justify the edge hardware’s upfront cost within the first year.

The reverse mistake is just as common: businesses invest in edge hardware for a workload that doesn’t generate enough data volume or urgency to justify it, then pay for underused local compute that a cloud subscription would have handled more cheaply with none of the maintenance burden. Neither direction is automatically cheaper it depends entirely on data volume, urgency, and how much local hardware maintenance your team is actually equipped to take on.

Model Size Is the Constraint People Forget to Check

A surprising number of “edge vs. cloud” decisions get made before anyone checks whether the model in question can physically run on the edge hardware being considered. Large models, particularly large language models and complex vision transformers, often can’t be compressed enough to run within the memory and power constraints of typical edge devices without a meaningful accuracy tradeoff.

This is where AI workflow automation projects run into trouble most often: a team prototypes a feature using a large cloud-hosted model, gets impressive accuracy in testing, then discovers during deployment planning that the model has no realistic path to running at the edge without significant compression, and compression frequently costs several accuracy points that weren’t part of the original pitch to leadership. Check model size and hardware constraints before promising a latency target, not after.

A Decision Framework That Isn’t a Flowchart

Most architecture decision frameworks turn into elaborate flowcharts that look thorough and get ignored in practice. Here’s the shorter version that’s actually held up across client engagements: four questions, asked in this order:

1. What’s the actual latency requirement, in milliseconds, for this specific feature? Not “fast” a number.
2. Can the model that hits that latency target actually run on the hardware you’re willing to deploy? If not, the latency requirement and the model choice conflict, and one of them has to give.
3. What’s the connectivity reality where this runs? Consistent, high-bandwidth connection, or intermittent and unreliable?
4. What does moving the raw data off-site cost or risk in transfer fees, in regulatory exposure, in bandwidth strain?

Answer those four honestly, in that order, and the edge-vs-cloud decision for a given feature usually resolves itself without needing a broader philosophical stance on which architecture is “the future.”

Mistakes We See Teams Make

A handful of patterns show up often enough to name directly:

• Choosing an architecture based on industry trend rather than the specific feature’s latency and hardware constraints, edge computing is having a moment in AI conversations right now, which leads some teams toward it for workloads that had no real latency pressure to begin with.
• Promising a latency target before confirming the model can run on the intended hardware. This is the single most common cause of late-stage AI deployment delays. The model works perfectly in a cloud test environment, but it can’t be compressed enough to hit the promised response time on the actual edge device.
• Treating “hybrid” as a compromise rather than the default. Teams sometimes feel pressure to pick a side cleanly, when in practice the train-centrally, infer-locally pattern is the normal shape of a mature AI deployment, not an indecisive middle ground.
• Underestimating data transfer costs at scale when defaulting to cloud for continuous sensor or video-based AI features, only to discover the real cost once usage scales past the pilot.

Closing Thought

Edge versus cloud was never really a question about which architecture is more advanced; it’s a question about where a specific piece of your AI system needs to live, given its latency requirement, its connectivity environment, and the size of the model doing the work. Cloud remains the right default for training, for large models, and for anything that can tolerate a delay. Edge earns its place when milliseconds matter, when connectivity can’t be guaranteed, or when moving raw data off-site costs more than it’s worth.

The businesses getting this right aren’t picking a side; they’re asking the four questions above for each specific feature and letting the answer fall out naturally, rather than deciding on an architecture philosophy first and forcing every feature to fit it.

# # #

About the Author

Ketan Barad is the Co-founder & CTO of encodedots, leading technology and operations with a strategic, innovation-driven approach. With strong expertise in technology and business management, he drives scalable solutions, process optimization, and consistent delivery excellence. His leadership in custom web application Development enables encodedots to build high-performing, future-ready Digital products, helping clients worldwide achieve sustainable growth and long-term success.

The post Edge Computing vs. Cloud Computing: Choosing the Right Architecture for AI Applications appeared first on Data Center POST.

TL;DR:

• Green Capital officially advanced Atlas GigaHub with the signing of a Strategic Partnership Agreement in Buenos Aires.
• Planned for up to 3 GW of IT capacity, the project pairs hyperscale AI infrastructure with dedicated renewable energy.
• The milestone highlights a new model for AI infrastructure built through collaboration between government, industry, and energy partners.

# # #

Artificial intelligence is reshaping the digital economy, but the race to deploy AI infrastructure is increasingly constrained by one critical factor: access to reliable, scalable power. As utilities face growing demand, interconnection queues lengthen, and permitting timelines become more complex, developers are reevaluating where and how the next generation of AI infrastructure should be built.

Against this backdrop, Green Capital S.A. (“Green Capital”) recently reached an important milestone in the development of Atlas GigaHub, its planned hyperscale AI infrastructure project in Argentina. During a Strategic Partnership Agreement signing ceremony hosted at the Embassy of the Republic of Poland in Buenos Aires, government leaders, technology companies, energy providers, investors, and industry stakeholders came together to mark the project’s next phase.

While the ceremony celebrated a significant achievement for Atlas GigaHub, it also reflected a broader shift taking place across the digital infrastructure industry. Increasingly, AI infrastructure is moving beyond traditional data center markets toward locations that can deliver abundant renewable energy, long-term power certainty, and room to scale.

A Different Approach to AI Infrastructure

For years, hyperscale development has largely followed established data center markets where network connectivity, cloud ecosystems, and customer demand were already concentrated. Today, however, many of these regions are experiencing mounting pressure from limited grid capacity, rising electricity costs, and increasingly competitive access to power.

Rather than asking where power is currently available, many organizations are beginning to ask where future AI infrastructure can be developed alongside dedicated energy resources. This shift is creating opportunities for regions that combine renewable energy potential, available land, supportive public policy, and the ability to develop large-scale infrastructure from the ground up.

Atlas GigaHub represents one example of this evolving approach.

Planned to support up to 3 GW of IT capacity powered by approximately 10 GW of renewable energy generation, Atlas GigaHub is designed around dedicated micro-grid infrastructure that provides long-term energy reliability while reducing dependence on constrained public transmission networks. By pairing hyperscale data center capacity with renewable energy development, the project aims to create an integrated platform capable of supporting the growing demands of AI workloads.

Argentina’s Opportunity

Argentina offers many of the characteristics becoming increasingly important for next-generation AI infrastructure.

The country offers significant renewable energy resources, vast areas suitable for development, favorable climatic conditions in regions such as Chubut, and growing interest in attracting technology investment. As AI continues driving unprecedented demand for digital infrastructure, these advantages position Argentina to play an increasingly important role in supporting global compute capacity.

During the signing ceremony, Ignacio Torres, Governor of Chubut, noted that this project is a key step in  bringing Argentina into the future. His statement reflects more than support for a single development; it highlights the broader economic opportunity associated with attracting advanced digital infrastructure and positioning Argentina within the rapidly expanding global AI ecosystem.

Building an Ecosystem for Gigawatt-Scale Development

Projects of this scale require far more than land and power. Successfully delivering hyperscale AI infrastructure demands expertise across renewable energy generation, electrical systems, engineering, financing, legal frameworks, and digital infrastructure design.

The signing ceremony reflected this collaborative approach, bringing together representatives from Wärtsilä, Schneider Electric, Envision Energy, Ore Energy, Green Capital, government agencies, and investment organizations. Speakers including Green Capital CEO Michał Polanowski, Governor Ignacio Torres, Adam Ryszka of the Embassy of the Republic of Poland in Buenos Aires, and Victoria Castro of PromArgentina highlighted both the scale of the opportunity and the partnerships required to bring it to life. Marta Zlotkowska then outlined the Atlas GigaHub vision before joining an expert panel discussing what it takes to deliver AI infrastructure at scale.

The discussion reinforced an increasingly important reality: delivering next-generation AI infrastructure requires coordinated collaboration across the public and private sectors.

From Concept to Execution

Earlier this year, Green Capital introduced Atlas GigaHub as a new model for hyperscale AI infrastructure, one designed around dedicated renewable energy resources and large-scale deployment rather than traditional grid-dependent expansion.

The Strategic Partnership Agreement represents an important step in transforming that vision into execution.

As AI adoption accelerates, demand for digital infrastructure will continue to expand into regions capable of supporting the power requirements these facilities demand. Success will increasingly depend not only on access to capital and technology, but also on long-term partnerships between governments, infrastructure developers, energy providers, and technology companies.

From Milestone to Momentum

The future of AI infrastructure will not be defined solely by advances in computing technology. It will also be shaped by where reliable, scalable, and sustainable energy can be developed to support that growth.

Atlas GigaHub represents one vision for how that future may unfold, bringing together renewable energy, hyperscale infrastructure, international partnerships, and government collaboration to create a new model for AI development.

As the industry continues searching for ways to deliver capacity faster and more sustainably, projects like Atlas GigaHub illustrate that the next generation of AI infrastructure may emerge in places where power, policy, and long-term vision converge.

See more from the event here.

The post From Vision to Reality: Why Atlas GigaHub Signals a New Era for AI Infrastructure appeared first on Data Center POST.

TL;DR

• Decisions made during design and installation establish long-term maintainability, enabling data centres to scale for AI workloads and higher rack densities without costly downtime.
• High-quality installation and precise construction records simplify operations, troubleshooting, and future upgrades.
• Thorough Integrated Systems Testing (IST) establishes vital performance baselines and transfers critical knowledge to operational teams before live workloads are deployed.

# # #

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

Data center lifecycle management begins long before a facility goes live. Decisions made during design and installation establish the foundation for long-term resilience, efficiency, and scalability. Designing with operational maintainability and future growth in mind, such as ensuring clear access routes, disciplined cable routing, and modular expansion capacity, allows facilities to accommodate rising rack densities and AI workloads without requiring disruptive redesigns. High-quality execution during construction further protects asset longevity, as accurate labelling, precise equipment installation, and thorough project documentation simplify ongoing maintenance and troubleshooting.

Commissioning and handover serve as the critical transition point where a completed facility becomes a fully functional operational asset. Comprehensive functional testing and Integrated Systems Testing (IST) validate how interconnected systems perform under real-world stress and failure scenarios, resolving issues before live workloads are deployed and establishing performance baselines for future health monitoring. Coupling rigorous testing with structured knowledge transfer, including as-built documentation, asset registers, and operator training, equips facilities teams to maintain uptime and minimize fault resolution times throughout the building’s operational life.

As critical infrastructure ages, proactive lifecycle replacement helps facilities keep pace with evolving capacity and resilience requirements without compromising service availability. Phased upgrade strategies allow operators to modernize end-of-life power, cooling, and control systems within live, mission-critical environments. Embedding comprehensive lifecycle thinking across every phase from initial layout to eventual hardware renewal minimizes operational risk, lowers total cost of ownership, and maximizes long-term investment value.

This content originally appeared on the Datalec Precision Installations website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post Lifecycle Management Starts Before a Data Center Goes Live appeared first on Data Center POST.

TL;DR

• Lightpath reports a 32% annualized increase in IP backbone traffic over the past 18 months.
• The growth reflects accelerating enterprise adoption of AI and large language model platforms.
• AI inference workloads are shifting network demand from hyperscale data centers to enterprise environments.
• The trend reinforces the growing importance of dense, high-capacity fiber infrastructure for enterprise AI.

# # #

For much of the AI conversation, attention has focused on the massive data centers used to train large language models. But once those models are built, the real work begins. Every time a hospital queries an AI diagnostic tool, a financial institution scores a transaction, or a university runs an AI-assisted research workload, that request travels across a network.

This stage of AI, known as inference, is becoming one of the clearest indicators of how quickly enterprises are adopting artificial intelligence. Unlike model training, which takes place inside a relatively small number of hyperscale facilities, inference happens wherever organizations are putting AI to work in their day-to-day operations. As adoption grows, the supporting network becomes just as important as the models themselves.

Lightpath’s latest network data provides a real-world view of that shift. The company reported that IP traffic across its backbone has grown at an annualized rate of 32% over the past 18 months, significantly outpacing its historical annual growth rate of 19%. Rather than being driven by a single customer or industry, the increase has been broad-based across its enterprise customer base, reflecting how AI is becoming embedded across multiple sectors.

This is an important distinction because enterprise AI generates a different pattern of network demand than model training. Training workloads are concentrated within large AI campuses, while inference creates millions of distributed requests originating from businesses, healthcare providers, financial institutions, universities and other organizations. Each interaction requires reliable, high-capacity connectivity between enterprise locations and AI platforms, placing increasing demands on metro and regional fiber networks.

As organizations integrate AI into everyday business processes, network performance becomes increasingly mission-critical. Low latency, high-capacity connectivity and resilient backbone infrastructure are no longer simply desirable characteristics—they are becoming foundational requirements for organizations that depend on real-time AI applications to support operations and decision-making.

Lightpath’s network is designed to support this growing demand through 12,100 route miles of AI-grade fiber spanning 11 major U.S. metropolitan markets and direct connections to more than 18,000 service locations. As enterprise AI adoption continues to accelerate, this type of dense metro infrastructure will play an increasingly important role in ensuring organizations can reliably access the AI platforms they depend on.

Perhaps most importantly, the announcement highlights how network traffic itself is becoming an early indicator of AI adoption. While new model launches and hyperscale investments often dominate headlines, increasing IP backbone traffic provides tangible evidence that enterprises are moving beyond experimentation and embedding AI into everyday operations.

As AI continues to evolve from a specialized technology into standard business infrastructure, the networks connecting enterprises to AI platforms will increasingly determine how effectively organizations can realize the technology’s full potential.

Read more in the press release here.

The post Lightpath’s IP Backbone Reveals the Next Phase of Enterprise AI Adoption appeared first on Data Center POST.

TL;DR

• Enterprise infrastructure planning has evolved from default public cloud adoption to a workload-driven approach balancing public cloud, dedicated infrastructure, and regional colocation.
• Specialized Neocloud platforms provide accessible, high-performance GPU compute required for AI training without the capital burden of dedicated hardware.
• Edge colocation in secondary markets supplies high-density power, cooling, low latency, and geographic diversity to keep distributed systems connected.
• Combining Neocloud compute, edge facilities, and public cloud creates a unified framework optimized for cost, performance, and long-term scalability.

# # #

Written by iMiller Public Relations on behalf of fifteenFortySeven.

Enterprise IT strategy has shifted from a “cloud-first” mandate toward a “cloud-smart” philosophy. Rather than forcing all applications onto public cloud platforms, organizations evaluate workloads individually based on specific performance demands, operational costs, and governance requirements. Industry research from Gartner underscores that hybrid cloud adoption is accelerating as businesses seek to match each application with its optimal operational environment rather than standardizing on a single deployment model.

The rise of artificial intelligence has intensified these infrastructure demands, as training and deploying AI models requires specialized GPU capacity that is expensive to build in-house. Neocloud platforms have emerged as a purpose-built solution, enabling companies to access compute-heavy capacity without the complexity of managing dedicated AI infrastructure. However, as McKinsey notes, moving AI initiatives from experimentation into day-to-day production requires reliable, low-latency connectivity back to enterprise systems, public cloud services, and end users.

Regional edge colocation bridges these environments by delivering local physical infrastructure, high-density power, cooling, and network interconnection. As highlighted by JLL, secondary markets continue to attract infrastructure investment due to geographic diversity, capacity expansion, and disaster recovery benefits. By integrating Neocloud platforms with regional colocation, organizations establish an interconnected hybrid architecture where AI model training, sensitive enterprise data, customer-facing applications, and public cloud services seamlessly interoperate.

This content originally appeared on the fifteenFortySeven website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post From “Cloud-Only” to Cloud-Smart: Why Neocloud and Edge Colocation Belong Together appeared first on Data Center POST.

TL;DR

• Workload demands have forced a shift away from single-environment standardization, as no individual cloud or data center satisfies every application’s need for latency, scale, and control.
• Success requires an operating model approach focused on strict governance, design, and cross-platform visibility to prevent cost surprises and duplicate tools.
• Workload segmentation is key: secure colocation or private environments are best for data-heavy and mission-critical systems, while public clouds handle burstable, customer-facing services.

# # #

Written by iMiller Public Relations on behalf of 365 Data Centers.

Enterprises are shifting away from standardizing around a single cloud or primary data center model because no single environment can fit every workload’s unique demands for latency, control, scale, and flexibility. Modern hybrid infrastructure and multi-cloud strategies have become practical responses to real business needs, allowing organizations to place specific workloads where they perform best rather than forcing everything into one rigid model. This balanced approach directly addresses challenges associated with rising cloud costs, compliance requirements, performance demands, and data gravity, while adding layers of resilience and pricing flexibility.

Successful implementation requires treating these combined environments as a comprehensive operating model rather than just a tool choice. Organizations must prioritize design, governance, and visibility. Without clear policies, they risk running into duplicate tools, inconsistent security controls, and unexpected costs. A practical starting point is workload segmentation, where mission-critical, data-heavy, or highly regulated systems are assigned to secure private or colocation environments, while burstable or customer-facing services utilize the public cloud.

Ultimately, this flexibility delivers business continuity, procurement adaptability, and lower-risk, phased migrations instead of high-risk “big bang” transitions. To make this operational model sustainable over time, companies must design connectivity, security, and support together, ensuring security controls travel with workloads and teams maintain clear views of latency, cost, and utilization across platforms.

This content originally appeared on the 365 Data Centers website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post Adopting Hybrid Data Center and Multi-Cloud Strategies: The Way Forward appeared first on Data Center POST.

TL;DR

• During major outages, generators rarely fail on their own; the true vulnerability is navigating blocked routes, terminal congestion, and surging demand to deliver the fuel.
• Operational continuity requires proactive fuel lifecycle management, including realistic tank sizing, real-time monitoring, and continuous filtration, long before a storm hits.
• Emergency fuel deliveries must only serve as a backup to a robust, pre-planned system, rather than acting as the system itself.

# # #

When major outages hit, generators usually do what they are supposed to do. The harder problem is often fuel logistics: getting the right fuel to the right place, at the right time, when roads are blocked, terminals are constrained, and demand surges all at once. Hurricane Helene was a clear reminder that continuity plans are only as strong as the fuel systems behind them.

In many affected areas, the issue was not whether generators could run, but whether fuel could reliably reach the facilities that depended on them. Emergency managers and critical infrastructure operators faced blocked access routes, terminal disruptions, and competing demand across the region. That same pattern has shown up in other hurricanes and grid failures as well, reinforcing a simple point: Runtime on paper is not the same as runtime in practice.

For organizations that support mission-critical operations, these events are not isolated lessons. They show that fuel resilience has to be built into planning from the start, not improvised during an emergency. Hurricane Helene made that especially clear because it exposed how quickly logistics, access, and fuel visibility can become the real limiting factors.

What Real Events Have Taught Us

Looking across Hurricane Helene, Hurricane Sandy, severe derechos, winter storms, and multi-day grid failures, one lesson stands out: Last-mile fuel delivery is where most plans are tested. In earlier events like Sandy, terminal congestion, driver shortages, and access restrictions quickly turned routine replenishment into a critical operation. ESI Total Fuel Management’s role in these situations has often been to coordinate across terminals, carriers, and authorities so fuel actually reaches tanks when it matters most.

Recent storms have shown the same operational pressures in different forms. Fuel may exist in the region, but access to the right terminal or route can still be constrained. Carrier capacity disappears quickly when multiple sectors are competing for the same resource. Facilities without real-time visibility into fuel levels and usage tend to react late, which makes recovery harder.

The practical lesson from Hurricane Helene and other recent events is that last-mile logistics cannot be treated as an afterthought. They need to be part of the resilience plan before an outage begins.

From Delivery to Resilience

Over time, ESI has evolved from a fuel supplier and systems provider into what it views as an integrated fuel resilience strategy for mission-critical environments. That approach includes tank design, monitoring systems, polishing and filtration, automated controls, and long-term maintenance programs tailored to data centers, hospitals, and other critical facilities.

This matters because emergency deliveries are only one part of the picture. A resilient program also depends on physical infrastructure and operational planning: Are tanks sized for realistic outage durations? Are monitoring and alarms in place so operators can act early? Are fuel quality processes built in so generators start cleanly on day one and day three?

When these pieces work together, fewer things have to go right during a crisis. This is the difference between a fuel program that simply exists and one that can be trusted under stress. Hurricane Helene underscored that distinction in a very visible way.

Renewable Diesel and Runtime Expectations

The next challenge is that runtime expectations are changing. AI growth, denser data centers, and stricter service-level requirements are pushing many facilities to plan for extended generator operation, sometimes 48, 72, or more hours. At the same time, organizations are still working toward net-zero or lower-carbon goals.

That is where renewable diesel and hydrotreated vegetable oil (HVO/R99) enter the conversation. ESI’s expansion of HVO services reflects a practical response to this tension: facilities need fuel options that support sustainability goals without compromising reliability.

The key is to treat renewable fuels with the same rigor as conventional diesel. They still need to be sourced correctly, stored properly, monitored carefully, and managed across their full lifecycle so they perform when needed. The transition to cleaner fuels does not remove the need for discipline in fuel logistics; it increases it.

Fuel Lifecycle Management

ESI describes its approach as Fuel Lifecycle Management, a framework that treats fuel as a managed asset from procurement to combustion. That means the work starts long before an emergency delivery is ever required.

The model includes:

• Designing and maintaining storage and distribution systems that match realistic runtime needs.
• Implementing monitoring, alarms, and analytics so operators have real-time visibility into fuel conditions.
• Managing fuel quality through testing, polishing, and filtration rather than relying on occasional spot checks.
• Building logistics planning into the resilience strategy, including alternate terminals and carriers.

This is where the idea of fuel resilience becomes operational, not theoretical. It is not just about getting more fuel delivered faster during a storm; it is about combining infrastructure, maintenance, and planning so the system is ready before the storm arrives.

Final Takeaway

If Hurricane Helene reinforced anything, it is that fuel should be treated as critical infrastructure and a managed asset, not just another commodity. Storage, quality, monitoring, logistics, and emissions planning all have to work together if facilities want to maintain continuity under pressure.

Emergency deliveries will always be part of the response model, but they should be the backup to a resilient system, not the system itself. That systems-level perspective is what allows mission-critical facilities to weather outages with fewer surprises and more control.

Contact ESI Total Fuel Management today to ensure your systems are fuel ready for the next emergency.

The post Emergency Fuel Delivery: Lessons from Hurricane Helene and Recent Grid Failures appeared first on Data Center POST.

TL;DR

• Liquid cooling controls automatically adjust pump speeds or open valves to compensate for early deviations, meaning temperature graphs often remain falsely calm while critical evidence is erased.
• Record the alarm source, exact timestamp, measured values, and system trends before resetting any device or changing setpoints.
• Use clean-loop commissioning records (between minutes 5 and 15) to evaluate how current flows, pressures, and control efforts compare to known healthy baselines.
• Define owners, confirmation signals, escalation paths, and telemetry snapshot protocols for every alarm before an incident occurs on the night shift so facilities and IT teams can coordinate seamlessly.

# # #

By Rupesh Mainali, Senior Member of Technical Staff, Reliability Engine

A low-flow alarm appears on one rack during an overnight shift. Supply temperature is normal. The GPUs are not throttling. Clearing the alarm and watching the trend may feel reasonable, but that reset can erase the best evidence the team has.

Liquid cooling controls are built to compensate. A pump can speed up, a valve can open farther, or another branch can give up some capacity while the temperature graph remains calm. By the time temperature finally moves, the filter, valve, sensor, fluid, or connection that started the event may have been drifting for hours.

The first 30 minutes should protect people and equipment, preserve the event, and narrow the fault to a useful boundary. Root cause can take longer. Good incident response keeps the early evidence intact so the later investigation starts with facts.

This sequence sits underneath the site’s approved emergency procedures. Electrical safety rules, environmental requirements, equipment limits, and trained-personnel requirements always take priority, especially when fluid may be present near energized equipment.

Start With the Alarm You Actually Have

A moisture detector, a low-flow warning, high filter pressure loss, a return-temperature excursion, and a conductivity alert are different events. Each needs different confirmation. Grouping them under a single label such as cooling problem makes the first response slower and less precise.

The initial question is practical: is this a containment issue, a hydraulic issue, a thermal issue, a fluid-condition issue, or some combination? That answer determines which data matters and who needs to join the response.

First Five Minutes: Preserve the Event

Before anyone resets a device or changes a setpoint, capture the alarm source, exact time, location, threshold, measured value, and the trend immediately before it. Add the rack or branch identity, workload state, pump command, valve position, and any recent maintenance. Synchronized timestamps matter because facilities telemetry and IT telemetry often live in different systems.

Then look for a second signal. Confirm a moisture alarm against reservoir level, pressure decay, or a safe visual check at likely collection points. Read low flow beside differential pressure, pump command, valve position, and neighboring branch flow. Read a temperature excursion beside workload and coolant inlet conditions.

If an automatic interlock has already acted, preserve its state and follow the site’s approved path. Restoring a green dashboard is not the same as restoring the physical system.

Minutes 5 to 15: Compare Like with Like

The strongest reference is the same system at a comparable load when it was known to be healthy. A clean-loop commissioning record should connect fluid identity, temperature, branch flow, differential pressure, pump and valve commands, filter condition, chemistry, and workload. It turns an alarm into a comparison instead of a debate about what normal used to look like.

Read the relationships between signals. Falling flow with rising pump effort or pressure loss directs attention toward a restriction, filter loading, branch balance, or valve position. Falling flow while pump command also falls points somewhere else, such as controls, power, or instrumentation. These patterns guide inspection; they do not prove a diagnosis on their own.

Peer racks are useful controls. When one branch changes and the others do not, the search can stay local. When several branches move together, look upstream at the coolant distribution unit, shared controls, facility-water conditions, or a common maintenance event.

Fluid readings also need context. A conductivity or pH change after a fill, filter replacement, or service visit means something different from the same movement during stable operation. Verify sensor condition, temperature compensation, sample location, and recent additions before changing chemistry.

Control effort is often the clue that arrives before temperature. A loop may still hold setpoint while the pump works harder or the valve runs closer to its limit. In a longer trend, thin deposits can steal thermal margin before they create an obvious blockage. Temperature, hydraulics, fluid condition, and control effort are most useful when read together.

Minutes 15 to 30: Choose a Response Path

By minute 15, the team may still be working toward root cause. It should know whether the alarm is credible, which part of the system is affected, and which response path is justified.

A credible leak indication follows the site’s containment and isolation plan. Logs, safe photographs, and the location of any fluid should be preserved before cleanup removes the evidence. The affected boundary also needs to be checked for fluid migration beyond the first visible point.

A hydraulic event calls for a controlled review of recent service, filter pressure loss, branch balance, valve position, pump behavior, and possible air entry. A thermal event needs workload verification, peer comparison, inlet-condition review, and sensor checks. A fluid-quality event needs measurement validation and, where the site procedure requires it, a representative sample.

When the procedure allows, change one variable at a time and watch for the expected response. Several simultaneous adjustments can recover service while leaving the team unsure which action worked.

A Cleared Alarm Is Not Recovery

Closure should mean that the affected boundary is understood, the relevant signals are stable against the right baseline, the intervention has been verified, interlocks are restored, and a named owner has the remaining follow-up. A temperature graph returning to normal is only one part of that decision.

Keep the original alarm, synchronized trends, workload context, maintenance history, actions taken, samples or photographs, and the evidence used for return to service. That record helps the next shift, the post-incident review, and any discussion with equipment or fluid suppliers.

Make the Response Routine Before the Night Shift

A good alarm response should feel routine. Each alarm definition should already identify its owner, confirmation signal, authorized action, telemetry snapshot, escalation path, and recovery criteria. Facilities, IT operations, hardware reliability, and environmental health and safety teams should know where their responsibilities meet.

No team will solve every cooling problem in half an hour. A successful first 30 minutes leaves the system safer, the fault tree smaller, the evidence intact, and the next action assigned. That is how a small cooling event stays an operational task instead of becoming an expensive mystery.

# # #

About the Author

Rupesh Mainali is a Senior Member of Technical Staff at Reliability Engine, where he works on predictive reliability for liquid-cooled AI infrastructure. His focus includes coolant condition, loop behavior, commissioning evidence, and the operating signals that connect cooling performance to compute risk.

The post The First 30 Minutes After a Liquid Cooling Alarm appeared first on Data Center POST.

TL;DR

• Building Momentum: Sabey Data Centers celebrated the “Going Vertical” milestone at SDC Umatilla, marking the start of vertical construction as the first steel was raised at the Northeast Oregon campus.
• Designed for the Future: Located in the Columbia Cloud Corridor, SDC Umatilla is being developed to support hyperscale, enterprise and colocation customers with scalable infrastructure for growing cloud and AI demand.
• Investing in the Region: The project reflects Sabey’s long-term commitment to Northeast Oregon through community partnerships, sustainable design and continued economic investment.

# # #

Every major project reaches a point where years of planning become something tangible. For Sabey Data Centers, that moment arrived at SDC Umatilla last week as the first steel was raised, marking the campus’ “Going Vertical” milestone and the beginning of visible vertical construction.

Held on July 15, the ceremony brought together local and state leaders, community partners, utility partners, contractors, construction teams and Sabey leadership to celebrate the progress of the Northeast Oregon campus. More than a construction milestone, the event reflected the collaboration that has shaped the project from the earliest planning stages and the long-term investment Sabey is making in the region.

Located in the Columbia Cloud Corridor, SDC Umatilla is being developed to meet growing demand for resilient data center infrastructure driven by cloud computing and artificial intelligence. The campus will provide hyperscale, enterprise and colocation customers with the flexibility and connectivity to scale that today’s digital economy requires.

“Watching the first steel go up is always a significant moment because it turns years of planning, engineering and hard work into something you can finally see,” said Tim Mirick, President of Sabey Data Centers. “We are excited to build here, excited to become part of this community, and committed to serving our customers and this region for many years to come.”

The project also highlights Sabey’s commitment to responsible development. State Representative Bobby Levy recognized both the economic impact of the campus and its focus on sustainability:

“The construction of this new data center is an exciting investment in Northeast Oregon and a strong vote of confidence in our communities. Facilities like this keep our hospitals, schools, and small businesses connected while creating new economic opportunities. I’m especially pleased that Sabey’s use of a closed-loop cooling system demonstrates that innovation and responsible resource stewardship can go hand in hand. We’re proud to welcome Sabey Data Centers to Northeast Oregon,” Levy said.

As construction continues, the project is expected to create opportunities for skilled tradespeople while supporting local businesses throughout the region. The campus reflects Sabey’s approach to building infrastructure that is supported by strong partnerships and designed for long-term success.

The post From Planning to Progress: SDC Umatilla Reaches a Major Construction Milestone appeared first on Data Center POST.

TL;DR

• Edge data centers process and store data closer to users, reducing network hops and cutting application latency from 80 milliseconds to less than 20 milliseconds.
• Driven by 5G, AI, and the rapid expansion of active IoT devices (which are expected to exceed 75 billion by the end of this decade), the global edge data center market is projected to surge from $22.4 billion in 2025 to $124.8 billion by 2034.
• Managing highly dispersed sites increases security attack surfaces, raises power and cooling challenges in compact spaces, and requires organizations to address complex interoperability in hybrid cloud environments.
• By the end of this decade, 50% of data created by business corporations will be processed outside of centralized data centers, increasingly utilizing automated AI management and modular construction.

# # #

There has been a significant change in the way that data is generated, processed and consumed due to the rapid rise of connected devices, artificial intelligence, cloud software, and immersive digital experiences. Traditional centralized data centers continue to play their role in large-scale computing; however, they are usually located at distances from end-users in kilo meters. This long distance results in network latency affecting responsiveness of time-critical applications.

Edge data centers are viewed as a significant leap of modern digital architecture. Rather than sending every request to centralized cloud facilities, edge data centers enable processing and storing data closer to users, devices, and enterprises’ locations. Estimates made by industry experts indicate that in recent years internet traffic reached more than 180 zettabytes, while the number of active IoT devices will exceed 75 billion by the end of this decade.

Market Intelo has projected that by 2025 the global Edge Data Center market is valued at $22.4 billion and will hit $124.8 billion by 2034 with a CAGR of 21.3% during the period of 2026-2034. Sustainable growth of the industry is mainly caused by the increasing need for low-latent computing, distributed cloud computing, and real-time processing across many industries.

Knowing Edge Data Centres

A data centre at the edge is a relatively small, but well-placed facility that helps execute workloads close to the original data. In contrast to hyperscale campuses that can have millions of square feet of space, edge centres are mostly limited to urban areas, industrial locations, transport junctions, and communication networks.

This type of decentralised architecture permits a decline of the number of hops between an individual and his/her application. In regard to sensitive latency services, the time of 80 milliseconds decreased to less than 20 milliseconds has a considerable influence on the capability of the application.

The Importance of Low Latency in Today’s World

Latency has become one of the major performance standards in areas of industry such as telecommunications and logistics. Today, as the field of digital technology is evolving, the response speed measured in milliseconds impacts operations, customer experience and product reliability.

There are several technologies that are responsible for this trend:

• 5G networks which enable communication of below 10 milliseconds in optimal environments.
• Industrial Internet of Things where factories can use around 100,000 or more sensors at the same time.
• AI inference workloads requiring immediate processing for decision-making.
• Connected vehicles generating several terabytes of data daily, much of which requires rapid local analysis.

For applications such as remote robotic operations, intelligent traffic systems, financial trading platforms, and healthcare monitoring, delays of even 20–30 milliseconds may reduce operational effectiveness. Processing information closer to its source minimizes these delays while reducing unnecessary backhaul traffic across core networks.

Main Benefits of Edge Data Centers

Edge computing offers the benefits of measurable operational efficiencies in enterprise and government operations.

This demonstrates the fact that various industries including telecommunications, manufacturing, retail and smart cities apply edge technology.

Sector Applications Are Driving Usage

The edge infrastructure is no longer confined to telecommunications companies. Businesses from various industries are beginning to use localized computing for operational purposes.

Manufacturers use predictive maintenance technologies that examine the functioning of machines in real time, allowing the detection of problems on time. Today’s manufacturing plants generate numerous terabytes of operational information every day, thus sending such information to remote clouds is irrelevant.

Medical institutions take advantage of edge computing in order to improve services such as medical imaging, patient monitoring and emergency response. Since processing takes place locally, medical personnel receive important information without delay and transmit less relevant data.

Retail companies use edge computing technologies for inventory analysis, automated checkout technologies, and providing custom digital experiences. Large shop owners can serve thousands of customers per hour, so local analysis helps to increase efficiency in times of high demand.

Obstacles in Developing Distributed Infrastructure

Despite the multiple benefits of edge computing, implementing it brings additional technical and operational hurdles.

In contrast to centralized sites, companies might have to monitor and control dozens or even numerous dispersed sites, making operational burden in terms of control, maintenance, cybersecurity, and infrastructure management bigger.

Power availability can also pose problems. Smaller locations need to ensure high reliability despite having less physical space, which can involve efficient cooling systems, compact productive units, and reliable backups.

Cybersecurity is essential when it comes to edge computing, since every edge location increases the attack surface. To mitigate operational risks in traditional settings, organizations can use the same security measures, policies, automated updates, encrypted connections, and monitoring.

Interoperability is a matter of concern as well. Companies may have their infrastructure and use hybrid environments consisting of private infrastructure, hyperscale cloud infrastructures, and edge computing sites.

Prospects of Low-Latency Infrastructure

The ongoing development of AI, 5G, and connected devices may enhance expenditure in distributed computing in the near future. Analysts project that by the end of this decade, 50% of data created by business corporations will be processed outside the centralized data centers, consequently proving the significance of local infrastructure.

The new edge facilities are supposed to feature the use of AI automation for predictive maintenance, dynamic load management, as well as energy efficiency. Improvements in technologies concerning liquid cooling, modular construction, and software-defined infrastructure can contribute to boosting the effectiveness of operations.

However, cooperation between cloud service providers, network operators, and colocation service providers will be crucial in creating the ecosystem that combines centralized and decentralized services.

Conclusion

Edge data centers signify a major advancement in digital infrastructure as opposed to being a substitute for conventional data centers. They help to meet the increasing demand for low latency, rapid analytics, and effective utilization of the network by providing computing power immaterially closer to clients and connected devices.

Given that the volume of data keeps increasing and many real-time processes become prevalent, the distributed framework will become more and more vital in supporting efficient, reliable, and scalable digital services in the near future.

# # #

About the Author

Raksha Sharma is a Jr. Team Lead at MarketIntelo 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 Edge Data Centers Are Reshaping Low-Latency Digital Infrastructure appeared first on Data Center POST.

TL; DR:

• PowerBridge and Liberty Energy have announced a strategic joint venture to develop integrated powered campuses.
• The partnership combines PowerBridge’s powered campus platform with Liberty Power Innovations’ expertise in power generation and energy management.
• The first project will support the planned 2 GW Alpha Digital Campus in West Texas, with initial power delivery anticipated in the fourth quarter of 2027.
• The joint venture creates a scalable model for future powered campus developments designed to support hyperscale, AI, and other large-load customers.

# # #

PowerBridge has announced a strategic joint venture with Liberty Energy to combine powered campus development with integrated power generation. The partnership brings together PowerBridge’s powered campus platform and Liberty Power Innovations’ operational expertise to support the delivery of gigawatt-scale powered campuses.

The joint venture will initially support the planned 2 GW Alpha Digital Campus in West Texas. The campus is expected to include more than 300 megawatts of generation capacity during its first phase, with initial power delivery anticipated in the fourth quarter of 2027 and continued deployment through the first half of 2028. Beyond this first project, the partnership establishes a framework for future powered campus developments across PowerBridge’s growing portfolio, with a commitment to working alongside local communities to create long-term economic opportunities and support workforce development.

The joint venture is designed to provide customers with an integrated solution that simplifies execution while reducing uncertainty around infrastructure readiness and power availability. PowerBridge will continue to lead campus development, including site infrastructure and regional fiber connectivity, while Liberty Power Innovations will design, develop, and operate the modular power generation systems that support each campus.

“We are delighted to partner with Liberty as we execute our powered campuses at gigawatt scale,” said Alex Hernandez, Founder and Chief Executive Officer of PowerBridge. “PowerBridge was founded on the premise of driving and solving the accelerating convergence between energy and digital infrastructure. Our mission is to serve our digital infrastructure customers with an integrated powered campus solution while building new power generation assets that strengthen grid reliability. By aligning our powered campus development assets with Liberty’s comprehensive power services and operational expertise, we are creating a model that can be scaled across future gigawatt-scale powered campuses in West Texas.”

Liberty Energy views the partnership as a response to changing customer expectations for digital infrastructure delivery.

“The most sophisticated customers are increasingly seeking partners that can simplify the delivery of large-scale digital infrastructure projects,” said Ron Gusek, Chief Executive Officer of Liberty Energy. “As power availability becomes a defining factor in data center development, customers need integrated solutions that align power generation, energy management, and campus infrastructure planning from the outset. With PowerBridge, we are helping customers secure reliable power faster and more efficiently, supporting the accelerating growth of AI and digital infrastructure.”

As demand for AI-ready digital infrastructure continues to grow, this joint venture positions PowerBridge and Liberty Energy to deliver an integrated development model that brings together power infrastructure and digital campus development under a single strategy. The result is a platform designed to accelerate deployment while providing customers with greater confidence in project execution and infrastructure readiness, while working alongside local communities to support workforce development, create long-term economic opportunities, and responsibly develop the infrastructure needed to power the digital future.

Read the full release here.

The post Building the Future of Powered Digital Infrastructure: PowerBridge and Liberty Energy Form Strategic Joint Venture appeared first on Data Center POST.

TL;DR

• DC BLOX has been selected as the digital infrastructure partner in a consortium chosen by the U.S. Department of Energy’s National Nuclear Security Administration (NNSA).
• The proposed project includes a 1-gigawatt AI data center campus and approximately 2 gigawatts of dedicated on-site power generation at the Savannah River Site in South Carolina.
• The initiative combines advanced digital infrastructure with reliable energy generation to accelerate AI innovation and support long-term economic growth.
• The project builds on DC BLOX’s investments across the Southeast and reinforces its commitment to delivering resilient, next-generation infrastructure for hyperscale AI and cloud computing.

# # #

The infrastructure powering the next generation of artificial intelligence is taking shape, and DC BLOX is helping lead the way. The company has been selected as the digital infrastructure partner in a consortium with Amentum chosen by the U.S. Department of Energy’s National Nuclear Security Administration (NNSA) to enter negotiations for a phased lease at the Savannah River Site in South Carolina. The proposed development represents an innovative public-private partnership designed to support America’s growing demand for AI computing while strengthening the nation’s long-term technology leadership.

The project calls for the development of a 1-gigawatt AI data center campus supported by approximately 2 gigawatts of dedicated on-site power generation. By pairing computing infrastructure with reliable energy production, the initiative addresses one of the industry’s biggest challenges: ensuring enough power is available to support increasingly intensive AI workloads. The effort also supports broader federal priorities aimed at expanding AI infrastructure and advancing next-generation nuclear energy technologies.

For DC BLOX, the selection builds on years of investment across the Southeast and reflects the company’s experience developing and operating high-performance digital infrastructure. The company will contribute that expertise as the consortium works to establish a new model for deploying AI infrastructure on federal land while supporting economic growth and American innovation.”DC BLOX is honored to join Amentum and this visionary consortium to support a program that accelerates our nation’s economic growth through technology advancement,” said Jeff Uphues, CEO of DC BLOX. “To lead, we must innovate not just in how we build data centers, but in how we power them.”

The opportunity also reflects DC BLOX’s continued commitment to the future of energy. As a founding member of the Palmetto Nuclear Coalition, the company is working with industry leaders and policymakers to help ensure nuclear energy remains a reliable foundation for America’s digital future. Aligning dependable power generation with modern digital infrastructure will be essential as demand for AI continues to grow, and initiatives like this demonstrate how those goals can move forward together.

The Savannah River Site project highlights a new approach to building the infrastructure that will support the next era of innovation. By combining proven digital infrastructure expertise with a long-term vision for reliable energy, DC BLOX is helping create the foundation that will power AI and strengthen the country’s competitive advantage for years to come.

Read the full release here.

The post DC BLOX Selected for Landmark Federal AI and Energy Infrastructure Initiative appeared first on Data Center POST.

TL; DR

• Sustainability in digital infrastructure starts before a data center opens, with construction methods playing a major role in environmental impact.
• Modular construction reduces disruption by shifting much of the work off-site, which can lower noise, traffic, waste, and the project’s overall carbon footprint.
• The article frames human sustainability as part of the equation, arguing that factory-based work can improve worker stability, safety, retention, and quality of life.
• Long-term community value matters as much as speed or efficiency, so Mod42 emphasizes workforce development, local partnerships, and future community initiatives alongside infrastructure delivery.

# # #

Digital infrastructure is expanding at unprecedented rates. AI workloads, cloud migration, and edge computing are driving data center demand to record levels. The International Energy Agency estimates global data center power consumption could reach 1,050 terawatt-hours by 2026, approximately equivalent to Japan’s entire electricity use.

This growth creates a critical question: How do we build the infrastructure the future needs while minimizing impact on the environment and communities where it sits?

The answer isn’t just about cleaner energy or better cooling. It also requires rethinking how data centers are delivered.  Sustainability and community aren’t separate concerns. They’re interconnected challenges that demand integrated solutions.

Sustainability is often discussed in terms of what a facility consumes once it is online. But for data centers, sustainability also starts long before operations begin. It includes how the project is built, how much disruption it creates, and how it affects the people and places around it.

Modular Construction: Reducing Physical Impact

Traditional data center construction can create years of noise, dust, debris, heavy truck traffic, and ongoing disruption for nearby communities. Modular construction changes that pattern by moving much of the work off-site while site preparation happens in parallel. Instead of a long period of active construction on location, the completed modules are delivered and installed in a much shorter window, reducing the duration of construction-related disruption. That makes modular construction a more community-friendly approach from the start, not just a faster one.

By shifting a large share of the work into factory environments, teams can reduce site activity, limit deliveries, and shorten the overall construction window. At Mod42, much of the data center is pre-engineered and manufactured off-site, which helps move the work from a crowded job site into a more controlled setting.

That shift matters for more than speed. It reduces on-site disturbance, lowers material waste, and shrinks the project’s overall carbon footprint. Because modules are built in a factory with lean, repeatable processes, material use is optimized and excess is minimized. Construction is one of the largest contributors to landfill waste; modular’s lean approach significantly cuts that waste at the source.

Modular construction is also more land efficient. Building in a controlled factory environment allows more capacity to be delivered on a smaller site. For example, Mod42 can fit up to 15 MW of data center capacity on a single acre, whereas a traditional build would typically require more land to deliver the same power. That reduced land disturbance means less grading, fewer impervious surfaces, and a smaller physical footprint for the same MW; this is another way modular lowers environmental impact before the facility even turns on.

The environmental benefit is not only in what gets built, but in how it gets built. Factory manufacturing creates far less construction waste because materials are used more efficiently, processes are repeatable, and excess is reduced. Construction remains one of the largest contributors to landfill waste, so minimizing waste at the manufacturing stage is an important part of a more sustainable model. Factory precision also helps improve consistency and quality while reducing unnecessary material use.

This is the key point: Construction methods and community impact are also central to sustainability.

Human Sustainability: The Workforce Angle

Sustainability is often discussed in terms of energy, water, and emissions. But there is another important dimension: human sustainability. That means thinking about the skilled workers who build this infrastructure, the stability of their jobs, and the quality of life that comes with a more predictable work environment.

Traditional construction projects can require long commutes, temporary relocation, and extended time away from home. That affects worker quality of life, retention, and consistency across projects.

Factory-based modular construction helps create stable, consistent employment for skilled workers. Instead of moving from project to project and spending more time traveling, employees return to the same workplace each day, build deeper expertise, and spend more time with their families. That consistency can reduce burnout, improve retention, and support a stronger workforce over time. It also allows experienced teams to build the same products repeatedly, helping deliver greater consistency and quality across every project.

When workers are not constantly moving between distant job sites, they are better able to stay connected to their communities and families. Over time, that stability can benefit both the workforce and the projects they support.

Manufacturing off site also creates a safer, more stable working environment. Workers are inside, protected from extreme heat, cold, rain, and/or sun. That sheltered setting reduces weather-related downtime and fatigue, supports steadier productivity, and helps keep projects on schedule. In other words, protecting the human element directly supports project performance.

Community Engagement Beyond Construction

A data center may be built in months, but its relationship with a community lasts much longer. That means the conversation should extend beyond construction timelines and operational performance. The broader discussion is also about workforce sustainability, employee retention, long-term skilled labor development, and the quality of life of the people doing the work.

Some of the most practical ways that shows up include:

• Sharing more transparent and efficient building practices.
• Supporting sustainability education and green software learning.
• Creating pathways for students and local residents to enter technical trades.
• Partnering with workforce and industry organizations that build long-term capacity.

Looking ahead, Mod42 is also exploring ways to give back in the communities where it builds. That could include initiatives such as tree planting, park enhancements, and other community projects that create lasting value beyond the infrastructure itself. Combined with the company’s support for workforce development, these efforts reflect Mod42’s belief that building digital infrastructure also means making a positive contribution to the communities it serves.

These efforts are not separate from infrastructure development. They are part of what makes that infrastructure more sustainable over time.

What’s Coming Next for Digital Infrastructure

Several trends are changing how digital infrastructure is planned and evaluated.

As AI infrastructure demand accelerates, communities are paying closer attention to how data centers are built, not just how they operate. At the same time, owners are looking for ways to reduce waste, shorten construction schedules, and minimize disruption without compromising quality.

Talent is another factor. As the industry grows, so does the need for skilled workers who can support design, manufacturing, and deployment at scale. Companies that make workforce quality a priority are likely to be better positioned for the future.

Taken together, these trends point in the same direction. The industry needs infrastructure models that are faster, more adaptable, and more considerate of both environmental and human impact.

The Path Forward

The future of digital infrastructure will depend on more than speed. It will depend on whether the industry can build in ways that are efficient, responsible, and respectful of the communities involved.

Sustainability, workforce, and community impact are no longer separate conversations. They are all part of delivering better digital infrastructure.  They are part of the same conversation. Reducing construction impact, supporting workers, and investing in local relationships all contribute to better infrastructure outcomes.

Mod42 is continuing to explore additional sustainability initiatives and industry partnerships that support long-term environmental stewardship, workforce development, and community engagement.

This is the perspective behind Mod42’s approach: Build infrastructure that can meet demand while also reducing disruption and supporting the people around it.

Ready for a faster, more sustainable approach to AI infrastructure deployment?

Connect with the Mod42 team to learn how volumetric modular construction can help reduce disruption, improve efficiency, and accelerate delivery. Get started today: mod42.com/quote.

The post Why Digital Infrastructure Needs a New Approach and Why This Matters Now appeared first on Data Center POST.

TL;DR

• Driven by AI requirements, global hyperscale data center construction is booming. U.S. construction spending reached an annualized $40 billion in June 2025, with major inventory surges across top North American markets.
• To support massive bandwidth and low latency, cabling is shifting from traditional 12-fiber trunks to ultra-high-density fabrics carrying up to 6,912 fibers, compressed into smaller spaces using emerging very-small-form-factor (VSFF) connectors.
• Operators are moving away from field termination to factory-built, 100% pre-tested assemblies.
• Modular, high-density interconnection strategies lower total cost of ownership (TCO) and maximize deployment speed, which has become a vital competitive requirement in the AI era.

# # #

Paulo Campos, President, R&M USA Inc.

The largest Hyperscale deployments are concentrated in the United States and China. Synergy Research Group reported that hyperscale operators ran 1,136 large data centers at the end of 2024, twice as many as five years earlier; the USA accounted for more than half of global critical IT load, with Europe and China each accounting for about a third of the balance.

Bank of America Institute and U.S. Census data shows that U.S. data-center construction spending reached an annualized $40 billion in June 2025, 30% higher year on year after a 50% surge in 2024. CBRE says inventory across the four largest North American markets — Northern Virginia, Atlanta, Dallas-Fort Worth and Chicago — rose 33% year on year in Q1 2026.

As a result of this growth and expansion, Hyperscale data centers are entering a brand-new cabling phase. AI and machine-learning clusters are demanding vast east-west bandwidth, low latency and repeatable build quality. What should DC operators bear in mind in order to optimize cabling for the next wave of AI driven requirements? For one thing, we’re seeing changes to the physical layer. In the past, many operators planned around 12-fiber trunks. Today’s AI fabrics may specify 6,912 fibers: 432 connectors per side, each carrying 16 fibers. Current hyperscale and 5G fiber counts can already reach 6,912, while SN-MT and MMC connectors are emerging to compress density into less space. (both connectors offer a 16-fiber connector with 2.7 times the density of 16-fiber MPO, and are a very-small-form-factor connectors that can be used in high-fiber-count interconnects, transceivers, and onboard optics).

The procurement model is changing too. Instead of buying bulk cable and leaving contractors to terminate on site, operators increasingly want assemblies manufactured to specification. If these are 100% tested in the factory, labeled, and delivered ready to install speed, quality control and supply-chain planning benefit: orders can be split across qualified facilities or partners, then combined without compromising documentation or consistency.

Pretermination can also significantly help reduce dependence on scarce specialist labor. Preterminated cabling can cut installation time by up to 80% compared with field termination. That matters during rollout, but also during changes and maintenance, because plug-and-play links are easier to identify, replace and restore.

Microtubes and blown fiber add another layer of flexibility and are rapidly being taken up by data centers. Microducts allow operators to install pathways first, then blow in fiber as architectures change. Dataintelo valued the air-blown fiber systems market at $3.8 billion in 2025 and forecasts $6.9 billion by 2034, driven by FTTH, 5G, data-center proliferation and modernization.

The best interconnection strategy is therefore modular, high-density and avoids splicing. Ideally, it  should combine preterminated trunks, VSFF multi-fiber connectors, microduct options, automated documentation, flexible manufacturing capacity and rigorous factory testing. High-density infrastructure can reduce TCO by saving space and operating cost, while modular cabling can be reconfigured or expanded without complete overhauls. That’s vital to hyperscale Data Center operators – in the AI-era, faster deployment is not a convenience; it is a competitive requirement.

The post Hyperscale Data Centers: Plug-and-Play Fiber for the AI Era appeared first on Data Center POST.

TL; DR

• AI workloads are changing how data centers think about battery infrastructure.
• Battery strategy is becoming a key part of scalable, AI-ready power design.
• Industry experts will discuss practical approaches to supporting AI growth while improving efficiency and resilience.
• Join DCD AI Week on July 28 to learn what’s next for AI-powered infrastructure.

# # #

As AI workloads push data centers toward higher density and faster power swings, operators are rethinking the role of battery infrastructure in supporting both backup power and increasingly dynamic AI workloads. On July 28, DCD will host a webinar for its AI Week to look at how smarter battery technology can help support scalable, AI-ready infrastructure. The discussion, “Enabling AI factory scalability through smarter battery technology,” will stream 1-2 p.m. EST on DCD’s Compute, Storage & Networking Channel.

AI is changing the rules for data center power. Traditional backup approaches were built for a different era, but today’s facilities need battery technologies that can handle more dynamic loads, tighter space constraints, and growing pressure to improve efficiency and resilience.

Rather than simply requiring more power, AI is introducing highly dynamic power demands that are prompting operators to rethink how infrastructure responds to rapid changes in load.

That shift is why this webinar matters now. Bringing together experts in power infrastructure, battery technology, and digital infrastructure strategy, the session will explore how next-generation battery technologies can better support AI-ready infrastructure. The discussion will examine topics including power density, performance, safety, sustainability, and deployability, as well as how battery-enabled infrastructure, modular system design, and emerging technologies can support GPU-intensive workloads while helping reduce both capital and operating costs.

Who:

• Thomas Schuldt, VP Business Development & Strategy, DG Matrix
  
• Brandon Smith, VP Global Sales & Product, ZincFive
• Stephen Worn, CTO & Managing Director, Datacenter Dynamics

What: “Enabling AI factory scalability through smarter battery technology” webinar

When: Tuesday, July 28, 1-2 p.m.

For operators, the core issue is practical: How do you scale AI compute while maintaining reliability, maximizing available space, and avoiding unnecessary infrastructure costs?  That’s where battery architecture becomes more than a backup decision and starts to shape overall infrastructure strategy.

Whether you’re designing new AI infrastructure, expanding existing capacity, or evaluating future power strategies, the session offers practical perspectives on one of the industry’s fastest-evolving challenges.

As AI infrastructure continues to evolve, understanding how battery technology fits into the broader power strategy is becoming increasingly important. To register for the conversation, visit:  www.datacenterdynamics.com/en/dcd-broadcasts/ai-week-2026/enabling-ai-factory-scalability-through-smarter-battery-technology.

The post AI Growth Is Forcing a New Look at Battery Strategy appeared first on Data Center POST.

TL;DR

• While Dallas-Fort Worth is one of the fastest-growing data center markets in North America, colocation vacancy has dropped to an all-time low of 1.8%, and 88% of the 700+ MW currently under construction is pre-leased.
• Hyperscalers dominate the market, structurally excluding mid-market AI operators, regional healthcare systems, and financial firms because developers prioritize multi-megawatt deals and design retail colocation for traditional IT rather than GPU workloads.
• Traditional air-cooling systems fail beyond 30 to 40 kW per rack, whereas modern GPU servers push densities to 80 to 130 kW, making immersion cooling an architectural necessity for next-generation AI workloads.

# # #

Dallas-Fort Worth is one of the largest and fastest-growing data center markets in North America. However, actually securing available space has become nearly impossible. According to CBRE’s latest market data, colocation vacancy in DFW has dropped to an all-time low of 1.8%, and roughly 88% of the more than 700 MW currently under construction is already leased out before it’s even launched.

This market works just fine for hyperscalers and large AI labs. They sign leases years in advance, lock down entire campuses, and negotiate directly with developers and utilities. Meanwhile, the reality for everyone else, including mid-market AI operators, regional healthcare systems, and regulated financial services firms, is completely different as they are structurally excluded.

Excluded by design, not by price

There are three layers working against mid-market buyers in DFW today. One of them is the deal size, as new developments are underwritten around anchor tenants taking tens of megawatts, so a requirement of one to five megawatts rarely gets picked up. On top of that, there’s the timing, with the majority of new supply preleased before groundbreaking, capacity never reaches the open market. And when retail colocation space does become available, it turns out to be largely designed for enterprise IT loads, not for GPU clusters.

“The mid-market has been structurally excluded from the infrastructure buildout happening in DFW. Hyperscalers are building for hyperscale. The minimums are too large, the architecture is shared, and the compliance posture doesn’t hold for regulated industries,” says Dennis Finster, Chief Executive Officer of EG AI Corp. “We built EG AI Corp to close that gap. Single-tenant, immersion-cooled, built specifically for the operator that needs dedicated infrastructure and cannot compromise on data sovereignty.”

The thermal ceiling

That need for dedicated, high-density infrastructure becomes more critical every year. Traditional air-cooled data halls were built to handle racks drawing 5 to 15 kW, and even optimized air-cooling designs will not perform well beyond 30 to 40 kW per rack. The problem is that modern GPU servers push rack densities to 80 to 130 kW, and sometimes even higher.

Immersion cooling offers a solution to this challenge. By submerging servers in a dielectric fluid that transfers heat far more efficiently than air, immersion systems support extreme rack densities while reducing cooling energy consumption, eliminating evaporative water use, and removing the fans and hot-aisle setups that tend to result in both noise and mechanical failure. For companies running continuous AI training or inference, this translates to more compute power per square meter and much more predictable thermal performance.

“Immersion cooling isn’t a feature, it’s an architectural decision that determines what the facility can actually do. Air-cooled environments have a ceiling on rack density that makes them structurally incompatible with next-generation GPU workloads,” says Conrad Eaton, CRO of EG AI Corp. “We designed this facility from the ground up around the thermal and compliance requirements of the operators we’re building for, and that starts with the cooling architecture before anything else goes in the ground.”

The compliance ceiling

For regulated industries like healthcare systems and financial services firms, thermal capacity is not the only obstacle. Frameworks such as HIPAA, GLBA, and PCI DSS require demonstrable control over physical access, audit boundaries, and data environments. In a shared colocation hall, those boundaries are negotiated and documented around other tenants’ operations. For many compliance teams, the complex auditing process outweighs the convenience of shared infrastructure.

Single-tenant architecture solves this problem completely with one tenant, one physical security perimeter, a clean audit boundary, and dedicated power and network paths. Compliance stops being an issue and becomes a property of the building itself.

What purpose-built looks like

A small number of projects are beginning to target this gap in the market. In Dallas, EG AI Corp broke ground in late May on a 5,000-square-meter single-tenant, immersion-cooled GPU data center, with completion targeted for 2027. The facility is being built to directly address the two main bottlenecks: immersion cooling architecture that eliminates the thermal ceiling of air-cooled systems, while the single-tenant model provides complete data isolation, avoiding shared physical infrastructure or third-party data exposure — a structural requirement for HIPAA, SEC, and fintech compliance workloads. It is built specifically for the mid-market operators, healthcare systems, and regulated financial firms the current buildout ignores, and  similar single-tenant, high-density models are likely to follow in other supply-constrained markets.

While the industry’s attention is fixed on gigawatt-scale hyperscale campuses, the middle market is where regional hospital networks train diagnostic models, where community financial institutions deploy AI under regulatory scrutiny, and where daily applied AI actually takes place. Meeting that demand will require infrastructure built for it, not whatever is left over.

Rendering of EG AI Corp’s immersion-cooled GPU data hall, Dallas, Texas.

The post The DFW Data Center Boom Has a Blind Spot: The Mid-Market appeared first on Data Center POST.

TL;DR

• The explosive growth of AI requires massive speed and scale, exposing the severe labor shortages and logistical bottlenecks of traditional data center construction.
• To increase efficiency and safety, Compu Dynamics Modular (CDM) is moving data center construction into controlled factory environments, assembling them like automobiles.
• To ease community concerns, the company is using closed-loop cooling systems that don’t evaporate local water. They also suggest rebranding these highly specialized facilities as “computer manufacturing sites.”

# # #

The digital infrastructure industry is moving at a breathtaking pace, driven by the massive power and density requirements of the artificial intelligence era. In episode 67 of the NEDAS Live!, host Ilissa Miller sat down with Steve Altizer, President and CEO of Compu Dynamics and Compu Dynamics Modular (CDM), to discuss how the industry must evolve to meet these unprecedented demands.

With more than two decades of experience in the data center ecosystem, Altizer shared his insights on why traditional construction methods are reaching their breaking point and how factory-built, modular infrastructure provides a critical strategic advantage for the future.

The Bottlenecks of Traditional Construction

The sheer speed and scale required by today’s hyperscalers and AI deployments have exposed the limitations of traditional data center construction. Building a massive 500,000 to 1 million square-foot facility often leads to coordination nightmares, with multiple trades fighting for space, staging equipment, and causing preventable delays.

Furthermore, the labor required to build these traditional campuses is becoming unsustainable. “Well, you’ve heard stories of Stargate and other large data centers where they have 7 or 8,000 people on a single campus,” Altizer noted, explaining that developers basically have to build temporary cities to house the workforce. Trying to staff these massive projects, especially outside of major metropolitan hubs, leads to extraordinary costs and logistical hurdles.

The Modular Revolution: Building Data Centers Like Cars

To overcome these bottlenecks, Altizer advocates for moving the complexities of data center construction into controlled factory environments. By building data centers on an assembly line, similar to the automobile industry, companies can drastically improve safety, efficiency, and speed.

However, Altizer is quick to clarify that today’s modular solutions are far beyond the retrofitted shipping containers of the past. Today’s modules are highly sophisticated, large-scale enclosures designed to handle massive amounts of cabling, liquid cooling, and high-density hardware. While a typical data center averages around 400 watts per square foot, CDM’s enclosures support staggering densities of 2,000 to 4,000 watts per square foot. By placing these optimized IT and power modules side by side, operators can rapidly assemble massively dense, gigawatt-sized data halls with a significantly smaller footprint.

Vertical Integration and a Major Merger

To realize this vision of speed and scalability, CDM recently acquired a majority stake in R&D Specialties, a highly qualified manufacturing facility based in Odessa, Texas. Finding that their teams and visions meshed perfectly, Altizer announced on the podcast that the two companies are officially merging into a single, vertically integrated entity.

Inspired by the vertical integration models of Elon Musk’s Tesla and SpaceX, Altizer explained that controlling the design, engineering, fabrication, and integration under one roof eliminates redundant overhead and saves the consumer money. Ultimately, CDM aims to standardize its product profiles so effectively that a customer could theoretically go to a website, select their power requirements, enter a credit card, and order a data center online.

Addressing Community Pushback Through Innovation

The conversation also tackled the growing community opposition to data center developments across the country. Altizer views this pushback not as a roadblock, but as a necessary catalyst for innovation. For example, to address community concerns regarding water scarcity, CDM utilizes closed-loop liquid cooling systems that do not evaporate and deplete local water resources.

Miller and Altizer also discussed the idea of rebranding these advanced facilities. Because modern AI infrastructure is highly specialized, referring to them as “computer manufacturing sites” rather than “data centers” could help communities better understand their function, predictability, and footprint.

To continue the conversation, listen to the full podcast episode here.

The post Building What’s Next: A Conversation with Steve Altizer on the Modular Revolution and the AI Data Center Boom appeared first on Data Center POST.

TL;DR

• New agreement: Duos Edge AI signed a five-year colocation agreement for 10 MW of critical IT-load capacity at its Columbus, Georgia campus.
• Revenue visibility: The agreement is valued at approximately $111 million in contracted revenue over five years.
• Campus growth: The deal is expected to expand Columbus capacity to 20 MW by the end of the fourth quarter of 2026.

# # #

Duos Technologies Group Inc. (Nasdaq: DUOT), through subsidiary Duos Edge AI, has signed a new five-year customer agreement that adds 10 MW of critical IT-load capacity at its Columbus, Georgia campus. The deal adds meaningful long-term contracted revenue while continuing to build out the company’s AI-ready infrastructure footprint.

With the new agreement in place, the Columbus campus is expected to grow to 20 MW of critical IT-load capacity by the end of the fourth quarter of 2026. Duos also expects its previously announced initial 10 MW deployment at the site to begin generating revenue in August.

Per Duos, the Columbus buildout is part of a broader strategy to own the underlying real estate and supporting infrastructure behind customer deployments. That approach is intended to create durable recurring revenue as demand for high-density AI capacity continues to scale.

“This agreement shows how we are investing in infrastructure in key markets so we can quickly add capacity for our customers,” said Duos CEO Doug Recker. “The Columbus campus gives us the ability to rapidly deploy high-density AI infrastructure while generating durable recurring revenue. We believe this model positions Duos to meet growing customer demand and create meaningful long-term value for our shareholders.”

The Columbus campus now supports a growing base of contracted deployments, including previously announced AI infrastructure agreements. Duos said the site reflects its continued focus on developing owned infrastructure in markets where long-term customer demand is building.

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

The post Duos Signs 5-Year, 10 MW Colocation Agreement Valued at ~$111 Million appeared first on Data Center POST.

TL;DR

• Legacy LAN cabling is struggling to keep pace with the high-speed connectivity and power demands driven by AI and advanced PoE devices.
• Organizations frequently delay infrastructure upgrades because active equipment is more visible, yet hardware cannot perform better than the underlying cable link.
• Upgrading legacy connectivity reduces troubleshooting, secures power delivery, and prevents expensive emergency network replacements.

# # #

Originally posted on Telecom Newsroom.

Local area network (LAN) infrastructures across the United States are facing unprecedented demands due to the rapid expansion of artificial intelligence, faster connectivity requirements, and high-power Power over Ethernet (PoE) demands. Many organizations delay essential cabling upgrades because their current networks seem functional or because they prioritize budgets for active equipment like switches and access points. However, active hardware cannot outperform the physical cabling layer underneath, and neglecting infrastructure upgrades leads to slow network speeds, unstable connections, overheated cable bundles, and underperforming Wi-Fi 7 systems.

The transition to higher Ethernet speeds (GBASE-T) and higher-power PoE classes (up to 90W) presents significant physical and technical risks to older brownfield installations. Legacy Cat.5e and Cat.6 cabling often fail to reliably support newer 2.5G, 5G, or 10G standards and can suffer from excessive direct current (DC) resistance unbalance, causing distorted signals, bit errors, and damaged contacts. To mitigate these issues, organizations must perform rigorous link certification testing, using frameworks like TIA’s TSB-5021, and adopt Cat.6A cabling as the default for new installations to ensure reliable high-speed performance and safe heat dissipation.

Rather than relying on non-standard, extended-reach copper solutions that increase installation weight and cost, forward-looking designs should utilize fiber optics or Single Pair Ethernet (SPE) for links exceeding 90 meters. Fiber delivers immune, virtually unlimited bandwidth over long distances, while SPE provides efficient, space-saving IP connectivity for automation. Ultimately, upgrading modular RJ45 connectivity and planning with zone cabling modernizes legacy systems, which prevents costly emergency replacements and fully unlocks the value of modern network electronics.

To read the original article, please click here.

The post Brownfield LAN in the USA: How Future-Proof is Today’s Cabling? appeared first on Data Center POST.

TL;DR

• Ecolab delivers “chip to chiller” cooling solutions globally to help operators manage power-intensive AI environments reliably and responsibly.
• In 2025, Ecolab helped its data center customers save 2.67 billion gallons of water, 40.5 million therms of energy, and 512,000 metric tons of CO2 emissions, driving over $61 million (USD) in total value.
• The company is driving a strategic shift from reactive maintenance to proactive, real-time data monitoring, supported by lifecycle programs like Cooling-as-a-Service to simplify deployment and reduce operational risk.

# # #

Data Center POST had the opportunity to connect with Matthew Branson, Vice President of Global Corporate Accounts, Data Center and Microelectronics for Ecolab, Inc. Over Matthew’s 17-year career with Ecolab, he has held a variety of positions of increasing responsibility in Marketing, Sales Operations, Corporate Account Sales Management, Field Sales Management and General Management in Ecolab’s Institutional and Food & Beverage Sectors.

Matthew is an active member of several Ecolab Employee Resource Groups, ERGs, and sits on the advisory board for the Black Leaders Forum (BLF) and EcoEssence. Matthew graduated from Michigan State University with a bachelor’s degree in Mechanical Engineering. He earned his MBA from the Carlson School of Management at the University of Minnesota.

What does Ecolab do?

At Ecolab, we help data center operators solve one of the biggest challenges in the AI era: how to cool increasingly dense, power-intensive environments reliably, efficiently, and responsibly. Cooling is among the largest drivers of both energy use and operational risk in a data center, directly impacting key metrics like PUE and uptime. We help customers improve cooling performance, protect uptime, and manage water and energy with confidence as demand accelerates. Ecolab’s data center business is one of 40 industries supported across 170 countries. A global leader in water, hygiene and infection prevention, Ecolab combines science‑based solutions, data‑driven insights, and world‑class service to drive performance and protect vital resources for millions of customers worldwide.

What problems does Ecolab solve in the market?

The market is moving fast. AI and high-performance computing are putting new pressure on data center infrastructure – especially around heat, energy, water, and reliability. Even small inefficiencies in cooling systems can drive meaningful increases in energy consumption and operating cost at scale. Our role is to help customers take complexity out of cooling, reduce operational risk, and create practical strategies that support growth without compromising uptime or resource stewardship. In 2025 alone, we helped data center customers worldwide save 2.67 billion gallons of water, 40.5 million therms of energy and 512,000 metric tons of CO2 emissions, valued together at more than $61 million (USD).

What are Ecolab’s core products or services?

Our portfolio supports data center cooling from the chip to the chiller, including Cooling-as-a-Service, advanced water management programs, field service expertise, digital monitoring, and technologies such as 3D TRASAR™ Technology for Direct-to-Chip Liquid Cooling. These solutions are designed to proactively identify issues, reduce rework, and accelerate commissioning timelines for new capacity. We help operators run more reliably and move faster as cooling requirements become more demanding.

With CoolIT, Ecolab is now uniquely positioned to offer end-to-end solutions across the full AI value chain, from ultra-pure water to produce the most advanced chips, to high-performing water systems for optimized power generation, to direct liquid cooling for the most powerful AI data centers. This further enables Ecolab to help customers scale AI faster with global solutions that protect people, natural resources, and communities.

What markets do you serve?

We serve data center operators globally, including hyperscale, colocation, and enterprise environments. Our core competency is helping customers scale AI infrastructure in regions facing power constraints, water stress, and/or aggressive deployment timelines, and where reliability and efficiency are critical to supporting mission-critical digital infrastructure.

What challenges does the global digital infrastructure industry face today?

AI demand is driving higher rack densities, greater cooling complexity, tighter speed-to-market expectations, and increased scrutiny on water and energy use. Demand is accelerating rapidly as data center capacity expands globally. At the same time, water and power are becoming more visible constraints as the industry scales. Operators need scalable, practical solutions built to protect reliability.

How is Ecolab adapting to these challenges?

We are adapting by driving a pivot from reactive maintenance to proactive, data-driven cooling management to bolster uptime, speed, resource efficiency, and operational confidence. PG25 coolant parameters, for example, often begin to shift before operational issues arise. Monitoring key fluid health indicators in real time can help operators catch and correct out-of-spec conditions before they escalate.

Ecolab combines field expertise, water science, chemistry, digital insight, and service to help customers manage cooling performance proactively instead of reacting after issues occur. With the recent acquisition of CoolIT, we’ll continue to work directly with our customers and bring together stakeholders across the power and water ecosystem to support the design of next-generation AI data centers that accelerate AI deployment while maximizing performance and minimizing impact.

What are Ecolab’s key differentiators?

Our biggest differentiator is the combination of deep water and process expertise with a global service model that operates where our customers operate. We have more than 100 years of experience helping customers run critical systems safely and efficiently. Even small improvements in cooling efficiency or reliability can translate into significant financial and operational impact at scale. We help customers manage cooling as a strategic operational lever, not just a utility.

What can we expect to see/hear from Ecolab in the future?

Ecolab will stay focused on practical innovation for the AI era. As the industry continues shifting toward liquid cooling, we are investing in solutions that support next-generation architectures. We will continue delivering solutions that help operators move faster, reduce risk, protect uptime, and manage water and energy more intelligently as cooling demands evolve.

What upcoming industry events will you be attending?

We will continue to stay close to the conversations shaping the future of digital infrastructure. You can expect to see Ecolab at DCD NOVA, OCP Global Summit, and SC26, where we will engage with operators, engineers, and technology partners on the future of data center cooling and next-generation operating models.

Do you have any recent news you would like us to highlight?

A key area of focus is our Cooling-as-a-Service program. It is designed to reduce the complexity of cooling management by supporting project planning, commissioning, start-up, ongoing service, and maintenance. The model improves speed to deployment, reduces operational risk, and delivers more predictable performance across the lifecycle of the asset, giving customers greater confidence and consistency as they scale.

In addition, at SC26 in November 2026, Ecolab plans to introduce a new end-to-end 3D TRASAR™ cooling platform combining CoolIT technologies, including coolant distribution units (CDUs) and high-performing cold plates, with integrated 3D TRASAR™ digital optimization and advanced cooling fluids. Supported by global expertise, the platform will optimize water, power, and compute performance at scale for next-gen AI architectures like NVIDIA Vera Rubin, Grace Blackwell, and more.

Is there anything else you would like our readers to know about your company and capabilities?

The scale and complexity of AI infrastructure is driving a more integrated approach to cooling, water, and energy management. The future of data centers will be shaped by strong partnerships between operators, technology providers, engineers, and service partners. Ecolab’s role is to bring practical expertise, disciplined execution, and collaborative problem-solving to help the industry scale efficiently, responsibly, and reliably.

The post Company Profile: How Ecolab Helps Data Centers Scale Responsibly appeared first on Data Center POST.

TL;DR

• Award-winning again: Empire Fiber Internet earned the Gold Winner title for Best IT/Office Technology in the 2026 FLX Finest Awards, its fourth consecutive year taking home the honor.
• Community-driven recognition: The win came from a pool of over 179,000 votes cast across more than 1,800 entrants competing in 214 categories throughout the Finger Lakes region.
• What it means locally: The recognition reflects the trust residents and businesses place in Empire Fiber Internet’s 100% fiber network and hometown customer service.

# # #

Empire Fiber Internet, a leading fiber optic internet service provider serving communities across New York and Pennsylvania, has been named Gold Winner of the FLX Finest Award for Best IT/Office Technology for the fourth year in a row. The community-voted award recognizes the businesses and residents shaping the Finger Lakes region, with this year’s contest drawing record participation across hundreds of categories.

The win underscores how much local customers value fast, dependable connectivity paired with a genuinely local approach to service. Empire Fiber Internet’s locally based teams and round-the-clock customer support continue to back its growing fiber network, helping homes, businesses, schools, and healthcare providers across the region stay reliably connected.

“We’re incredibly honored to receive this recognition from the Finger Lakes community for the fourth year in a row,” said Kevin Dickens, CEO of Empire Fiber Internet. “This award is especially meaningful because it comes directly from the people and businesses we serve every day. We’re proud to continue supporting this region’s growth with reliable connectivity, local support, and the exceptional customer experience our communities deserve.”

Empire Fiber Internet extends its thanks to everyone who voted this year. This recognition reinforces the company’s mission of strengthening Finger Lakes communities through fast, reliable fiber and hometown service.

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

The post Empire Fiber Internet Wins FLX Finest Gold for Fourth Straight Year appeared first on Data Center POST.

TL;DR

• Grand opening: Duos Edge AI will host a community open house for its Abilene communications hub on July 23 in Abilene, Texas
• Expansion milestone: The site reflects Duos Edge AI’s continued buildout of AI-ready infrastructure across underserved markets.
• Community event: Attendees can join a ribbon-cutting, tour the facility, and meet Duos Edge AI leadership and technology experts.

# # #

Duos Technologies Group Inc. (Nasdaq: DUOT), through subsidiary Duos Edge AI, is marking the public debut of its Abilene communications hub with a grand opening and community open house following the successful launch of operations. The event highlights the company’s continued expansion of AI-ready infrastructure across rural school districts and underserved markets.

The Abilene communications hub is part of Duos Edge AI’s growing network of modular sites designed to deliver scalable, high-performance computing closer to end users. By reducing latency and improving resiliency, the platform supports artificial intelligence, enterprise applications, education, healthcare, and other mission-critical workloads.

“Abilene is an important example of how Duos Edge AI is executing on its strategy to bring advanced AI-ready infrastructure closer to the communities and organizations that need it,” said Duos CEO Doug Recker. “This Grand Opening reflects our continued progress in expanding a scalable edge platform across underserved markets.”

During the event, attendees will be able to participate in the ribbon-cutting ceremony, tour the Edge facility, and meet Duos Edge AI leadership and technology experts. The open house is being hosted with Region 14 Education Service Center, the Abilene Chamber of Commerce, and the Abilene Industrial Foundation.

“Here at Region 14 ESC, we continually seek partners who are on the cutting edge of technology,” said Kevin Hill, Director of Technology Services at Region 14 Education Service Center. “With this partnership with Duos Edge AI, we hope to create even more opportunities for our schools, communities, and local businesses.”

To learn more about Duos Edge AI and register for the open house, visit https://freeevite.com/event.php?e=BFqIEVZmFZqz8t2x4_lvkg.

The post Duos Edge AI Celebrates Abilene Communications Hub Grand Opening appeared first on Data Center POST.

TL;DR

• Lightpath has secured a second infrastructure tenant for its Eastern Pennsylvania AI-grade fiber network.
• The new 10-year anchor agreement expands the planned regional network by approximately 30%.
• More than 60% of the combined Eastern Pennsylvania build is now committed, reflecting increasing demand for AI-ready fiber infrastructure.
• The expansion strengthens regional connectivity while supporting future hyperscale and enterprise growth.

# # #

Artificial intelligence is changing more than where data centers are built. It is also changing how the fiber networks supporting them are designed. As hyperscale development accelerates, network infrastructure is increasingly being planned at a regional level rather than around the needs of a single customer.

Lightpath’s latest expansion in Eastern Pennsylvania reflects this shift. The company has announced a second infrastructure tenant on its AI-grade fiber build in the Susquehanna region, expanding the planned network by roughly 30% through a new 10-year anchor agreement. Rather than constructing a separate network for each development, the regional build is designed to support multiple large-scale campuses while providing the flexibility to accommodate future growth.

This approach reflects how AI infrastructure is evolving. Today’s hyperscale campuses require far more than basic connectivity. They depend on resilient, high-capacity fiber capable of supporting massive volumes of data traffic between data centers, cloud platforms and enterprise networks. As more AI developments cluster around regions with available power and land, shared regional fiber infrastructure becomes increasingly valuable.

The announcement also highlights a broader trend across the AI infrastructure market. Network planning is shifting from reactive, customer-by-customer builds toward proactive investment in regions expected to experience sustained demand. By designing networks with future growth in mind, providers can scale more efficiently while giving customers faster access to the connectivity required for AI, cloud and data-intensive workloads.

Eastern Pennsylvania continues to emerge as one of those strategic regions. With growing hyperscale investment and access to major metropolitan markets, the area is becoming an increasingly important location for next-generation digital infrastructure. Lightpath’s network is positioned to provide high-bandwidth connectivity between Eastern Pennsylvania and key markets including New York, Ashburn and Columbus, with future connectivity extending to Chicago through the company’s recently announced long-haul corridor.

As AI infrastructure continues expanding across the United States, the supporting fiber network will play an equally important role in determining how quickly new campuses can come online. Building scalable regional infrastructure ahead of demand allows providers to support multiple generations of AI growth while creating a stronger foundation for the broader digital economy.

Read more in the press release here.

The post Lightpath Expands Eastern Pennsylvania Fiber Network to Support the Next Phase of AI Infrastructure appeared first on Data Center POST.

TL;DR

• The global edge data center market is projected to surge from USD 14.7 billion in 2025 to USD 71.9 billion by 2035 at a rapid 17.5% CAGR.
• Demand is fueled by the need for low-latency processing, real-time computing, near-source AI inference, and strict data residency compliance.
• Leading global providers are prioritizing modular deployments, energy efficiency, automation, and strategic telecom partnerships to handle rising workloads.

# # #

The global edge data center market was valued at USD 14.7 billion in 2025 and is estimated to grow at a CAGR of 17.5% to reach USD 71.9 billion by 2035, according to a recent report by Global Market Insights Inc.

Market expansion reflects the accelerating demand for low-latency digital infrastructure as real-time computing requirements continue to rise across connected ecosystems. Organizations increasingly require computing resources to be deployed closer to end users and data generation points to support instantaneous data processing, faster response times, and uninterrupted service delivery. The rapid rollout of next-generation connectivity and distributed computing frameworks is reshaping infrastructure strategies, encouraging decentralized architectures that reduce reliance on centralized cloud facilities.

As data volumes surge, localized processing enables enterprises to manage workloads more efficiently while reducing network congestion and improving operational continuity. Artificial intelligence adoption is further reinforcing the need for geographically distributed computing, as near-source inference enhances speed, accuracy, and reliability. Additionally, compliance requirements related to data residency and governance are pushing enterprises and public institutions to store and process information within defined geographic boundaries, reinforcing the strategic importance of edge data centers in modern digital ecosystems.

The edge data center market from the solution-as-a-service segment accounted for 74% share in 2025 and is forecast to reach USD 50.3 billion by 2035. This dominance is driven by demand for integrated platforms that enable seamless coordination across distributed infrastructure while ensuring consistent performance, scalability, and operational resilience.

The hyperscale and enterprise data center segment held a 45% share in 2025 and is expected to reach USD 32.1 billion by 2035. Large operators continue to deploy smaller, distributed facilities closer to population and data hubs to improve latency performance, workload resilience, and computing efficiency.

China Edge Data Center Market is projected to reach USD 9.7 billion by 2035. Policy support, digital infrastructure expansion, and national technology initiatives are accelerating localized data processing investments.

Key companies active in the edge data center market include Schneider Electric, Dell Technologies, Vertiv, Huawei Technologies, Hewlett Packard Enterprise Company, Eaton, IBM, Fujitsu, EdgeConnex, Rittal, Flexential, Vapor, and EdgePresence. Companies operating in the global edge data center industry are strengthening their market position through strategic infrastructure expansion, modular deployment models, and technology integration. Providers are investing in scalable and standardized solutions to accelerate deployment timelines and reduce operational complexity. Partnerships with telecom operators, cloud platforms, and enterprise clients support ecosystem integration and long-term contracts. Firms are also prioritizing energy efficiency, automation, and remote management capabilities to improve reliability and cost performance.

The post Edge Data Center Market to Surpass USD 71.9 Billion by 2035 appeared first on Data Center POST.

TL;DR

• Implementing automation cuts manual errors by 60% and slashes server provisioning times by 80%, reducing setup tasks that once took several hours down to under 30 minutes.
• Nearly 70% of IT infrastructure failures are caused by human mistakes, equipment issues, or delayed troubleshooting responses. Automated monitoring detects issues instantly to trigger pre-established recovery processes, reducing response times from hours to minutes.
• Cooling represents up to 40% of total data center energy consumption. Smart automation dynamically manages cooling, airflow, and resources to handle the extreme power densities and thermal loads created by massive AI clusters and edge computing deployments.

# # #

The quick growth of cloud computing, artificial intelligence (AI), edge computing, and digital services is changing how today’s data centers work. Canters that used to depend on human control have to now oversee thousands of servers, sophisticated networking equipment, and escalating workloads with the least downtimes possible. With systems getting bigger and more dispersed, it is more difficult to stick to old management approaches than before.

As reported by Data Intelo, the world’s data center automation market was worth $12.8bn in 2025 and is expected to grow to $42.6bn by 2034, showing considerable CAGR (14.3%) within the 2026-2034 period. The high market dynamics indicate growing demand for smart management of infrastructure in line with IT advances.

Experts estimate that global data center capacity increases by 15% or more each year, while enterprise data increases by over 25% every year. Such rapid growth means that automated approaches are becoming a must instead of an additional facility. Companies use automation to get the advantages of stabilized operations, optimized resource use, and increased resilience of infrastructure without increasing complexities in operations.

Increased Complexity of Infrastructure Calls for Smart Operations

Data centers of today do not consist only of several server racks used for running internal business processes. Current facilities have thousands of servers to monitor, along with virtual machines, containers, networks defined by software and environments based on hybrid clouds.

Hyperscale data centers have smartphones and tablets and therefore are capable of hosting the work of over 100,000 servers simultaneously. Even medium-sized companies have about 5,000 devices, and finding an effective way to monitor all of them becomes incredibly challenging.

Automation platforms are responsible for monitoring the performance of infrastructure, actions are taken without human intervention. Automation capabilities allow to minimize the number of repetitive processes, thus allowing engineers to focus on planning, development and protection.

Boosting Operational Efficiency Using Automation

Automation has been crucial in minimizing the amount of maintenance tasks performed while also increasing the reliability of infrastructure. Many repetitive tasks like server provisioning, software deployment, backup scheduling, and resource allocation can now be done automatically in minutes instead of hours.

According to industry estimates the use of automation technologies can cut provisioning times by as much as 80% while automation software can decrease the number of manual errors by 60% and speed up deployment which in turn speeds up application delivery.

The following picture illustrates the role of automation in different operational areas.

These improvements help keep high levels of service provided by infrastructure teams despite the growth of digital loads.

The Importance of Automation in Ensuring System Reliability and Availability

Unless they haven’t experienced an unexpected outage of their operation, data center owners understand that it has disastrous consequences and can affect business applications, cloud services, banking systems, and even customers’ experiences.

Research shows that nearly 70% of IT infrastructure failures are caused by human mistakes, equipment failures, or untimely measures undertaken in response to occurring glitches in the operation of data centers. With the introduction of automated monitoring systems, such regularity is eliminated, as the systems detect performance deviation at once and start executing pre-established recovery processes when problems arise.

Some of the regular functions of automated monitoring that ensure reliability of IT operations are:

• Constant health monitoring of servers, storages, and networks
• Automatic distribution of workloads in case of excess demand
• Predictive maintenance based on the hardware operation history
• Automatic switching to a backup equipment in case of failure
• Immediate notification in cases of vital events

The use of automation reduces response time from hours to minutes and increases reliability as well.

Energy Efficiency is Becoming a Priority Area

Electricity costs have become one of the biggest running costs for today’s data centers. These facilities can demand between 20 to 100 megawatts of electricity during operation, with cooling systems often representing up to 40% of total energy consumption.

Automation enables the intelligent management of electricity through constant manipulation of cooling systems, airflow, workload placement, and server use. The trend towards sustainability now forces automation to help energy efficiency versus the growing needs of computing.

Assistance AI, Cloud as well as Edge Computing Growth

Artificial intelligence as well as edge computing create some new operational challenges for traditional management of infrastructure since it is hard to cope with such challenges. Ah, that would mean that if AI training clusters need thousands of GPU working at the same time, that would create much greater power density as well as thermal loads than that of traditional computing venues.

World spending on AI infrastructure is expected to exceed hundreds of billions in times of the next decade while various deployments of edge computing are continuing on the way expanding through many spheres of our existence including manufacturing, healthcare, telecommunications and smart cities.

The process of automation allows infrastructures to adapt to their environment dynamically. That is to say that they are able to manage computing resources, manage the load on the system, and cool the system without any interference from people.

Anticipating the Future

The new data center of the future does not only require computing power. It also requires smart systems capable of managing increasing operational complexity. As cloud services, AI operations, and edge deployments continue to grow, automation evolves into an essential element of contemporary data center strategy.

Automation is not meant to replace human specialists. It is meant to free specialists from routine work and allow them to use their talent and skills for more important engineering decisions. As infrastructure is likely to get more distributed and become more data-driven in the next few years, automated operations are likely to remain at the core of building reliable, sustainable, and scalable data centers.

# # #

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 Why Data Center Automation Is Essential for Next-Generation Infrastructure appeared first on Data Center POST.

TL;DR

• Community-driven collaboration can deliver more sustainable infrastructure solutions.
• Local knowledge is essential when addressing complex infrastructure challenges.
• Scalable models can improve public health and quality of life in underserved communities.
• Successful infrastructure projects balance engineering expertise with community engagement.

# # #

Written by iMiller Public Relations on behalf of Nomad Futurist.

Community collaboration helps create lasting solutions.

Infrastructure challenges are often viewed as technical problems, but lasting solutions require community involvement. In the latest Nomad Futurist article, Student Advocate Prathyaj Mantha highlights how Alabama’s Black Belt Unincorporated Wastewater Program (BBUWP) is transforming rural sanitation through collaboration among residents, engineers, environmental health professionals, and government partners.

Rather than applying one-size-fits-all solutions, the program addresses local challenges with community-driven planning and tailored infrastructure. The result is a scalable model that improves public health while demonstrating how collaboration can create sustainable, long-term impact for underserved communities.

This content originally appeared on the Nomad Futurist website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post It Takes a Community to Solve a Problem appeared first on Data Center POST.

TL;DR

• Replacing sequential on-site construction with parallel factory production allows modular units to deliver in just 8-16 weeks, compared to the 18-36 months required for traditional builds.
• Modular infrastructure is roughly 15-25% cheaper to construct, with total project costs (including financing and risk-adjusted contingency) favoring modular by 15-25% for deployments below 10 MW. This rapid deployment significantly reduces exposure to specification obsolescence driven by fast-evolving AI workloads.
• Rather than assembling separate components on-site, modular units integrate power, cooling, racks, and security. These systems undergo full integration testing in a controlled factory environment before delivery, eliminating the most common sources of on-site commissioning delays.
• While traditional builds remain the most cost-effective choice for massive hyperscale projects above 50 MW, modular architectures are ideal for timeline-constrained projects, phased capital deployments, and sites with physical, geographical, or logistical limitations.

# # #

What Is a Modular Data Center, and How Does It Differ from a Traditional Build?

A modular data center refers to a pre-engineered, factory-built infrastructure unit that delivers IT capacity, power distribution, precision cooling, racks, monitoring software, and physical security as an integrated system rather than as separately specified components assembled on-site.

A traditional brick-and-mortar build refers to constructing a data center facility in place, with civil works, electrical infrastructure, mechanical cooling plant, fire suppression, and physical security all designed, procured, and installed sequentially at the deployment site.

5 Key Differences Between Modular and Traditional Data Center Builds

The term “modular” is applied inconsistently across the industry. The precise definition is one where the core build work happens in a controlled factory environment, with on-site activity limited to foundation preparation, utility connection, and final commissioning.

Modular data center solutions significantly differ from traditional brick-and-mortar builds in cost and speed, with 60-75% faster deployment timelines compared to 18-36 months in traditional brick and mortar builds and around 15-25% cheaper infrastructure costs. This makes it an ideal choice for many companies starting out or seeking cost-effective solutions.

Why Is the Industry Moving Toward Modular Deployment Now?

The data center industry is operating under a deployment clock it has not had to watch this closely before. According to the International Energy Agency (IEA, 2024), global data center capacity is projected to more than double between 2024 and 2030, with AI workloads driving the majority of that growth. The window in which infrastructure decisions are tolerated as “in progress” has collapsed.

A facility that takes two years to come online today delivers its value into a market that has already moved past the assumptions it was designed around. This is the structural reason modular has moved from a niche option to a mainstream choice for hyperscalers, enterprises, and edge operators.

Did You Know?

AI rack densities have increased from an industry average of 8-15 kW per rack in 2020 to deployments now routinely exceeding 70-132 kW per rack, meaning infrastructure specifications written even 18 months ago may no longer match the workloads they were intended to support by the time a traditional build is completed.

How Do Modular Data Centers Cut Deployment Timelines by 30-50%?

The timeline compression is not a single saving; it is the cumulative effect of removing several sequential delays inherent in traditional construction. According to Uptime Institute’s deployment analysis (2024), the timeline difference between factory-built and stick-built data center projects of equivalent capacity ranges from 30% to 50% in favor of modular projects.

The compression comes from four specific mechanisms:

1. Parallel rather than sequential workflows. In a traditional build, design, procurement, civil works, electrical installation, and cooling commissioning largely occur in sequence. In a modular deployment, factory production runs in parallel with on-site preparation. The two streams converge at delivery rather than alternating across an extended timeline.
2. Factory testing replaces on-site commissioning iterations. Integration problems between independently installed power, cooling, and monitoring systems are typically discovered during on-site commissioning in traditional builds. Modular units complete this integration testing in the factory before delivery, eliminating the most common source of commissioning delay.
3. Reduced civil and structural lead times. Permitting cycles in urban environments routinely exceed six months for traditional builds. Modular deployments require significantly less civil work because the infrastructure unit provides its own structural enclosure, environmental control, and physical security.
4. Standardization removes specification cycles. Modular units use pre-validated reference designs where infrastructure calculations have already been performed. The design timeline collapses from months to weeks because the design choices have largely been made and tested in prior deployments.

What Does the Cost Comparison Actually Look Like?

The cost discussion around modular versus traditional builds is more nuanced than the timeline discussion. On a per-megawatt basis, large traditional builds at scale can achieve lower unit capital costs than modular equivalents, particularly at 50 MW and above, where extensive customization is justified by deployment size.

The cost advantage of modular shows up under different conditions:

• Deployments at the 1-10 MW scale where traditional construction overheads do not amortize efficiently
• Deployments where time-to-revenue matters more than absolute unit cost
• Deployments where capital must be released in phases rather than committed upfront
• Deployments where physical site constraints make traditional construction impractical

According to a 2024 451 Research analysis of mid-scale data center deployments, total project cost, including financing, delayed revenue, and risk-adjusted contingency, favors modular by 15-25% for deployments below 10 MW even when raw unit costs are comparable.

Did You Know?

A six-month deployment timeline difference is not just a six-month delay in revenue; it is also six months of avoided exposure to specification obsolescence. AI workload projections from twelve months ago routinely look conservative today, meaning faster deployments carry less risk of being commissioned into a market that has already moved past the original design assumptions.

What Does a Modular Data Center Architecture Actually Include?

Several vendors operate in the modular data center space, and their architectural choices vary meaningfully. The Schneider Electric prefabricated modular data center range provides a useful reference point for what a full modular architecture typically covers.

The EcoStruxure Row Data Center configurations integrate the following components into pre-engineered, factory-tested units:

• Racks: NetShelter SX enclosures with cable management and aisle containment
• Power distribution: Metered rack PDUs (Power Distribution Units) with outlet-level monitoring
• UPS (Uninterruptible Power Supply): Rack-mount UPS in N+N redundancy with 10-minute backup at full load
• Cooling: InRow DX precision cooling in N+1 redundancy
• Management: EcoStruxure IT monitoring software for remote visibility across all components
• Physical security: Biometric access control, CCTV, and environmental sensors
• Fire suppression: Integrated detection and suppression systems

For deployments at scale, prefabricated power, cooling, and IT modules can be combined to deliver megawatt-class capacity with the same factory-built advantages. Reference designs co-developed with NVIDIA support AI rack densities up to 132 kW per rack, demonstrating that modular architectures are not limited to lower-density traditional workloads.

When Is Modular the Right Choice, and When Is Traditional Still Better?

The modular-versus-traditional question is rarely about which approach is universally better. It is about which approach matches the specific constraints of the deployment in question.

Four scenarios consistently favor modular deployment:

1. Timeline-constrained deployments where lease expiries, support contract endings, regulatory deadlines, or AI rollout commitments make a traditional construction timeline structurally incompatible with the business requirement.
2. Phased capacity scenarios where the deployment will scale incrementally over multiple years and the operator wants to align capital deployment with actual demand growth.
3. Site-constrained deployments with limited land, leased premises, multi-tenant buildings, or remote locations where traditional construction logistics are difficult.
4. Operator capability constraints, where the organization lacks in-house engineering capacity to manage a traditional build and prefers to consume infrastructure as a delivered solution.

Traditional builds remain the right choice for hyperscale deployments at 50 MW and above, where unit economics at scale outweigh timeline pressures; for highly customized facilities with unique architectural or operational requirements; and for organizations with established construction expertise and long-horizon capital deployment models.

Which Vendors Lead the Modular Data Center Market in 2026?

The modular data center vendor landscape has consolidated around several established providers, with meaningful differences in product range, scale capability, and integration depth.

The following comparison reflects vendor positioning based on publicly available product information, deployment case studies, and industry analyst coverage as of 2026:

Schneider Electric

• Capacity Range: 10 kW – multi-MW
• Differentiator: Full-stack integration: NetShelter racks, Galaxy UPS, InRow cooling, EcoStruxure IT monitoring, validated AI reference designs with NVIDIA up to 132 kW per rack
• Best Fit For: Enterprises and operators seeking end-to-end pre-integrated solutions with global service infrastructure

Seimens

• Capacity Range: 10 kW – multi-MW
• Differentiator: Industrial automation and digital twin capability via Siemens Xcelerator, strong building management system integration, and SIVACOM modular electrification platforms
• Best Fit For: Hyperscale and industrial deployments where automation, digital twin orchestration, and integrated building infrastructure are primary design drivers

ABB

• Capacity Range: Up to 1 MW per module
• Differentiator: Strong electrification heritage with HiPerGuard and DPA UPS portfolios, integrated medium-voltage switchgear, and a focus on grid-interactive modular power infrastructure
• Best Fit For: Operators prioritising electrical infrastructure depth, grid integration, and high-availability power architecture in mission-critical environments

Huawei

• Capacity Range: 10 kW – multi-MW
• Differentiator: FusionModule range with strong presence in Asia-Pacific markets; integrated cooling and IT modules
• Best Fit For: Operators in Asia-Pacific regions where Huawei’s deployment footprint is strongest

Stulz

• Capacity Range: 10 kW – mid-scale
• Differentiator: Cooling-led architecture with precision cooling expertise; CyberRow and CyberHandler product lines
• Best Fit For: Deployments where cooling specification is the primary design constraint

Evaluation should always be supplemented by direct reference checks and site visits to meet specific deployment requirements.

What Should Decision-Makers Take from This?

The 50% timeline compression is not the whole story of why modular has moved from niche to mainstream. Still, it is consistently the headline reason that justifies the deeper conversation about which approach is right for a specific deployment.

The infrastructure decisions that match the pace of AI-driven capacity demand are the ones that will look right in retrospect. For most deployments below 10 MW, edge and enterprise capacity expansions, and any scenario where time-to-deploy is materially constrained, modular data centers are the architecture that aligns with the speed at which the industry is now required to operate.

# # #

About the Author

Mayank Srivastav is Segment Director for Cloud & Service Providers (C&SP) at Schneider Electric, where he drives segment strategy and strategic client relationships across India’s cloud, colocation, and managed service provider market. With extensive experience across service delivery, sales leadership, business planning, and operations management, Mayank advises India’s leading cloud and AI infrastructure organisations on scalable modular power architecture and infrastructure planning for AI workload growth. He is a graduate of Jamnalal Bajaj Institute of Management, Mumbai University.

The post Speed-to-Deploy: How Modular Data Centers Cut Deployment Timelines by Up to 50% appeared first on Data Center POST.

TL;DR

• Operators must navigate high-density workloads, supply chain limits, and labor shortages that make traditional scaling timelines increasingly difficult to maintain.
• Power, cooling, and controls can no longer be treated as isolated workstreams; they must be integrated early to avoid late-stage bottlenecks.
• Factory-built modular methods allow teams to standardize core elements, reduce on-site complexity, and scale predictably phase-by-phase.

# # #

Written by iMiller Public Relations on behalf of Compu Dynamics.

Data center operators in the AI era face the critical challenge of rapidly adding digital infrastructure capacity without overbuilding or introducing operational risk. Traditional construction timelines are increasingly strained by supply chain constraints, limited labor availability, and complex site readiness issues. These pressures are particularly acute in artificial intelligence and high-performance computing environments, which require infrastructure capable of supporting extreme power densities, complex liquid cooling, and strict performance metrics.

To overcome these modern speed barriers, operators must shift away from slow, sequential on-site construction phases. Speed in today’s market is no longer about saving a few weeks, but rather shaving entire months off deployment schedules. Achieving this level of acceleration requires early, integrated coordination of highly interdependent systems to prevent costly, late-stage bottlenecks.

A hybrid approach combining traditional integration with modular infrastructure offers a highly flexible path forward. While traditional builds remain essential for certain environments, factory-built modular methods allow operators to standardize critical elements while customizing density, cooling, and layouts. This approach minimizes on-site complexity, enables predictable, phased expansion, and provides the agility needed to respond to urgent demand without sacrificing planning discipline.

This content originally appeared on the Compu Dynamics website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post Navigating Speed and Scale Challenges in the AI Era appeared first on Data Center POST.

TL;DR

• Network expansion: Harbor Link is extending its underground fiber network into CoreSite’s Washington, D.C. campus.
• Greater resiliency: The deployment adds new carrier diversity across multiple data center facilities.
• Regional connectivity: The expansion strengthens connectivity options for enterprises, carriers, cloud providers, and network operators.

Reliable connectivity depends on more than bandwidth alone. Diverse network paths improve resiliency, reduce risk, and provide organizations with greater flexibility as infrastructure requirements continue to evolve.

Harbor Link is expanding its underground fiber network into CoreSite’s DC2 data center, extending connectivity across the company’s Washington, D.C. campus and Northern Virginia facilities. The deployment will provide additional carrier diversity across DC2, DC1, VA1, VA2, and VA3, enhancing one of the country’s most interconnected data center ecosystems.

Purpose-built underground infrastructure has become increasingly valuable in dense metropolitan markets where adding new network routes can be challenging. By creating additional physical pathways beneath rivers, highways, and other major crossings, providers can improve network resiliency while giving customers more options for connecting critical workloads.

“This expansion brings additional fiber diversity directly into CoreSite DC2, which is a key interconnection point in the region,” said Felix Dialoiso, Founding Partner at Harbor Link. “It strengthens the options available to customers that rely on scalable, low-latency connectivity across the Washington, D.C. metro ecosystem.”

The Washington, D.C. and Northern Virginia corridor continues to be one of the nation’s most significant interconnection markets, supporting a dense concentration of enterprises, cloud providers, carriers, and government organizations. As demand for connectivity grows, additional route diversity helps organizations strengthen redundancy while maintaining access to highly interconnected environments.

Harbor Link’s latest deployment reflects the continued investment in underground fiber infrastructure needed to support future connectivity requirements across the Mid-Atlantic. By expanding access into CoreSite’s regional campus, the company is adding another pathway for organizations seeking resilient, high-capacity connectivity throughout the Washington metropolitan area.

Additional details are available in the full announcement here.

The post Harbor Link Expands Underground Fiber Network into CoreSite’s Washington, D.C. Campus appeared first on Data Center POST.

TL;DR

• Unprecedented electricity demands from massive AI workloads, where individual server racks can exceed 100 kW, are overwhelming traditional electrical architectures, making robust power design a key differentiator for hyperscale data centers.
• Operators are shifting to “super-grid” transformers that connect directly to high-voltage transmission networks (at 380 kV or above). This direct link bypasses local distribution networks, dramatically reducing transmission energy losses and boosting overall reliability.
• Connecting directly to the transmission grid allows the Wustermark site to operate on 100% certified renewable energy, utilize 2N electrical redundancy, and support a “zero-generator” backup strategy.

 # # #

As artificial intelligence (AI) reshapes the global digital economy, the conversation around data centers often focuses on graphics processing units (GPUs), liquid cooling, and high-density compute. Yet behind every AI workload sits a less visible but critical piece of technology – the transformer.

These systems rarely attract headlines, but they have become foundational to the future of hyperscale infrastructure. Transformers regulate voltage, stabilize power delivery, and enable electricity to move efficiently from generation sources to compute environments. Without them, large-scale AI deployment simply cannot happen.

Today’s hyperscale campuses consume unprecedented amounts of electricity. AI training clusters operate continuously at massive scale, while rack densities continue to climb well beyond traditional enterprise thresholds. In this environment, power infrastructure is no longer a utility consideration buried in the background of facility design. Instead, it has become a strategic differentiator.

This shift is particularly visible in Europe, where data center operators are increasingly redesigning electrical architecture around transmission-level connectivity, renewable integration, and long-term sustainability goals.

Why traditional electrical architectures are under pressure

Transformer technology is based on the electromagnetic induction principle first demonstrated by Michael Faraday in the nineteenth century. Alternating current flowing through a primary winding creates a magnetic field that induces voltage in a secondary winding. By adjusting winding ratios, transformers either increase or decrease voltage depending on operational requirements.

That capability is central to modern power transmission. Electricity transported at higher voltages experiences lower current flow, dramatically reducing load related energy losses during transmission. Since power loss rises in proportion to the square of the current, minimizing current becomes essential for efficiency at scale.

Conventional data center infrastructure has historically relied on multiple stages of voltage transformation between the utility grid and the server floor. However, AI-scale environments are exposing the limitations of this approach.

Modern AI workloads require enormous, uninterrupted power capacity. Individual racks can now exceed 100 kW, while large training environments consume tens of megawatts continuously. Every inefficiency across the electrical chain increases operational costs, cooling demands, and carbon emissions.

As a result, operators are moving toward transmission-level infrastructure strategies that reduce transformation stages and improve overall efficiency.

The rise of super-grid transformers

A new generation of “super-grid” transformers is emerging as the preferred solution for hyperscale campuses. Unlike conventional transformers connected through regional distribution networks, these systems connect directly to high-voltage transmission infrastructure, often at 380 kV or above.

This approach fundamentally changes how large data centers interact with the power grid.

By connecting directly to transmission networks, operators can reduce electrical losses, improve reliability, and access significantly greater capacity. The architecture also enables tighter voltage regulation and enhanced operational resilience.

Super-grid transformers themselves are enormous engineering assets. Typically oil-filled for cooling and insulation, they are designed for decades of continuous operation under fluctuating load conditions. Their size reflects the scale of the challenge they are intended to solve – that is to support AI-driven infrastructure growth without compromising efficiency or stability.

The importance of this shift cannot be overstated. Global data center electricity demand is projected to rise dramatically over the next decade as cloud computing and AI adoption accelerate. Legacy electrical models were simply not designed for this level of sustained consumption.

For hyperscalers, securing scalable and reliable power access has become as important as securing land or fiber connectivity.

Wustermark: A new blueprint for AI infrastructure

One of the clearest examples of this evolution is the new hyperscale campus being developed west of Berlin by VIRTUS Data Centres.

The Wustermark campus is designed around direct connection to the 50Hertz 380 kV transmission network, enabling an initial deployment of 300 MW with future expansion capability scaling to 500 MW.

At the center of the project are two 185 MVA super-grid transformers, among the largest deployed within a European data center environment.

The significance of the project lies not only in its scale, but also in its architecture. By operating at the transmission level, the campus minimizes dependency on local distribution networks, avoiding strain on residential and commercial energy users while improving overall reliability.

The site is also engineered around full 2N electrical redundancy, providing dual independent paths for critical infrastructure systems. Combined with the highest transmission-level connectivity, this creates exceptionally high levels of resilience and availability.

Equally important is the operational flexibility the design enables. While diesel backup generation capability remains available, customers can also pursue a zero-generator strategy supported by the stability of the transmission network and renewable energy integration.

The campus will operate using 100% certified renewable energy, aligning high-performance compute infrastructure with Europe’s long-term decarbonization objectives.

The infrastructure behind the AI economy

The future of AI will depend not only on advances in semiconductors and software, but also on the ability to deliver enormous quantities of reliable, sustainable electricity.

Transformers may not command the same attention as GPUs or AI models, but they are becoming one of the most strategically important components in digital infrastructure.

Projects such as Wustermark demonstrate that it is possible to combine hyperscale growth, operational resilience, and environmental responsibility within a single infrastructure model.

As AI adoption accelerates globally, the operators that master power architecture, transmission-level connectivity, and renewable integration will define the next generation of digital infrastructure leadership.

The transformer, once considered basic utility equipment, is rapidly becoming the backbone of the AI economy.

# # #

About the Author

Mike Golding, SVP Design & Build at VIRTUS Data Centres, leads Design and Construction activities in EMEA. He is a member of the VIRTUS executive leadership team, bringing over 30 years’ experience in the wider construction industry and 20 years of specific data center design and delivery expertise delivering over 750MW of capacity for lease providers and self-build programs of work during this time.  Mike is a strong advocate of construction safety, scheduling predictability, sustainability & quality and brings deep experience of end-to-end data center delivery in established and new geo expansions in EMEA.

The post The Super-Grid Transformer Revolution Powering AI Data Centers appeared first on Data Center POST.

TL:DR

• While early discussions focused mostly on land, power, and water, companies are increasingly forced to integrate logistics directly into their broader deployment, maintenance, and expansion plans.
• Because AI infrastructure projects must move much faster than traditional development cycles, logistics providers must offer accelerated, secure, and highly reliable deployment models to bring capacity online quickly.
• Companies are moving away from purely centralized distribution models in favor of regional logistics hubs near major infrastructure corridors to drastically reduce transit times and keep critical inventory close to data centers.

# # #

The rapid growth of artificial intelligence is driving unprecedented investment in digital infrastructure. Across the United States, hyperscalers and AI infrastructure providers are expanding their footprints to support growing demand for computing power, creating new opportunities for regions that can support large-scale data center development.

While discussions around AI infrastructure often focus on availability of land and technology or resources like power and water, logistics is becoming an increasingly important part of the equation. As data center development accelerates, companies are rethinking how supply chains, inventory, transportation, and deployment services support infrastructure growth. This shift is changing logistics strategy in several important ways.

Speed-to-deployment is becoming a competitive advantage

AI infrastructure projects are moving at a different pace than traditional development cycles. Organizations are under pressure to bring new capacity online quickly, creating demand for logistics models that can support faster deployment timelines.

As a result, logistics providers are being asked to do more than move products from one location to another. They must help ensure critical equipment is available when and where it is needed, while maintaining security and reliability throughout the process. The ability to support accelerated deployment timelines is becoming a key differentiator as infrastructure investment continues to grow.

As data center development expands across the country, logistics networks are evolving alongside it. Rather than relying solely on centralized distribution models, companies are increasingly evaluating how regional logistics hubs can support emerging data center markets. Facilities located near major infrastructure corridors can help reduce transit times, improve responsiveness, and position critical inventory closer to deployment sites. This approach reflects a broader shift toward logistics networks designed around proximity, flexibility, and speed.

Hyperscalers have new expectations for logistics partners

The needs of hyperscalers and AI infrastructure companies differ significantly from those of traditional logistics customers. Data center environments often require specialized handling, secure storage, coordinated deliveries, and services that support equipment throughout its lifecycle. As a result, logistics providers are increasingly expected to manage complex infrastructure deployments while meeting stringent operational and security requirements. With continued growth in AI infrastructure investment, these expectations are becoming a more prominent part of the logistics landscape.

Logistics is becoming part of infrastructure strategy

As AI investments accelerate, logistics is increasingly embedded in broader infrastructure planning. Organizations are evaluating not only where to build, but also how to support deployment, maintenance, and future expansion. The ability to position inventory strategically, manage complex delivery requirements, and enable faster deployment timelines is becoming an important consideration in how infrastructure projects are designed and executed.

The next phase of AI growth will require more than computing power and physical facilities. It will also depend on logistics networks that bring that infrastructure online efficiently and at scale.

# # #

About the Author

Andreas Podwojewski serves as Managing Director for Arvato North America & Brazil, leading the organization and overseeing strategic and operational performance across the company’s Americas footprint. Based in Pleasant Prairie, Wisconsin, he is responsible for driving growth, accelerating operational excellence, and ensuring cross‑regional alignment with Arvato’s global strategy.

Andreas provides executive leadership on enterprise‑critical initiatives, including footprint expansion, automation, and digital transformation, designed to embed continuous improvement and long‑term strategic discipline across the organization. With deep experience leading high‑impact teams and complex operations, Andreas brings a global perspective and forward‑looking leadership philosophy.

The post How AI Investment Is Changing Logistics and Infrastructure Strategy appeared first on Data Center POST.

TL;DR

• Responsibly planned data centers can provide lasting economic and community benefits.
• Local investments often extend beyond the facility to infrastructure and workforce development.
• Community engagement is essential to balancing growth with sustainability.
• Digital infrastructure can help strengthen local economies while supporting the AI-driven future.

# # #

Written by iMiller Public Relations on behalf of Nomad Futurist.

As communities evaluate new data center developments, public discussion often centers on power consumption, water usage, and environmental impact. While these are important considerations, Nomad Futurist Student Advocate Prathyaj Mantha encourages communities to also recognize the long-term economic and infrastructure benefits responsibly planned data centers can deliver.

Beyond supporting AI and the digital economy, modern data centers can generate local tax revenue, create skilled jobs, strengthen utility infrastructure, and expand workforce development opportunities through partnerships with educational institutions. The article emphasizes that thoughtful planning and community engagement can help ensure these facilities become valued long-term partners while supporting sustainable growth.

This content originally appeared on the Nomad Futurist website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post The Case for Community-Centered Data Centers appeared first on Data Center POST.

• Permits secured: VFE Networks has cleared the key licenses and permit needed to move Phase 1 forward.
• Local partners: Express-tek and S&N Infrastructure are helping support the build and deployment.
• Phase 1 scope: A 126-mile diverse fiber route from Northern Virginia to Richmond-Sandston, designed for redundancy and AI-ready capacity.

# # #

Virginia Fiber Express Networks LLC (VFE Networks), a wholly owned subsidiary of Infraforward Strategies LLC, has secured the licenses and permit needed to advance Phase 1 of its planned fiber network. In partnership with Express-tek and S&N Infrastructure, the company is moving forward with a project built to support rising demand for digital infrastructure across Virginia and beyond.

Phase 1 will deliver a 126-mile diverse fiber route from Northern Virginia to Richmond-Sandston, creating a redundant path outside the I-95 corridor. The network is being designed to support conduit, dark fiber IRUs, dark fiber, waves, and spectrum capacity, with an AI-ready focus built for edge inference.

“These licenses and permit unlock our ability to build and operate infrastructure designed for the demands of the data center industry pushing south out of an increasingly constrained Northern Virginia market,” said Joel M. Allen, CEO of VFE Networks. “Our diverse, non-I-95/ Route 1 route and interconnected network phases reflect a commitment to infrastructure that is resilient, future-proof and purpose-built to support autonomous network deployment while being aligned with Virginia’s long-term support of broadband expansion.”

Express-tek brings deep engineering and permitting experience to the project, while S&N Infrastructure adds decades of communications construction expertise across the Mid-Atlantic and Southeast.

“Our local knowledge of Virginia’s right-of-way landscape positions VFE Networks to move efficiently from licensure to construction,” said Stephen Brown, President of Express-tek. “Two generations of engineering and permitting expertise go into every route we design, and we are proud to help build infrastructure that will define the region’s digital future.”

“S&N has built and maintained communications networks across the Mid-Atlantic and Southeast for over 40 years, and we know how to execute in Virginia,” said Sean Davis, President and CEO of S&N Infrastructure. “This project is a meaningful step forward for the Commonwealth’s digital infrastructure, and we are committed to bringing the same reliability we are known for to Phase 1.”

With operator licenses and the initial permit now in place, VFE Networks is advancing construction and targeting a Ready for Service date in 2027.

To learn more about VFE Networks, visit vfenet.net.

The post Virginia Fiber Express Networks Advances Phase 1 of Statewide Fiber Build appeared first on Data Center POST.

TL;DR

• The data center industry is no longer evaluated solely on speed-to-market, power capacity, or capital deployment; long-term success is now heavily defined by community trust and relationship building.
• Responsible developers communicate with local officials and residents early before plans are finalized to address community concerns (such as noise, traffic, and water use) directly, preventing misinformation from filling the vacuum.
• Industry leaders align their projects with existing zoning frameworks and actively prioritize brownfield revitalization over rezoning farmland or green spaces.
• Operators build credibility by sharing real sustainability metrics and investing in advanced, water-efficient cooling systems that challenge outdated public narratives regarding resource consumption.
• The strongest players move past transactional mindsets to focus on multi-decade planning, working collaboratively with utilities to fortify the local grid and educating communities on the broad economic ecosystem they support.

# # #

As I take a moment to reflect on the past couple of months, and since an incredible conversation that was had at ITW 2026 during the panel titled: Debunking the data center misinformation dilemma. One thing became clear. the data center industry is no longer being judged solely on speed-to-market, power availability or capital deployment. It is being judged on trust.

And while headlines often focus on conflict, opposition and controversy, the reality is more nuanced. Many developers, operators and investors are already demonstrating what responsible digital infrastructure development looks like,  and those organizations are increasingly separating themselves from companies still relying on outdated development models.

The difference is becoming impossible to ignore. The companies succeeding today are not simply building faster. They are building smarter, more transparently and with greater respect for the communities they enter.

Based on our lively conversation, here are 10 ways industry leaders are earning trust, and how the industry can continue distinguishing responsible development from the practices that typically fuel opposition and distrust.

1. Responsible Developers Engage Early

One of the clearest distinctions discussed during the discussion at ITW 2026 was timing. The best operators engage communities before projects escalate into controversy.

They:

• Hold listening sessions
• Meet with local officials early
• Educate stakeholders
• Explain impacts transparently
• Answer difficult questions directly

An approach quickly becoming a red flag to communities and those opposed to data center developments is showing up with finalized site plans and expecting communities to simply approve them. It was discussed that many communities are overwhelmed by the pace and complexity of AI and digital infrastructure growth. So, when developers fail to communicate early, misinformation fills the vacuum. Industry leaders today need to understand that community engagement is not a permitting exercise. It is part of development itself.

2. Strong Projects Respect Existing Zoning

Some of the industry’s best-performing projects are succeeding because they align with:

• Industrial zoning
• Existing infrastructure corridors
• Long-term municipal planning
• Compatible land use

Our discussion at ITW 2026 repeatedly reinforced the importance of proper siting and respecting community planning frameworks. In contrast, some projects continue attempting to force incompatible development into areas never intended for large-scale infrastructure.

As a result, we are seeing this approach create:

• Political backlash
• Community distrust
• Unnecessary litigation
• Long-term reputational damage for the broader industry

The panelists convened informally after the session and shared that they are seeing developers pursuing parcels with little chance of community or zoning alignment, which is contributing to the problems and highlights that this creates a cascading effect for future and other developers that follow.

The strongest developers understand that just because land is available does not mean it is appropriate.

3. The Best Developers Revitalize Brownfields

One of the most compelling themes from the panel was the growing focus on brownfield redevelopment.

Leading organizations are transforming:

• Former factories
• Industrial sites
• Contaminated properties
• Aging manufacturing campuses

Examples discussed included projects that:

• Remediated environmental contamination
• Restored tax revenue
• Reused existing infrastructure
• Reduced prior industrial water consumption

These projects demonstrate what responsible digital infrastructure development can look like.

In contrast, projects that seek to rezone farmland or green space without broader community value creation, tend to become flashpoints for opposition. The lesson is not that communities oppose development.  It is that communities increasingly expect thoughtful development that fits the community and is mutually beneficial.

4. Transparent Operators Share Data

Another major differentiator emerging in the industry is transparency. The strongest operators increasingly provide:

• Sustainability reporting
• Energy metrics
• Water usage data
• Third-party certifications
• Measurable environmental benchmarks

The discussion among the panelists highlighted growing efforts by the Energy Information Administration to collect energy and water usage data from data centers as part of broader policy planning discussions. The industry leaders are already adapting. Communities today expect evidence, not talking points. And increasingly, regulators do too.

5. Responsible Companies Acknowledge Impacts

One of the panel’s most important undercurrents was the recognition that not all community concerns are irrational. As a matter of fact, the top concerns include: Noise, traffic, visual impact, power planning and water consumption.

The best companies address those concerns directly and explain things like:

• Mitigation strategies
• Operational realities
• Tradeoffs
• Infrastructure investments
• Long-term planning considerations

The weaker approach is dismissiveness. Communities lose trust quickly when developers imply residents are simply “anti-growth” or uninformed. Responsible operators understand that acknowledging concerns builds credibility.

6. The Industry’s Best Players Explain the Full Economic Ecosystem

Another major distinction discussed during the session was how economic impact is communicated. The industry often undersells itself by focusing only on direct facility employment. The reality is far broader.

Data centers support:

• Engineering
• Utilities
• Construction
• Hospitality
• Remote workforce enablement
• Local service businesses
• Environmental consulting
• Technology ecosystems
• Municipal tax bases
• And more

The  positive economic ripple effect of data center developments extends far beyond the facility footprint itself. Think of all of the newspapers sold; environmental firms raising membership; service providers; printers and other ancillary business or opposition support. It is an economic driver from all directions.

The strongest companies help communities understand the full ecosystem impact beyond what the local facility can employ.

7. Modern Developers and Operators Invest in Efficiency

During the panel conversation, water usage became one of the most heavily discussed topics. Importantly, the conversation highlighted how dramatically cooling technologies have evolved.

For example:

• Campuses using less water than prior industrial uses (golf courses, papermills, are two examples that notoriously use a lot of water)
• Facilities reducing agricultural-era consumption
• modern cooling systems lowering overall water demand substantially

The industry’s best operators are investing heavily in:

• Liquid cooling
• Efficiency optimization
• Advanced thermal management
• Closed-loop systems
• Smarter infrastructure design

Meanwhile, outdated public narratives often still assume every modern data center operates like facilities built decades ago. And, it’s this assumption that is causing consternation in the market and driving opposition narratives that are false and misleading.

The industry must continue educating communities on how rapidly technology is improving.

8. Responsible Operators Build with and for fortification of the Grid

Another myth directly addressed during the panel was the idea that data centers operate continuously on diesel generators or receive priority over homes during outages.

The panel discussion made it clear::

• Generators are backup systems
• Utilities prioritize homes and hospitals first
• Many operators invest in their own infrastructure capacity
• Grid coordination is becoming increasingly sophisticated

The strongest operators work collaboratively with utilities and municipalities on long-term planning. The weakest approach infrastructure as purely transactional, focusing only on securing capacity instead of helping strengthen regional resilience. That distinction matters more every year.

9. Industry Leaders and Proven Developers Think Long-Term

One of the more subtle themes throughout the session was the difference between long-term infrastructure planning and speculative development behavior that we are seeing from a lot of land developers seeking to get into the data center arena.

The best organizations are:

• Planning multi-decade infrastructure investments
• Engaging stakeholders consistently
• Adapting designs thoughtfully
• Integrating sustainability goals
• Aligning with regional planning frameworks

Others continue chasing opportunistic land acquisitions and rapid entitlement strategies without broader alignment. And, this is where communities are becoming increasingly sophisticated at identifying the difference.

Unfortunately, once trust is broken in a market, it becomes significantly harder for responsible developers to operate there afterward.

10. The Future Belongs to Companies That Understand Trust Is Infrastructure

Ultimately, the panel reinforced a powerful reality: Digital infrastructure development is no longer only an engineering exercise nor is it for the land-use attorneys to navigate. It is a relationship exercise.

The companies leading the industry forward understand that successful projects require:

• Technical excellence
• Community trust
• Political awareness
• Transparency
• Long-term accountability
• Respect for land use planning

And importantly, many companies are already demonstrating what that future looks like. The goal now is not to reinvent the industry. It is to scale the best practices already emerging across it, and continue separating responsible infrastructure leadership from the behaviors that continue to create resistance.

Because the strongest projects are not simply the ones that get approved fastest. They are the ones communities are ultimately proud to have.

To learn more about how Ilissa Miller and the iMiller Public Relations team can help well intended developers to approach the market with compassion, empathy and respect, visit  www.imillerpr.com.

The post 10 Ways the Data Center Industry Is Earning Community Trust — And Where the Industry Still Gets It Wrong appeared first on Data Center POST.

TL;DR

• Lightpath is extending its AI-grade fiber network to support two new hyperscale data center campuses under construction in Saline, Michigan and Port Washington, Wisconsin.
• Each campus is planned to exceed one gigawatt of capacity, requiring route-diverse, multi-terabit fiber connectivity.
• The projects continue Lightpath’s expansion into the markets where hyperscale AI infrastructure is being built.
• The deployments highlight the growing role of fiber infrastructure in supporting AI-driven compute demand and next-generation digital infrastructure.

###

Artificial intelligence is reshaping where digital infrastructure is built and how networks are designed to support it. As hyperscale data center development accelerates, fiber infrastructure has become a critical component of ensuring these facilities can move the massive volumes of data AI workloads require.

To support this next phase of AI infrastructure growth, Lightpath has announced new fiber builds serving two hyperscale data center campuses currently under construction in Saline, Michigan and Port Washington, Wisconsin. Each campus is planned to exceed one gigawatt of capacity and will be connected through route-diverse, multi-terabit fiber infrastructure designed to support high-capacity, low-latency connectivity.

The Saline deployment is expected to be completed by the end of 2026, with the Port Washington build scheduled for the second quarter of 2027. Both projects are being delivered in partnership with an anchor hyperscale customer, extending Lightpath’s AI-grade network into markets experiencing significant investment in AI and cloud infrastructure.

These deployments build on Lightpath’s broader network expansion strategy, following recent growth across Phoenix, Eastern Pennsylvania, Columbus, and the company’s first long-haul corridor connecting Columbus and Chicago. Together, these investments reflect a continued focus on expanding owned fiber infrastructure into the regions where hyperscale computing capacity is growing most rapidly.

“Lightpath is playing an increasingly central role in partnering with hyperscalers to build new fiber infrastructure to address AI-driven demand across the U.S.,” said Chris Morley, CEO of Lightpath. “Fiber infrastructure remains a critical component in the evolving and accelerating AI ecosystem.”

As AI campuses continue increasing in both scale and power requirements, connectivity demands extend beyond simply bringing fiber to a site. Large-scale deployments require engineered, route-diverse networks capable of delivering resilient, multi-terabit capacity while integrating new construction with existing network assets and strategic partner infrastructure.

These latest projects reinforce how AI infrastructure development is changing the geography of network investment. As hyperscalers expand into new markets selected for available land and power, the supporting fiber infrastructure must evolve alongside them to deliver the connectivity these facilities require.

Read more in the press release here.

The post Lightpath Expands AI-Grade Fiber to Meet Growing Hyperscale Demand appeared first on Data Center POST.

TL;DR

• A Stronger Platform for Growth: Sabey Data Center Properties (SDCP) has welcomed a minority equity investment from Ares Secondaries funds, expanding its ownership base and strengthening its long-term growth strategy.
• Meeting Rising Infrastructure Demand: The investment comes as demand for data center capacity continues to grow, creating additional opportunities to expand SDCP’s existing campuses.
• A Vote of Confidence in SDCP: The partnership reflects confidence in SDCP’s operating platform and positions the company to continue delivering mission-critical infrastructure for customers across the United States.

# # #

As demand for digital infrastructure ramps up, the companies building tomorrow’s data centers are increasingly looking beyond traditional financing to support long-term expansion. Strategic investment has become an important part of that equation, providing both capital and confidence in a platform’s ability to meet growing customer demand.

Sabey Data Center Properties (SDCP) has taken another step in that direction with a minority equity investment from Ares Secondaries funds. The investment expands SDCP’s ownership base while strengthening the long-standing partnership between Sabey Corporation and National Real Estate Advisors. It also provides additional institutional capital to support the company’s disciplined approach to future growth.

Driven by the rapid growth of cloud computing and artificial intelligence, demand for data center infrastructure continues to rise. SDCP currently operates six energized campuses totaling approximately 251 megawatts of operating capacity, with existing land holdings that provide the opportunity to nearly triple that capacity by 2036. The additional investment helps position the company to accelerate development across its current campuses while pursuing future opportunities in key markets.

“Welcoming Ares as an investor is a strong endorsement of SDCP’s platform, our team and the long-term demand we’re seeing for scaled data center infrastructure,” said Tim Mirick, President of SDCP. “With Sabey, National and now Ares aligned behind the business, we are well positioned to execute our disciplined growth plan and deliver the mission-critical capacity our customers need.”

The investment also reinforces the strength of the partnership that has guided SDCP’s growth. National Real Estate Advisors described the addition of Ares as a natural extension of its long-standing relationship with Sabey, while Ares noted that SDCP’s high-quality operating portfolio and future development opportunities made the platform an attractive investment. Together, the three organizations are aligned around supporting the next phase of SDCP’s expansion.

As enterprises continue investing in AI and other digital technologies, the need for reliable data center infrastructure is expected to grow alongside them. With additional institutional support and a clear path for expansion, SDCP is well positioned to continue developing the capacity needed to support customers while reinforcing its role as one of the nation’s leading privately held data center platforms.

The post Strategic Partnership Accelerates Growth at Sabey Data Center Properties appeared first on Data Center POST.

TL;DR

• The global data center chillers market is projected to grow from USD 2.8 billion in 2026 to USD 6.4 billion by 2035 at a 9.6% CAGR, propelled by escalating rack densities from generative AI and high-performance computing workloads.
• Water-cooled systems remain the dominant technology, holding a 61.8% market share (USD 1.6 billion) in 2025 with a projected 10% CAGR. Meanwhile, air-cooled systems (32.9% share) are scaling rapidly in water-scarce regions.
• Technological innovation is shifting toward oil-free and magnetic-bearing platforms, which can cut compressor power consumption by up to 49% at part-load cycles to help operators hit strict Power Usage Effectiveness (PUE) targets.

# # #

The global data center chillers market is poised for significant expansion, projected to grow from USD 2.8 billion in 2026 to USD 6.4 billion by 2035. According to a comprehensive industry report published by Global Market Insights Inc., the market will advance at a steady compound annual growth rate (CAGR) of 9.6% during the forecast period. This follows a strong baseline year in 2025, where the market was valued at USD 2.6 billion.

The primary catalyst for this aggressive growth is the global transition into a hyper-digital era. The exponential rise of generative artificial intelligence (AI), machine learning, and high-performance computing (HPC) workloads has radically increased rack power densities. Because cooling demands scale directly with IT loads, modern multi-megawatt hyperscale and colocation facilities require massive, industrial-grade chiller systems to maintain operational stability and prevent thermal stress.

Water-Cooled Systems Dominate, Air-Cooled Solutions Scale in Arid Regions

• Water-Cooled Chillers: This segment dominated the landscape in 2025, capturing a 61.8% market share and generating USD 1.6 billion in revenue. Prized for their superior heat transfer efficiency and scalability, water-cooled systems are projected to grow at a 10% CAGR through 2035, especially across large-scale hyperscale campuses.
• Air-Cooled Chillers: Holding a 32.9% market share in 2025, air-cooled units are expanding rapidly due to speed-to-capacity advantages, simpler installation, and strict water-conservation mandates in drought-prone regions like the Western U.S. and Northern China.

The Rise of Oil-Free and Magnetic-Bearing Innovations

In response to tightening environmental policies and stringent Power Usage Effectiveness (PUE) targets, the industry is shifting rapidly toward technological innovations. Magnetic-bearing centrifugal chillers and oil-free platforms are emerging as key growth drivers. Validated data indicates these advanced systems can cut compressor power draw by up to 49% at part-load cycles, making them highly attractive to procurement teams retrofitting older halls or building out dedicated AI zones.

Regional Outlook: North America Leads, Asia-Pacific Accelerates

North America remains the largest market, valued at USD 1.3 billion in 2025, heavily propelled by massive domestic investments in AI infrastructure. Meanwhile, the Asia-Pacific region is slated to log the fastest growth rate with a 10.8% CAGR through 2035. Rapid infrastructure updates in China and aggressive government-backed digital initiatives in India are positioning APAC as a primary hub for future data center deployments.

Consolidated Competitive Landscape

The data center chillers market is highly consolidated, with the top five players, Johnson Controls, Carrier Global, Daikin, Trane Technologies, and Vertiv, collectively commanding 74.2% of the market in 2025. Johnson Controls emerged as the individual market leader, holding a 20.9% market share. Prominent players are actively introducing modular, water-free, and high-capacity units (such as recent 2026 launches from Panasonic and Johnson Controls) to capture emerging demand from generative AI infrastructures.

The post Data Center Chillers Market to Surpass $6.4 Billion by 2035 appeared first on Data Center POST.

TL;DR

• Every Industrial Revolution has fundamentally changed workforce requirements and influenced the evolution of education.
• Digital technologies and AI are transforming how knowledge is delivered, accessed, and personalized for future learners.
• Universal access to education remains essential to developing the skilled workforce required for continued innovation.
• Digital infrastructure organizations can play an important role by supporting workforce development and expanding educational opportunities.

# # #

Written by iMiller Public Relations on behalf of Nomad Futurist.

How Industrial Revolutions Have Influenced Learning

Throughout history, every Industrial Revolution has transformed not only the way people work, but also how they learn. From mechanized manufacturing and electrification to digital technologies and artificial intelligence, each wave of innovation has reshaped workforce demands and driven changes in educational systems designed to prepare future generations.

In the latest Nomad Futurist thought leadership article, Melissa Reali-Elliott explores how advances in technology have influenced education over time, rom apprenticeships and public schools to standardized curricula, digital learning, and today’s AI-powered educational tools. As technology has evolved, so too has the need for broader access to knowledge and new approaches to workforce development.

Preparing for the Next Era of Opportunity

The article suggests that the next Industrial Revolution will be driven by artificial intelligence and personalized learning experiences that make education more accessible and adaptable than ever before. As digital infrastructure continues to expand, organizations have an opportunity to contribute by supporting educational initiatives, sharing industry knowledge, and helping prepare the workforce that will build tomorrow’s technologies.

Understanding how previous industrial revolutions shaped education provides valuable perspective for today’s digital infrastructure leaders as they invest in both technology and the people who will power the industry’s future.

This content originally appeared on the Nomad Futurist website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post The Ripple Effect of Industrial Revolutions: How Innovation Continues to Shape Education appeared first on Data Center POST.

TL;DR

• New agreement: Duos has signed a Master Services Agreement with Nistar for up to 2 MW of critical IT-load capacity and related utility infrastructure.
• Columbus expansion: The deployment will be located at Duos’ Columbus, Georgia campus, bringing the current contracted total there to 10 MW.
• Built for AI: The deployment is designed for high-density AI and HPC workloads with Tier III-equivalent infrastructure and N+1 redundant power and cooling.

# # #

Duos Technologies Group Inc. (Nasdaq: DUOT), through subsidiary Duos Edge AI, has executed a Master Services Agreement with Nistar to deploy up to 2 MW of critical IT-load capacity together with supporting utility infrastructure. The agreement marks another step in the continued expansion of Duos’ AI infrastructure platform, as it’s designed to support efficient server installation, commissioning, and operations while allowing final footprint and technical requirements to align with Duos’ deployment schedule.

“This important step with Nistar establishes a framework for material expansion of our existing compute footprint, and we look forward to advancing toward full deployment in the coming months,” said Duos CEO Doug Recker. “The growing demand for AI infrastructure continues to create significant opportunities for scalable, edge-focused data center deployments, and we believe our platform is well positioned to support these emerging requirements.”

The deployment will be located at Duos’ Columbus, Georgia campus and delivered as a fully Ready for Service environment for high-density AI and high-performance compute workloads. The agreement brings the current contract total at the site to 10 MW, with billing expected by the end of August 2026.

“Duos’ high-density data center infrastructure gives Nistar a practical path to serve an institutionally backed AI-compute client requiring approximately 2 MW of critical IT-load capacity to support a planned deployment of 1,024 NVIDIA B200 GPUs,” said Jay Sivam, Chief Executive Officer of Nistar. “…It represents an important step in Nistar’s broader strategy of delivering scalable, financeable AI and high-performance-computing infrastructure where speed to deployment, power availability, and high-density design are critical.”

The dedicated deployment will be supported by Tier III-equivalent infrastructure with N+1 redundant power and cooling systems to support high availability and reliability. Duos also plans to install an additional 10 MW between October and November 2026 as it continues scaling the Columbus campus.

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

The post Duos Edge AI Signs Agreement for an Additional 2 MW Deployment appeared first on Data Center POST.

TL;DR

• AI continues to influence facility design, infrastructure planning and long-term operational strategies.
• DICE National reinforced the importance of integrating power, capital, construction and workforce planning into every phase of development.
• Regional DICE events demonstrated how local market conditions continue to shape investment, site selection and deployment strategies.
• Construction innovation, operational readiness and industry collaboration remain essential to delivering the next generation of digital infrastructure.

# # #

The second quarter of 2026 continued the momentum of BISNOW’s Data Center Investment Conference & Expo (DICE) series, bringing together leaders from across the digital infrastructure ecosystem to examine the issues influencing data center development around the world. From regional markets in the United States to one of Europe’s fastest-growing investment hubs, each conference provided a platform for discussions on AI infrastructure, power availability, construction, capital investment and operational strategy.

The quarter featured Data Center Investment Conference & Expo (DICE) Southwest, DICE National, DICE Pacific Northwest, DICE Ohio, Data Center Construction, Design & Development Conference West, and the UK Investment & Development Conference. While each event addressed the priorities of its respective market, the conversations revealed many of the same opportunities and challenges facing developers, operators, investors and infrastructure providers.

AI Continues to Influence Infrastructure Planning

Artificial intelligence remained at the center of nearly every t discussions focusing on how AI workloads are changing infrastructure requirements across every stage of the project lifecycle. Developers are planning campuses capable of supporting higher rack densities, larger power demands and next-generation g cooling technologies while maintaining the flexibility to accommodate future advances in compute infrastructure.

Those conversations extended well beyond facility design. Sessions throughout DICE National, DICE Pacific Northwest and Construction, Design & Development West examined how AI is influencing network architecture, construction sequencing, operational planning and long-term investment decisions. Rather than designing for today’s requirements alone, organizations are building infrastructure that can adapt as AI technologies continue to mature.

Regional Markets Are Defining New Growth Opportunities

The regional DICE conferences highlighted how local market conditions continue to shape development strategies while contributing to broader industry growth. Each location presented its own set of opportunities, demonstrating that successful projects require solutions tailored to regional infrastructure, regulatory environments and available resources.

In the Southwest, conversations centered on site selection, water availability, sustainability and operational planning as Arizona continues to attract AI-driven investment. The Pacific Northwest focused on modular design, construction technologies and operating high-density facilities while balancing reliability, efficiency and deployment speed. Meanwhile, DICE Ohio examined one of the country’s fastest-growing emerging markets, with discussions addressing power systems, resilient infrastructure, construction delivery and the practical realities of supporting AI-ready facilities in a rapidly expanding region.

Across the Atlantic, the UK Investment & Development Conference examined many of the same issues from a European perspective. Investment activity, planning approvals, land availability and energy procurement remained central topics as organizations evaluated strategies for expanding capacity while meeting sustainability objectives and navigating an increasingly competitive marketplace.

DICE National Focused on the Industry’s Biggest Priorities

As BISNOW’s flagship conference , DICE National, brought together industry leaders from across the digital infrastructure ecosystem for three days of discussions spanning the entire project lifecycle. The topics ranged from capital deployment and site selection  to AI-ready network architecture, power economics, cooling technologies, workforce development and responsible AI.

Power availability remained one of the event’s defining  themes. Sessions explored how organizations are evaluating utility partnerships, powered land, battery storage, on-site generation and diversified energy strategies while balancing cost, sustainability and long-term resiliency. Additional discussions examined community engagement, permitting, construction delivery and workforce development, reflecting the increasingly interconnected nature of modern data center projects.

Construction Innovation and Operational Readiness Continue to Advance

Construction remained another major focus throughout the quarter as the industry looks for ways to accelerate deployment without compromising quality or reliability. Modular construction, prefabrication, standardized designs and digital project management technologies were recurring topics across multiple conferences, reflecting the industry’s effort to improve predictability while reducing project schedules.

Operational excellence also received significant attention as higher-density computing environments continue to introduce new challenges for facility management. Sessions explored advanced monitoring, automation, liquid cooling, thermal management and operational controls designed to improve efficiency while supporting increasingly demanding AI workloads. These discussions reflected the growing recognition that successful projects require equal attention to both construction execution and long-term operations.

Collaboration Remains Central to Industry Progress

One theme consistently e emerged across every DICE event: delivering digital infrastructure has become a collaborative effort that extends well beyond individual organizations. Developers, utilities, contractors, investors, engineering firms, technology providers and public agencies all contribute to successful project outcomes, particularly as developments become larger and more complex.

Early coordination, transparent communication and integrated planning were recurring themes throughout the quarter. Whether discussions focused on power procurement, permitting, construction sequencing or community engagement, participants agreed that collaboration is helping organizations reduce project risk while improving delivery timelines and long-term operational success.

Building Momentum Across the Digital Infrastructure Industry

The second quarter of BISNOW’s DICE event series reflected an industry focused on preparing for sustained growth rather than simply adding capacity. Organizations are approaching development with greater attention to energy strategy, construction planning, operational resilience and long-term flexibility as AI continues to influence infrastructure requirements around the world.

Although each conference addressed different regional priorities, the conversations consistently pointed toward a common objective: building digital infrastructure that can support future demand while responding to evolving technology, energy and market conditions.

To learn more about upcoming DICE events and register for future conferences, visit BISNOW’s events page at www.bisnow.com/events.

The post BISNOW DICE Event Series Spotlight the Evolution of Data Center Development appeared first on Data Center POST.

TL;DR

• Fuel degradation happens quietly over long storage periods, leading to clogged filters and catastrophic generator failures when emergency power is needed most.
• Hydrotreated Vegetable Oil (HVO/R99) offers excellent storage stability, direct engine compatibility, and cleaner emissions, helping facilities meet sustainability goals while securing reliable power.
• Under the 2026 Renewable Fuel Standard, end users (such as data centers) face higher RIN obligations, making supplier transparency and strict compliance documentation vital to managing regulatory and financial risks.

# # #

Mission-critical facilities do not get a second chance when fuel goes bad, storage conditions drift, or supply chains tighten. For data centers, hospitals, utilities, and emergency systems, the difference between resilience and disruption often comes down to fuel quality, testing, and planning.

Why fuel failures happen

Fuel failures usually start quietly. Water intrusion, microbial growth, oxidation, contamination, and long storage periods can all degrade diesel before anyone notices. In backup power systems, that can lead to clogged filters, poor starts, engine wear, or generator failure exactly when the system is needed most.

The risk is growing because many facilities are balancing longer runtime expectations, more frequent testing, and tighter environmental goals. At the same time, supply chain conditions and regulatory changes are making fuel selection and handling more complex.

What resilient facilities monitor

Strong fuel programs focus on a few basics. Regular sampling and testing help catch contamination early, while tank inspections and housekeeping reduce the chance of water and sediment buildup. Documentation matters too, because quality issues are easier to prevent when the facility can trace fuel origin, storage history, and compliance records.

For renewable diesel users, the same discipline applies. ESI Total Fuel Management notes that hydrotreated vegetable oil (HVO/R99) offers high storage stability, direct compatibility with existing diesel engines, and cleaner combustion than conventional diesel or biodiesel blends. ESI also highlights the importance of Certificates of Compliance, supplier transparency, and testing for ESG compliance.

Renewable diesel and compliance

Renewable diesel can improve resiliency, but only if the fuel is sourced and managed correctly. ESI’s materials explain that HVO is a hydrocarbon fuel made from waste oils and animal fats, with strong cold-weather performance and lower emissions than fossil diesel. The company also notes that biomass-based diesel is changing under the 2026 Renewable Fuel Standard, with mandates shifting to RIN-equivalent gallons and higher RIN obligations for end users such as data centers using renewable diesel for stationary generators.

That means cost, compliance, and reliability now sit closer together. Facilities need to know whether RINs are properly retired, whether supply is documented, and whether the fuel they buy is truly fit for backup power use.

Practical steps to reduce risk

A resilient fuel strategy usually includes a few non-negotiables:

• Test fuel on a schedule, not just after a problem appears.
• Keep tanks dry, clean, and monitored for water and sediment.
• Verify supplier documentation, including compliance and quality records.
• Use storage and replenishment plans that account for supply disruptions.
• Review whether renewable diesel or HVO fits both operational and regulatory needs.

For large facilities, the goal is not just to buy fuel. It is to create a system that keeps fuel usable over time, especially when weather, market conditions, or compliance changes put pressure on operations.

Closing perspective

Operational resiliency depends on fuel that performs as expected, on the day it is needed most. That is why ESI’s approach emphasizes stable supply, fuel quality, and clear compliance practices alongside the technical benefits of renewable diesel. Contact ESI Total Fuel Management today to ensure your operations are running at optimal resiliency.

The post Ensuring Operational Resiliency: Preventing Fuel Failures in Mission-Critical Facilities appeared first on Data Center POST.

TL;DR

• In the rapid deployment of AI and digital infrastructure, speed to market, rather than capital. has become the primary driver of success.
• A global developer managing 15 projects relied on quarterly reviews with three-month-old data, leaving leadership entirely unaware that a critical project was actually 245 days behind schedule.
• Drowning in Reporting Overhead: A Portfolio Management Office of over 30 employees spent 40% of their time reacting to ad-hoc executive information requests rather than supporting project delivery.

# # #

Written by iMiller Public Relations on behalf of Foresight.

Demand for artificial intelligence and digital infrastructure is at an all-time high, making time the critical constraint in delivery. However, many organizations manage billion-dollar programs with fragmented and delayed reporting, effectively driving projects blind. Research by Dr. Atif Ansar of Foresight and the University of Oxford details how a global data center developer managing 15 projects struggled with quarterly reviews based on three-month-old data, while a stretched Portfolio Management Office spent 40% of its time answering executive requests instead of supporting delivery. This lack of visibility created a false sense of security, leaving leadership unaware that a project they believed was ahead of schedule was actually 245 days behind.

To resolve these issues, Foresight consolidated schedule data from Primavera P6 and Microsoft Project into a single portfolio view within three weeks. This allowed the developer to transition from lagging quarterly reviews to monthly executive reviews supported by weekly project discussions, successfully shifting the focus from validating reports to actively solving delivery problems. As a result, annual reporting costs dropped by 90%, falling from $2 million to $200,000. More importantly, internal estimates showed a 20% reduction in project delays, preventing more than $100 million in lost revenue across the portfolio by enabling leadership to intervene before delays became irreversible.

In the accelerating race to build digital infrastructure, achieving clear, timely visibility is a critical competitive advantage. Better project outcomes do not stem from producing a larger volume of reports, but from providing leadership with a trusted, unified view of performance while there is still time to influence the results.

This content originally appeared on the Foresight website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post Visibility Is Becoming the Competitive Advantage in Data Center Delivery appeared first on Data Center POST.

TL;DR

• True data sovereignty goes beyond physical storage location (residency) to demand enforceable organizational control over encryption keys, backup systems, administrator access, and legal jurisdictions.
• Managing massive, proprietary, or regulated datasets for artificial intelligence creates high costs and compliance hurdles, pushing enterprises to run AI workloads closer to governed data via private or hybrid infrastructure.
• While private infrastructure has its own physical costs, running predictable, data-heavy workloads on-premises or in colocation facilities provides more stable economics and easier auditability compared to public cloud sprawl.

# # #

For years, the public cloud was treated as the natural destination for enterprise infrastructure. The logic was simple: move workloads to hyperscale platforms, gain speed and flexibility, and reduce the burden of managing physical hardware. That model still works for many applications, especially those with unpredictable demand, distributed users, and strong cloud-native design.

But a new pressure is reshaping enterprise cloud strategy: data sovereignty.

Enterprises are no longer asking only where their applications run. They are asking where their data resides, who can access it, which jurisdiction governs it, how it is encrypted, how it is backed up, and whether they can prove all of that during an audit or incident. For many organizations, especially in regulated industries, those questions are pushing certain workloads back toward private cloud, colocation, and on-premises data center environments.

Data Sovereignty Is More Than Data Residency

Data residency is about location. It answers the question: “Where is the data stored?”

Data sovereignty goes further. It asks whether the organization has enforceable control over the data, the infrastructure, the administrators, the encryption keys, the backup copies, the logs, and the legal environment around all of it. A workload can technically be hosted in the right country and still create sovereignty concerns if management access, support operations, replication, or metadata flows cross borders in ways the enterprise cannot fully govern.

That distinction matters because enterprise IT environments are rarely simple. Data may be created in one region, processed in another, backed up in a third, and accessed by support teams from multiple jurisdictions. In a single-cloud or multi-cloud environment, this can happen quickly and invisibly unless governance controls are built into the architecture from the beginning.

Why Cloud-First Is Becoming Workload-First

The movement back from the cloud is not a rejection of cloud computing. It is a sign that enterprise infrastructure strategy is maturing.

A decade ago, “cloud-first” became the default strategy for many companies. Today, more IT leaders are adopting a workload-first model. Instead of asking, “How do we move this to the cloud?” they are asking, “Where should this workload live based on risk, cost, performance, compliance, and control?”

That change is especially visible in workloads involving sensitive customer records, financial data, healthcare information, intellectual property, AI training datasets, backup repositories, and operational systems that cannot tolerate uncertain jurisdictional exposure. For these workloads, the economics and governance of public cloud can become more complicated than expected.

The result is selective repatriation. Enterprises may keep collaboration tools, web applications, and elastic development environments in public cloud while moving steady-state databases, high-volume storage, backup data, and latency-sensitive systems into infrastructure they control more directly.

Regulation Is Raising the Bar

Regulatory pressure is one of the strongest drivers behind this shift. Laws and industry frameworks around privacy, operational resilience, cross-border data transfer, and third-party technology risk are becoming more detailed. Boards and regulators increasingly expect organizations to demonstrate control, not simply outsource trust to a cloud provider.

For financial services, healthcare, government, defense, and critical infrastructure organizations, the question is no longer whether a provider has strong security. Most major providers do. The question is whether the enterprise can produce evidence that the right controls are in place, that data is handled according to policy, and that recovery can happen under pressure.

This is where private infrastructure, sovereign cloud models, and colocation environments are becoming more attractive. They allow organizations to define stricter boundaries around data location, administrator access, encryption key custody, network paths, and backup retention. In many cases, they also simplify audits because the environment is easier to map and explain.

AI Is Making the Problem Bigger

Artificial intelligence is adding another layer to the sovereignty discussion. AI systems depend on large volumes of data, and many enterprises are now trying to determine where that data should be stored, processed, indexed, and used for model training or inference.

Moving massive datasets into and out of cloud platforms can create cost, latency, and governance challenges. More importantly, organizations must understand whether sensitive data is being exposed to external services, retained in logs, used in model workflows, or replicated across regions.

For some AI use cases, public cloud will remain essential because of GPU availability and managed AI services. But for workloads involving proprietary datasets, regulated information, or long-term data retention, enterprises are increasingly exploring private AI infrastructure, hybrid architectures, and data-center-based processing models. The goal is not to avoid AI; it is to bring compute closer to governed data.

Cost and Control Are Connected

While sovereignty is often discussed as a compliance issue, it is also a cost issue. Large cloud environments can become expensive when workloads are predictable, data-heavy, or constantly moving between services and regions. Egress fees, storage growth, backup retention, replication, and underused resources can turn cloud flexibility into cloud sprawl.

Repatriating a workload does not automatically make it cheaper. Private infrastructure comes with its own costs: hardware, power, cooling, staffing, lifecycle management, and physical security. But for steady workloads with known utilization patterns, enterprises may gain more predictable economics by running them in private environments or colocation facilities.

That predictability matters. When infrastructure supports critical data, cost control and governance are linked. Organizations need to know not only what they are spending, but also what risk they are accepting in exchange for that spend.

What Enterprises Should Evaluate Before Repatriating

Pulling workloads back from the cloud should not be a reactionary move. It should be a structured decision based on technical and business requirements.

Enterprises should begin by classifying their workloads and data. Which systems contain regulated information? Which datasets are subject to regional restrictions? Which applications generate high data movement costs? Which workloads have stable demand? Which systems require low latency or strict recovery guarantees?

Next, they should review access and control. Who can administer the environment? Where are encryption keys stored? How are backups protected? Can the organization prove where data is located and how it is replicated? What happens if a provider outage, contract dispute, or geopolitical event affects access?

Finally, enterprises should design for portability. The strongest infrastructure strategies do not lock every workload into one environment. They use cloud where cloud adds value, private infrastructure where control is required, and colocation where resilient, connected, professionally managed facilities can bridge the two.

The Future Is Hybrid, but More Intentional

The future of enterprise infrastructure is not purely public cloud or purely on-premises. It is a more intentional hybrid model.

Public cloud will continue to support innovation, scalability, global reach, and managed services. But data centers, colocation facilities, private cloud platforms, and sovereign infrastructure will play a larger role in workloads where jurisdiction, auditability, cost predictability, and operational control are critical.

For many enterprises, the answer is not to abandon cloud completely, but to regain control over the parts of the environment that carry the most regulatory or operational risk. Private infrastructure, colocation, and hybrid storage models are becoming more important because they help organizations define clearer boundaries around data location, encryption, access control, backup retention, and recovery. In this model, private cloud storage can support a more controlled approach to storing and protecting sensitive workloads without fully giving up cloud-like scalability.

The data sovereignty problem is forcing enterprises to rethink an assumption that shaped the last decade: that moving to the cloud always means gaining control. In reality, control depends on architecture. For some workloads, the best answer is still public cloud. For others, the answer is bringing data and compute closer to infrastructure the enterprise can govern directly.

The companies that succeed will not be the ones that simply move everything back or push everything forward. They will be the ones that understand their data, classify their risk, and place each workload where it belongs.

# # #

About the Author

George Williams has over 13 years of experience in the data storage, backup and disaster recovery, 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 in 2006, George has been working to ensure that technical information is relayed in a simple and effective way to customers and targeted audiences. George helps curate content and works with numerous publishers and technology blogs to spread awareness and knowledge of data storage technology.

The post The Data Sovereignty Problem: Why Enterprises Are Pulling Workloads Back from the Cloud appeared first on Data Center POST.

TL;DR

• Traditional auto-scaling fails for AI workloads because LLM inference is constrained by GPU memory and KV cache rather than CPU. Relying on legacy CPU metrics leads to quiet, expensive budget bleeds while request queues bottleneck behind pinned GPUs.
• Inference latency is solved via weight compression and smart routing rather than traditional CDN caching. Shrinking model footprints via quantization and routing queries dynamically based on complexity prevents expensive frontier models from being wasted on simple requests.
• AI FinOps must automate the scale-down process to curb runaway idle GPU costs. Teams should adopt a “scale-to-floor” policy to maintain a minimal warm SLA pool, use cheap preemptible spot capacity for asynchronous batch work, and treat GPU utilization gaps as operational defects.

# # #

The first time a real-time inference cluster runs away with the budget, it rarely shows up as a spike. It shows up as a slow, expensive bleed that no alert ever fired on because every metric you trusted for the last decade is measuring the wrong thing. Your horizontal auto-scaler watched CPU hover at 30% and concluded all was well, while a queue of token-generation requests stacked up behind a GPU that was already pinned at 100% memory utilization. The pod count never moved. The latency graph climbed. The invoice climbed faster.

This is the quiet structural problem facing anyone running AI-powered applications in 2026: the assumptions baked into cloud-native architecture were written for stateless, CPU-bound web services. Large language model inference breaks almost all of them. If you are still sizing clusters, scaling policies, and storage tiers the way you did for a CRUD app, you are subsidizing your provider.

Let’s get specific about where the old playbook fails and what replaces it.

Compute: From CPU Sizing to GPU Orchestration

Classic horizontal scaling assumes work is fungible. Add a node, spread the load, done. Inference doesn’t behave that way. The binding constraint is GPU memory, not CPU, and a single accelerator can only hold so much. An H100 carries 80GB of HBM; an H200 stretches to 141GB; Blackwell-class B200 parts push past 190GB. Once a model’s weights plus its KV cache fill that memory, the next request doesn’t get “a little slower,” it waits in line or gets evicted.

That changes the unit of scaling. You are no longer scaling pods against CPU; you are scheduling sequences against finite GPU memory and managing the KV cache as a first-class resource. Modern serving stacks vLLM, TensorRT-LLM, and their orchestration layers exist precisely to solve this, using continuous batching to interleave requests at the token level and paged attention to stop the KV cache from fragmenting the card. Cluster health, in turn, stops meaning “are the nodes up?” and starts meaning “what is my real GPU utilization, my batch fill rate, and my queue depth per replica?” A fleet of H100s sitting at 40% utilization is not a healthy cluster. It is a six-figure monthly rounding error.

Two failure modes deserve naming because they stay invisible on a traditional dashboard. The first is cold start: spinning up a fresh GPU pod means pulling a multi-gigabyte model into VRAM, and that can take minutes or an eternity when a request is already waiting. The second is coarse allocation, where a small model monopolizes a large card it barely touches. Both are solvable in the scheduling layer through techniques like Multi-Instance GPU partitioning and model-aware bin-packing, but only if you architect for them up front. Bolted on after launch, they become a rewrite.

The teams building serious conversational products feel this first. A company offering production-grade chatbot development services quickly discovers that the model is the easy part; the hard part is keeping a GPU pool warm enough to hit sub-second first-token latency without paying for idle silicon around the clock. That tension responsiveness versus utilization is now the central design problem of the AI cloud, and it has no setting in your old auto-scaling group.

The Data Layer: Why SQL and NoSQL Quietly Fall Apart

Your relational and document stores were built to answer questions about exact values. Where user_id = 4471. Where status = ‘active’. They are exceptional at it. They are also structurally incapable of answering the question every AI application actually asks: what is most similar in meaning to this?

Semantic retrieval operates on high-dimensional embedding vectors with hundreds or thousands of dimensions, and “closeness” is a distance calculation across that space, not a B-tree lookup. Bolt a similarity search onto Postgres without the right index, and it degrades into a full scan that gets slower with every row you add. This is why the data layer is being re-stratified rather than replaced wholesale. Native vector capability now lives in three broad tiers: extensions like pgvector for teams that want similarity search alongside their existing transactional data; purpose-built distributed engines like Milvus when recall and throughput at billions of vectors matter; and managed services like Pinecone for organizations that would rather buy the operational burden than run it.

The deeper architectural shift is that retrieval has become stateful infrastructure, not an application-layer afterthought. Even the choice of index is now a real engineering decision rather than a default: graph-based indexes like HNSW deliver excellent recall at low latency but cost memory, while partition-based approaches like IVF trade some accuracy for a smaller footprint, and that tradeoff shifts again the moment your corpus updates in real time rather than sitting static. A retrieval-augmented generation (RAG) workflow is a pipeline embed the query, runs an approximate-nearest-neighbor search against the index, assembles the retrieved context, then calls the model, and every stage has its own latency, scaling, and consistency profile. Treating that pipeline as “just a database call” is how you end up with a chatbot that answers correctly in the demo and times out under load. The conversational front end and the vector store are one system, and they have to be containerized, versioned, and scaled as one.

Network and Edge: The Token Latency Problem

Web performance used to be a payload problem: ship fewer bytes, cache aggressively, push assets to a CDN. Inference performance is a generation problem. The model produces output one token at a time, and the user perceives quality through two numbers: time to first token, and tokens per second after that. Neither improves by moving a static file closer to the user.

What moves the needle is making the model itself cheaper to run, and that is where quantization has become non-negotiable rather than optional. Compressing weights from FP16 down to FP8 or INT8 using approaches like GPTQ or AWQ shrinks the memory footprint and raises throughput, often with negligible quality loss for well-chosen workloads. Smaller, quantized models are also what make edge and regional inference viable: instead of routing every request back to a central GPU farm, you can serve a distilled model from infrastructure physically nearer the user, collapsing round-trip latency for the requests that don’t need the largest frontier model. The architectural decision is no longer “which model?” but “which model, quantized how, served where, for which class of request?” Routing logic that grades requests by complexity and dispatches them accordingly is becoming a standard layer in any mature AI chatbot app development services stack, because serving every query with your most expensive model is the fastest way to make the unit economics collapse.

FinOps: Engineering the Scale-Down

Here is the part most teams under-build, because scaling up is exciting and scaling down is just discipline. Specialized compute is expensive enough that idle capacity is the dominant line item on most AI infrastructure bills, and traffic to conversational and inference workloads is rarely flat it follows business hours, regional patterns, and campaign spikes. If your GPU pool is provisioned for peak and never contracts, you are paying peak rates twenty-four hours a day.

Real FinOps for AI is structural, not a quarterly spreadsheet review. A few techniques that actually hold up in production:

• Tiered serving by request class route, cheap, common queries to small quantized models, and reserve frontier-model GPUs for the requests that genuinely need them.
• Scale-to-floor, not scale-to-zero, keep a minimal warm pool sized to your latency SLA, then expand on queue depth and token throughput rather than CPU, so you absorb spikes without paying for a full peak fleet overnight.
• Spot and preemptible capacity for batch and asynchronous work, embedding generation, re-indexing, and offline evaluation don’t need on-demand reliability and shouldn’t pay for it.
• Utilization as a first-class SLO track GPU-hours billed against GPU-hours actually doing useful inference, and treat a persistent gap as a defect, not a cost of doing business.

The organizations that win the next eighteen months won’t be the ones with the largest GPU allocations. They’ll be the ones whose architecture lets specialized compute breathe expanding under load and contracting the moment demand dips without a human in the loop.

The Reframe

None of this is a reason to slow down on AI. It’s a reason to stop treating AI workloads as ordinary applications that happen to call a model. The compute, the data layer, the network path, and the cost model have all changed shape at once, and the teams that re-architect deliberately instead of patching last decade’s patterns and hoping the bill stays reasonable are the ones that will ship reliable, affordable AI at scale.

If your inference costs are climbing faster than your usage, the problem usually isn’t the model. It’s the infrastructure underneath it, and that’s a solvable, architectural problem.

EncodeDots helps engineering teams design and operate the GPU orchestration, vector data layers, and FinOps controls that make AI-powered applications both fast and economical. If your AI infrastructure has outgrown your cloud playbook, let’s pressure-test the architecture together.

# # #

About the Author

Piyush Chauhan, CEO and Founder of EncodeDots, is a visionary leader transforming the Digital landscape with innovative web and mobile app solutions for Startups and enterprises. With a focus on strategic planning, operational excellence, and seamless project execution, he delivers cutting-edge solutions that empower thrive in a competitive market while fostering long-term growth and success.

The post Your Auto-Scaler Is Lying to You: What AI Workloads Actually Do to Cloud Infrastructure appeared first on Data Center POST.

TL;DR

• Belonging is a business imperative that directly influences collaboration, innovation, and organizational resilience.
• Diversity initiatives are most effective when supported by workplace systems that encourage inclusion and participation.
• Organizations that recognize diverse perspectives are better positioned to retain talent and improve performance.
• Building resilient digital infrastructure requires investing in people as intentionally as organizations invest in technology.

# # #

Written by iMiller Public Relations on behalf of Nomad Futurist.

Why Belonging Matters in Digital Infrastructure

As digital infrastructure becomes increasingly critical to supporting AI, cloud, and connectivity, organizations are investing heavily in technology and talent. However, according to Nomad Futurist, building a resilient workforce requires more than hiring a diverse team, it requires creating an environment where employees feel they genuinely belong.

The article argues that belonging should be viewed as an organizational asset rather than simply a cultural initiative. When employees feel comfortable sharing ideas, raising concerns, and contributing their expertise without unnecessary barriers, organizations become more innovative, collaborative, and resilient. Conversely, when individuals feel excluded, valuable knowledge is often lost, communication suffers, and business performance can be impacted.

Designing Organizations Where Talent Thrives

While many organizations have prioritized diversity initiatives, the article emphasizes that representation alone does not create inclusion. Sustainable workforce development requires intentional systems that recognize different perspectives, encourage collaboration, and enable employees to contribute without feeling pressure to conform to traditional norms.

As the digital infrastructure industry addresses workforce shortages and prepares for the next generation of innovation, fostering a culture of belonging can help organizations strengthen employee engagement, improve retention, and unlock the full potential of their teams.

This content originally appeared on the Nomad Futurist website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post Belonging, Not Diversity, Is Key to Building Stronger Digital Infrastructure Teams appeared first on Data Center POST.

TL;DR

• The acceleration of AI has thrust data centers from the invisible background of digital life into the active foreground of public debate.
• Infrastructure literacy is distinct from community engagement; it builds a shared baseline of understanding regarding how physical components connect to essential digital services.
• While improving public literacy does not eliminate community concerns, it makes public inquiry informed and serves as the essential foundation that makes better engagement possible.

# # #

As artificial intelligence, cloud computing and digital public services become more deeply embedded in everyday life, data centers are moving from the background of public awareness into the foreground of public debate.

For years, data centers were largely invisible to the average person. They were understood by engineers, operators, investors, planners and the organizations that depended on them, but rarely by the wider public. That is changing.

AI has made digital infrastructure more visible. Questions around energy, water, land use, planning, workforce development, community benefit and resilience are now part of broader public conversations. At the same time, many people still do not have a clear understanding of what a data center is, what it does, why it matters, or how it connects to the digital services they use every day.

That creates a challenge for the sector.

Community engagement asks people to respond to infrastructure. Infrastructure literacy helps people understand it.

That distinction matters.

Public concern about data centers is not always rooted in opposition to technology itself. Often, it reflects a deeper uncertainty about systems people rely on but cannot see. People use cloud storage, streaming platforms, healthcare systems, online banking, school platforms, AI tools and public services every day. Yet the physical infrastructure behind those services remains poorly understood.

When infrastructure is invisible, the public conversation can become fragmented. Data centers are discussed through the lens of energy, AI, jobs, planning, water, security, big tech or local impact, often without a shared baseline of understanding. These are all important issues, but they are not the same issue. Without infrastructure literacy, they can easily collapse into a single narrative: data centers are either good or bad.

The reality is more complex.

Data centers are part of the physical layer of digital life. They depend on buildings, power, cooling, fiber, servers, land, skilled people and local communities. They support services that many people now consider essential. At the same time, they raise legitimate questions about resource use, community benefit, ownership, resilience and the pace of AI-driven growth.

The sector needs a clearer public language for this complexity.

Many operators are already thinking seriously about community benefit, sustainability and future skills; infrastructure literacy can help connect those efforts into a clearer public conversation.

Infrastructure literacy is not about persuading communities to accept every development. It is about giving people the language, context and confidence to ask better questions. What is being built? Why is it needed? Who benefits? What are the trade-offs? How will energy and water be managed? What opportunities will be created for local people? How does this infrastructure connect to education, healthcare, public services and future skills?

These are not only communications questions. They are civic questions.

For data center operators, developers and industry bodies, this presents an opportunity. Public understanding should not begin only at the point of planning, permitting or local concern. It can be built earlier, through schools, libraries, community organizations, family learning, workforce pathways and accessible public education.

This is where education becomes strategic.

If children and communities understand that the internet is not just “in the cloud,” but depends on real places, people and systems, the conversation changes. Data centers stop being mysterious buildings at the edge of public imagination and become part of a wider story about modern infrastructure.

That does not remove public concern. It makes the concern more informed.

As founder of ThreadPoint, I have been developing an infrastructure literacy approach through children’s publishing, school workshops and community programs. My children’s book, Where the Internet Goes to Sleep, introduces young readers to data centers and the physical infrastructure behind the internet. Through our workshops, children and families explore cloud computing, servers, fiber, energy and digital systems through story-led learning and hands-on activities.

I have also been awarded a Churchill Fellowship to research public understanding of the physical infrastructure behind AI, cloud computing and digital life across the USA, UAE, the Netherlands, Sweden and Denmark. Through this research, I will be exploring how different countries approach infrastructure literacy, education, digital inclusion, community engagement and workforce pathways – bringing insights and recommendations back to support UK communities, schools, policymakers and industry.

For the data center sector, the timing matters.

AI growth is accelerating demand for compute. Public debate is becoming more active, and public trust is becoming more important. Communities are asking more questions. Governments and local authorities are considering how digital infrastructure fits into wider priorities around growth, resilience, sustainability and skills.

The sector can treat public understanding as a communications risk to manage, or as a long-term public value layer to build.

The second approach is the more sustainable one.

If data centers are now part of critical infrastructure, then public understanding should be part of responsible infrastructure delivery. That means moving beyond one-way messaging and creating accessible ways for people to understand the systems shaping their lives.

The next phase of digital infrastructure will not only be judged by what the sector builds. It will also be judged by how well the sector helps people understand what is being built, why it matters and how communities can be part of the conversation.

Infrastructure literacy is not a replacement for community engagement.

It is what makes better engagement possible.

# # #

About the Author

Willow Williams CF, MSc is the founder of ThreadPoint and author of Where the Internet Goes to Sleep, a children’s book introducing young readers to data centers and the physical infrastructure behind digital life. She has been awarded a Churchill Fellowship to research infrastructure literacy and public understanding of the physical systems behind AI, cloud computing and digital life.

The post Why Data Centers Need Infrastructure Literacy in the AI Era appeared first on Data Center POST.

TL;DR

• Reaching students at younger, earlier levels of education builds familiarity and comfort with technology, which helps both the students and their families understand its value.
• Industry leaders must treat education as a core component of workforce development to establish clear pathways into the field.
• Bridging the talent gap requires using accessible language to address local concerns and clearly demonstrate how digital infrastructure connects to everyday life.

# # #

This is reality, not just a mere notion: The future of digital infrastructure depends on more than servers and software; it depends on people, awareness, and the next generation of talent. In a recent Nomad Futurist conversation with co-hosts Nabeel Mahmood and Phillip Koblence, Brittani Clarke Clayman and Jason Clayman made that case with a rare mix of industry insight and classroom perspective.

Brittani, Director of Marketing and Sustainability at ESI Total Fuel Management, has spent her career helping make complex ideas more understandable and relevant to the communities they affect. Jason, an educator and curriculum leader at Dobyns-Bennett High School, sees the same challenge from the classroom: Students cannot pursue careers they have never been exposed to.

Together, they form a powerful voice for early education and real community engagement. Their message is simple: The data center industry cannot rely on technical growth alone. It also has to explain its value, listen to local concerns, and create clear pathways into the field.

This means speaking in plain language, reaching students earlier, and showing families how digital infrastructure connects to everyday life. It also means treating education as part of workforce development, not an afterthought. As Brittani and Jason discuss, awareness starts with conversation, but it only becomes meaningful when it leads to opportunity.

“I could speak at you, but if you’re not understanding what I’m saying, then we’re not communicating,” Brittani said. “I’m just throwing information at you.”

Reaching students early is key, said Jason: “If we do start at the younger and earlier levels of education, just getting them comfortable with it and kind of understanding what it is, not only does it help them, but it helps their families.”

Their story is a reminder that the industry’s next chapter will be shaped not just by innovation, but by trust. For data center leaders, that is both a challenge and an opening.

Read the full conversation here.

The post Why Education and Community Matter to the Future of Digital Infrastructure appeared first on Data Center POST.

Written by iMiller Public Relations on behalf of Nomad Futurist.

TL;DR

• The Nomad Futurist Foundation partnered with Bisnow to provide students with direct access to DICE National and the digital infrastructure industry.
• Students gained valuable mentorship, networking opportunities, and firsthand exposure to careers supporting AI, cloud, and data center infrastructure.
• Workforce development remains a critical priority as demand for digital infrastructure talent continues to accelerate.
• Programs that connect education with industry help build the next generation of professionals driving the digital economy.

# # #

Preparing the Next Generation of Industry Talent

As AI, cloud, and digital infrastructure continue to expand, the industry faces a growing challenge: developing the workforce needed to support that growth. To help address this need, the Nomad Futurist Foundation partnered with Bisnow during DICE National to provide nine master’s students with direct exposure to one of the digital infrastructure industry’s premier events.

The initiative gave students the opportunity to engage with industry leaders, explore career opportunities, and gain firsthand insight into the technologies and businesses driving the digital economy. Students represented disciplines including engineering, AI strategy, data science, finance, operations, and marketing, highlighting the diverse skill sets supporting today’s digital infrastructure ecosystem.

Creating Connections Beyond the Classroom

Each participant was paired with a Nomad Futurist mentor who provided guidance throughout the conference while facilitating introductions to executives and industry professionals. The experience helped students better understand career pathways within digital infrastructure while expanding their professional networks and industry knowledge.

The program reflects the Nomad Futurist Foundation’s ongoing commitment to workforce development through mentorship, education, and industry engagement. By connecting students with experienced professionals early in their careers, the Foundation is helping strengthen the talent pipeline needed to support the continued growth of AI and digital infrastructure.

This content originally appeared on the Nomad Futurist website and has been adapted for syndication on Data Center POST. Read the complete blog here.

The post From Classroom to Career: Nomad Futurist Bridges the Gap Between Students and Digital Infrastructure appeared first on Data Center POST.

TL;DR

• The digital divide has shifted beyond connectivity. Today’s challenge is helping communities understand, evaluate, and participate in decisions surrounding digital infrastructure and AI.
• Community engagement is essential for project success. Transparent communication and early stakeholder involvement build trust and lead to better long-term outcomes for digital infrastructure initiatives.
• Digital literacy is becoming a competitive advantage. As AI adoption accelerates, the ability to critically evaluate information and make informed decisions is just as important as access to technology.
• Collaboration drives sustainable growth. Industry, policymakers, educators, and local communities must work together to develop planning frameworks that support responsible digital infrastructure development and economic opportunity.

# # #

Written by Authority Magazine Editorial Staff featuring Ilissa Miller, Founder and CEO of iMiller Public Relations

The Digital Divide Has Evolved

The digital divide is often framed as an issue of internet access, but according to Ilissa Miller, Founder and CEO of iMiller Public Relations, today’s challenge is increasingly centered on understanding, trust, and informed decision-making.

As digital infrastructure and artificial intelligence continue reshaping economies and communities, the gap is widening between those who understand these technologies and those left trying to navigate complex decisions without sufficient context. This shift has implications far beyond connectivity. It influences economic development, workforce readiness, and community planning.

Miller explains that communities are frequently asked to evaluate projects such as data centers, AI infrastructure, and connectivity initiatives without the technical knowledge or planning frameworks needed to make confident, long-term decisions. The result is often uncertainty, misinformation, and resistance rather than productive dialogue.

Community Engagement Is Critical to Digital Infrastructure

Drawing from her experience working with large-scale digital infrastructure developments, Miller emphasizes that successful projects require more than engineering and investment, they require meaningful community engagement.

Rather than viewing digital infrastructure solely through the lens of technology deployment, she advocates for creating opportunities for education, transparency, and collaboration. Helping residents understand how data centers, fiber networks, and AI infrastructure contribute to economic growth, workforce development, and future innovation allows communities to participate more effectively in decisions that shape their future.

She also notes that closing the digital divide requires coordinated planning among industry leaders, policymakers, and local communities. As AI becomes more integrated into everyday life, digital literacy and critical thinking will become just as important as physical access to technology.

The conversation ultimately reframes the digital divide as a challenge of participation rather than connectivity, one that requires greater trust, education, and collaboration across the entire digital infrastructure ecosystem.

This content originally appeared in Authority Magazine on Medium and has been adapted for syndication on Data Center POST. Read the complete interview here.

The post Ilissa Miller on the Digital Divide: Why Understanding Matters More Than Access appeared first on Data Center POST.

TLDR;

•  Foresight has been named a Platinum winner in the 2026 Pinnacle Technology Awards for its Predictive Project Delivery platform, recognizing its role in helping infrastructure teams identify schedule risks earlier and improve delivery certainty.
• The award follows a year of strong momentum for the company, including winning Best AI Innovation at the 2026 Datacloud Global Congress and closing a $25 million Series A funding round.
• As AI and digital infrastructure investment continues to grow, Foresight is helping organizations reduce project delays, protect schedule value, and bring critical infrastructure online faster.

# # #

As the pace of digital infrastructure development continues to accelerate, organizations are looking beyond traditional project management tools to improve delivery certainty. That shift is reflected in the 2026 Pinnacle Technology Awards, where Foresight has been named a Platinum winner for its Predictive Project Delivery platform.

The Pinnacle Technology Awards recognize organizations that are driving innovation through technology while delivering measurable business outcomes and demonstrating market leadership. Foresight was honored for helping owners and operators of major infrastructure programs identify delivery risks earlier, improve schedule certainty, and accelerate time-to-revenue across data centers, energy infrastructure, advanced manufacturing, transportation, and other large-scale capital projects.

The recognition comes at a pivotal time for the industry. Global investment in AI infrastructure continues to surge, but delivering projects on schedule remains one of the sector’s greatest challenges. Nearly nine out of ten data center projects experience schedule overruns, making early visibility into delivery risk increasingly valuable for organizations racing to bring new capacity online.

Rather than relying on historical reporting alone, Foresight’s Predictive Project Delivery platform continuously analyzes project data to identify emerging schedule risks before they escalate. The platform provides leadership teams with actionable insights that help improve decision-making and reduce the impact of project delays. Across a global portfolio of 15 concurrent data center projects, Foresight helped reduce reporting overhead by 90%, while preventing more than $100 million in projected lost revenue through earlier identification of schedule risks.

“This recognition reflects a growing industry realization that project delivery is fundamentally a data and decision-making challenge,” said Dr. Atif Ansar, Co-Founder and Executive Chairman of Foresight. “Organizations have long invested heavily in systems that help manage cost and capital, yet time remains one of the least governed and most valuable assets in infrastructure delivery.”

The award adds to a series of recent milestones for Foresight, including being named Best AI Innovation at the 2026 Datacloud Global Congress and closing a $25 million Series A funding round led by Macquarie Capital Venture Capital.

To continue reading, read the full press release here.

The post Foresight Recognized with 2026 Pinnacle Technology Award for Advancing Infrastructure Project Delivery appeared first on Data Center POST.

TL;DR

• A severe voter backlash against Kevin O’Leary’s proposed 9-gigawatt Project Stratos AI data center led to a major political upset in Utah’s primary election, ousting Senate President Stuart Adams and two county commissioners.
• The project’s failure stemmed from bypassing local control using an unelected state authority (MIDA) and dismissing community anxieties over massive water and double-state-capacity power demands as foreign-funded misinformation.
• Utah has raised the regulatory bar for digital infrastructure through Governor Spencer Cox’s “Operation Gigawatt” initiative, enforcing a strict new framework of total transparency, rigorous environmental reviews, and utility ratepayer protections.
• Quietly succeeding developments like the Delta Gigasite, Joule Energy Campus, Pole Canyon, and Antelope Data Campus prove that engaging directly with local officials, securing local permits, and respecting community resource constraints is the only viable path to getting built.

# # #

When Utah Senate President Stuart Adams conceded his primary election defeat recently, it sent a shockwave through the state’s political establishment. For a politician so deeply entrenched, an upset by a political newcomer was considered virtually impossible just a few months ago. Along with Adams, two Box Elder County commissioners were also ousted by voters.

To outside observers, this sudden reshaping of Utah’s leadership seemed unforeseen. But to anyone closely tracking the digital infrastructure space, the cause was clearly a severe, localized voter backlash against Kevin O’Leary’s massive, 9-gigawatt Project Stratos.

The Shark Tank investor’s multi-billion-dollar AI data center campus has quickly become a textbook example of how not to build digital infrastructure. The trouble with Stratos was never a lack of capital or vision, or even the project’s merits; it was a fundamental failure of process, transparency, and respect for the community.

The Mistake of Bypassing Local Control

Project Stratos was aggressively pushed through using the Military Installation Development Authority (MIDA), an unelected state entity. Critics suggest this was a state-level workaround that effectively stripped residents of permanent control over land use, water rights, and taxation.

When residents voiced anxieties about how a 40,000-acre project that would demand double the state’s current electrical capacity could impact their water and agricultural way of life, they weren’t met with collaboration. Instead, they were met with an outsized executive ego in O’Leary that publicly dismissed grassroots opposition as misinformation funded by foreign interests.

This heavy-handed, top-down approach backfired. Sustained protests, lawsuits challenging MIDA’s authority, and hundreds of formal objections forced the developer to withdraw multiple water permit applications. Even after Senator Adams pressured O’Leary to slash the project’s physical footprint by 75% before the primary, the political damage was done. The post-facto work on the ground failed because the human element had been ignored.

Operation Gigawatt: Utah Raises the Regulatory Bar

This high-profile collapse stands in sharp contrast to Utah’s actual macroeconomic goals. Governor Spencer Cox has actively championed “Operation Gigawatt,” a sweeping state initiative designed to rapidly scale Utah’s energy infrastructure to support the AI boom. The state explicitly wants data center investment.

However, as the Stratos controversy peaked, leadership recognized the extent of public concern over unbridled tech growth, the lack of understanding of data centers, and the need to respect local voices and bring the community along. Governor Cox issued a decisive executive order to raise the bar for data center development. The mandate established a strict new framework requiring total transparency, rigorous environmental reviews, and strong protections for local utility ratepayers. Operation Gigawatt proved that Utah wants growth, but only if developers align with local values.

The Quiet Winners Getting Things Built

While Project Stratos dominated the news cycle and tanked political careers, a completely different story was playing out across the rest of the state. Other major hyperscale and AI developments have proved that when you respect local autonomy and invest time in a community, you can actually get things built. Several large-scale projects have quietly navigated the right path, securing their permits and putting shovels in the dirt without the political drama.

In Millard County, both the 9,700-megawatt Delta Gigasite and the 4,000-megawatt Joule Energy Campus are moving forward. Developers like Creekstone Energy and Joule Capital Partners skipped the state-level workarounds. They engaged directly with county officials, secured proper Heavy Industrial zoning, and obtained the necessary Conditional Use Permits for their on-site natural gas generation. Today, both multi-gigawatt sites are fully permitted and actively under construction.

The same holds true for Tract’s 1,700-megawatt Pole Canyon development in Eagle Mountain. Tract progressed through transparent planning with city officials to establish long-term infrastructure guidelines rather than using backdoor maneuvers. Down in Iron County, Pronghorn Development recently secured its local permits for the 1,500-megawatt Antelope Data Campus. They did so by strictly complying with local regulations, investing time in local community outreach, treating resource limits as engineering parameters rather than obstacles to bypass.

The Real Lesson: A Social License to Operate

The message out of Utah’s primary election is a stark warning to hyperscalers and developers nationwide. The AI expansion demands unprecedented amounts of power and land, but it still fundamentally requires a social license to operate.

You cannot buy, legislate, or steamroll past your way past a community’s real anxieties regarding resource scarcity—whether rooted in technical misunderstanding or deep-seated regional concerns. Utah remains one of the most promising frontiers for digital infrastructure in the world. But as the dust settles on the ballot boxes, the industry must absorb the real lesson of the Beehive State: a loud, top-down approach driven by overconfidence and insularity may trigger a fatal political reaction and mire a project in controversy.

The developers who treat local communities as partners, rather than obstacles to be bypassed, will be the ones building the future.

# # #

About the Author

Scott Cuthbertson is the Founder and CEO of Alpen Systems, a boutique advisory firm helping developers, communities, and contractors navigate large-scale infrastructure deployment in the energy and data center sectors globally. He previously served as the President and CEO of the Economic Development Corporation of Utah (EDCUtah), where he directed corporate recruitment and regional infrastructure strategy for one of the nation’s fastest-growing economies. Prior to his leadership in public-private economic development, Scott spent 15 years as a capital projects and infrastructure management consultant at PwC and Booz Allen Hamilton, advising global enterprises and investors on large-scale asset development.

The post The Limits of Top-Down Expansion: What Utah’s Primary Election Teaches the Data Center Industry appeared first on Data Center POST.

TL;DR

• AI Workloads are Driving Unprecedented Power Densities: The massive compute demand from artificial intelligence is pushing rack densities to extreme new highs, forcing an industry-wide transition from traditional air cooling to more efficient liquid and immersion cooling technologies.
• AI and Smart Tech are Powering Green Data Centers: Operators are increasingly relying on AI-driven energy management and continuous infrastructure monitoring, which can reduce cooling energy consumption by 20% to 40% and help achieve highly efficient Power Usage Effectiveness (PUE) ratings.
• Localized Manufacturing Enhances Speed and Sustainability: To quickly deploy complex, liquid-ready environments, providers like DPI are utilizing onshore, localized manufacturing (such as in the UAE) to eliminate global shipping bottlenecks, cut embodied carbon emissions, and reduce deployment timelines from months to weeks.

# # #

As digitization advances into our society so too do data centers, and they have become one of the largest users of electricity worldwide. According to the International Energy Agency (IEA) global electricity consumption by data centers is about 1%-1.5%. Given the extraordinary growth in AI, cloud, and edge computing, the demand for this will continue to grow. Therefore, the expansion of this industry has resulted in using sustainable principles in constructing data center infrastructure and providing a basis for furthering the development of green data centers.

Currently, there are no defined characteristics for identifying a data center as being a green facility. Instead, green data centers rely on energy efficient hardware, sophisticated software, usage of renewable energy, and future friendly cooling methods to increase performance while minimizing their impact on the environment.

Using AI Based Energy Management to Improve Operational Efficiency

AI technology is one of the main driving forces behind sustainable data center operation improvements. AI-driven monitoring platforms analyze thousands of operational parameters continually, including the use of the server, cooling demand, airflow and electricity.

Intelligent optimization systems can help lower cooling energy usage by 20% to 40% based on facility size and workload characteristics, according to research. In addition, machine learning algorithms predict potential equipment failure before it happens, thereby reducing the chances for unnecessary maintenance and hardware waste.

As the use of AI workloads continues to rise, operators are continuing to utilize AI not just in servers but also to optimize the supporting infrastructure.

Temperature Control in Data Centers is on the Rise

Cooling typically accounts for 30-40% of a conventional data centre’s energy consumption, which provides an opportunity to make substantial sustainability improvements.

There are several next-generation cooling techniques that are spurring change.

• Liquid Cooling – A liquid cooling technique that uses specialized coolants to transfer heat directly from processors, thereby enabling improved heat dissipation.
• Free Air Cooling – Adjusting cooling systems to take advantage of outdoor air’s temperature, which can significantly reduce mechanical cooling requirements.
• Immersion Cooling – Submerging servers in dielectric fluids, which reduces energy used for cooling and allows for the operation of very dense AI computing workloads.
• Intelligent Airflow Management – Implementing temperature sensors and automated control of airflow to minimize temperature variation across server racks.

The majority of the new hyperscale data center facilities that are currently under development incorporate the use of multiple cooling techniques to optimize energy efficiency and reduce total energy consumption.

Renewable Energy is becoming a Core Strategy for Infrastructure.

Renewables have shifted from simply being an option to becoming a fundamental part of how green data centers are built. There is now an increase in how much renewable energy companies are utilizing through either direct investments or long-term power contract agreements; many companies’ sources of electricity now occur from the sun, wind, hydro, or geothermal.

Previous industry data assessment suggests many large/massive data center companies are using 90%+ of their yearly power through renewables in locations around the world.

Another report conducted by Data Intelo suggests the Size of the green data centre Industry will grow from $92.4 billion in 2025 to $298.7 billion by 2034; thus, forecasting a CAGR rate of 13.9% between 2026-2034 should reflect increasing amounts being invested by companies towards energy-efficient/low carbon sources; as well as sustainable operations worldwide.

Energy Efficient Hardware Supports Sustainable Computing

Today’s CPUs, memory, storage and networking devices consume far less electricity than those of previous generations but provide greater computing power.

An energy-efficient power supply can achieve up to 96% efficiency which means lower losses during the conversion of alternating current (AC) into direct current (DC) and vice versa. Virtualizing your servers allows you to co-locate multiple applications on one physical machine which results in higher average utilization (historically 15%-20% of capacity) now exceeding 60% in most places.

SSD’s have also replaced many traditional HDDs reducing electrical demand while simultaneously improving performance across a variety of applications.

Continuous Optimization Through Smart Infrastructure Monitoring

Operators benefit from having immediate access to information about the performance of their facilities thanks to IoT sensors, digital twins, and the implementation of automated systems for managing buildings.

By utilizing a range of sensor technologies, such as monitoring temperature, humidity, airflow, electrical load and the health of equipment at least once every second, facilities can receive an ongoing supply of operational data to use in making automated adjustments that improve energy performance without compromising on reliability.

In addition to the use of the above technologies, many facilities monitor their PUE (power usage effectiveness) and are now seeing values as low as 1.2 for newer generation Green Data Centers, while older generation data centers typically operate at values between 1.7 and 2.0.

Sustainable Practices Are Evolving into Technology-Based Actions

The evolution of sustainable development and the transition of green data centres indicates that sustainability is more heavily reliant on technology (intelligent technology) versus traditional stand-alone initiatives focused on environmental improvements.

Many technologies, including AI-based optimisation of data centres; cooling systems; renewable energy integration; high-efficiency hardware; and continuous monitoring of the physical infrastructure, have come together to define how digital infrastructure is being created and operated.

With the ongoing growth of demand for cloud computing and artificial intelligence, technologies will increasingly provide a means to achieve an appropriate balance between sustainable long-term environmental responsibility, and the ever-increasing demand for computation.

# # #

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

The post What Technologies Are Powering the Evolution of Green Data Centers? appeared first on Data Center POST.

TL;DR

• Generative AI is changing data center planning by introducing higher power density, greater cooling requirements and more complex infrastructure.
• Traditional forecasting models were designed around relatively predictable enterprise and cloud workloads, not large-scale AI training and inference clusters.
• AI workloads create significant challenges in power provisioning, rack density, cooling design, network architecture and capacity planning.
• Data center operators are responding with new planning techniques, liquid cooling deployments, higher-capacity power systems and more flexible infrastructure designs.

# # #

The rise of generative AI is forcing a significant shift in data center planning. What began as a technological breakthrough has quickly evolved into an infrastructure challenge, as organizations race to deploy increasingly powerful AI models that demand vast computing resources.

Why Traditional Data Center Planning Is Struggling

Historically, data center planning focused on balancing anticipated business growth against available infrastructure capacity. Planners examined historical trends, projected future demand and designed facilities around gradual increases in utilization.

This approach worked because most workloads exhibited relatively consistent behavior. Storage growth, virtualization adoption and cloud migration initiatives typically followed multiyear trajectories that could be modeled with reasonable accuracy.

However, generative AI workloads behave differently. Training large language models and other advanced AI systems requires massive clusters of GPUs and accelerators operating simultaneously. These deployments can dramatically increase power demand within a short period, creating infrastructure requirements that exceed what many facilities were originally designed to support.

The Power Density Problem

Perhaps the most visible impact of generative AI is the large increase in rack power density. For years, many facilities were designed around rack densities ranging from 5 kW to 15 kW. Even in high-performance environments, conventional cooling and power distribution systems often remained within manageable thresholds.

However, modern AI deployments frequently operate at densities far beyond those levels. GPU-intensive racks can exceed 50 kW or even 100 kW per rack, depending on configuration.

These increases create significant challenges throughout the facility. Energy distribution systems must deliver more electricity to individual racks, and backup power infrastructure must accommodate larger loads. Utility connections will require expansion, and capacity planning models that once projected gradual increases now must account for sudden spikes driven by AI initiatives.

The U.S. Department of Energy reported that domestic data center electricity consumption is projected to double or triple by 2028, due to AI adoption and expanding digital infrastructure needs.

Cooling Systems Are Reaching Their Limits

Power and cooling have always been interconnected, but generative AI is intensifying that relationship. As computing density increases, heat generation increases accordingly. As such, traditional air-cooling systems that worked effectively for lower-density environments can struggle to remove heat from modern AI clusters.

Many data centers now face thermal constraints to the point that they may have sufficient electrical capacity available but lack the cooling capability needed to support additional AI infrastructure. Operators are recognizing that traditional cooling approaches alone may not be sufficient for future AI growth.

This reality is forcing planners to rethink facility design. Airflow management, containment strategies, cooling distribution and equipment placement are becoming increasingly critical considerations.

Network and Capacity Assumptions Are Changing

Generative AI is also challenging conventional assumptions about network architecture and capacity utilization. Large AI clusters require substantial east-west traffic as GPUs communicate during training and inference. Additionally, network bottlenecks that might have been acceptable in traditional environments can negatively impact AI performance.

Data center operators must consider higher-bandwidth networking infrastructure, low-latency interconnects and scalable architectures capable of supporting data-intensive workloads. Infrastructure decisions that once centered on server deployment now require a more holistic evaluation of power, cooling, networking and storage resources.

Similarly, utilization models are becoming more difficult to predict. AI projects can scale rapidly, creating infrastructure demands that exceed previous forecasts. Therefore, organizations often move from experimentation to production deployment much faster than the traditional application life cycle. This creates greater uncertainty and a need for more adaptive planning frameworks.

How Data Center Planning Is Evolving

To address these challenges, operators are fundamentally rethinking how data center planning is performed. They are redesigning facilities, reevaluating planning assumptions and making significant investments to support a new generation of high-density workloads.

Rather than relying primarily on historical growth trends, planners are incorporating scenario-based modeling. This approach evaluates multiple potential growth paths, including aggressive AI adoption scenarios that may exceed historical demand patterns. As such, flexibility is now becoming a key design principle.

New facilities are often designed with higher baseline power capacities, modular expansion capabilities and infrastructure to accommodate future increases in rack density. This shift to adaptable and modular design allows for capacity to be added incrementally as AI demand evolves.

Cooling infrastructure is also undergoing rapid transformation. While air cooling remains effective for many workloads, operators supporting large-scale AI deployments are now adopting liquid technologies. Direct-to-chip liquid cooling and other advanced thermal management solutions allow facilities to support rack densities that would be difficult or impossible to manage through conventional methods alone.

Power planning is also becoming more integrated with business strategy. Operators are securing utility capacity years in advance and exploring alternative energy strategies to ensure long-term scalability.

Real-World Adaptation in the AI Era

Some of the industry’s largest operators are already making substantial adjustments. For example, many colocation providers are upgrading existing facilities to support higher-density deployments and redesigning power distribution systems to accommodate AI customers.

As AI loads grow, many colocation data centers are moving toward single-tenant facilities. This allows operators to dedicate power, cooling and network resources to large-scale AI deployments without the constraints that can arise in multitenant environments.

Additionally, hyperscale providers have invested heavily in AI-optimized infrastructure, including advanced cooling technologies, expanded power capacity and purpose-built data center designs. For example, Microsoft will build a renewable, scalable next-generation hyperscale AI data center valued at $6.2 billion in Narvik, Norway.

The Future of Data Center Planning

Generative AI reveals the limitations of traditional data center planning models and drives innovation in the industry. The assumptions that guided facility design for decades, including predictable growth rates, moderate rack densities and stable power requirements, are being challenged by AI-driven computing demands.

Infrastructure needs will evolve rapidly as organizations integrate generative AI into products, services and operations. Data center planners need to use flexible forecasting, adopt higher-density architectures, and prepare for higher power and cooling demands.

# # #

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 Generative AI Workloads Are Breaking Traditional Data Center Planning Models appeared first on Data Center POST.

Originally posted on 1547realty.

TL;DR

• Hybrid and multi-cloud strategies depend on connectivity, not just colocation space.
• Carrier-rich, interconnection-focused facilities improve performance, flexibility, and scalability.
• 1547’s connected data centers help organizations build low-latency, future-ready infrastructure.

# # #

Modern colocation facilities have evolved far beyond providing basic power, cooling, and space, transforming into the critical connective tissue for hybrid and multi-cloud architectures. Driven by concerns over latency, egress costs, and vendor lock-in, the vast majority of organizations now blend private, on-premises systems with multiple public cloud platforms. This architectural complexity is increasingly managed within carrier-dense interconnection hubs, which provide the resilient and flexible foundation needed to connect private deployments directly with hyperscaler environments.

1547 Critical Systems Realty addresses this shift by offering facilities that function as rich, interconnected ecosystems rather than simple hardware storage. These hubs grant enterprises access to diverse carriers, direct cloud on-ramps like AWS Direct Connect and Azure ExpressRoute, and private peering opportunities. By anchoring infrastructure in strategically located, connectivity-rich markets, organizations can avoid unnecessary network hops and maintain the low-latency performance essential for modern and AI-driven workloads.

As U.S. data center demand is projected to nearly triple by 2030, evaluating infrastructure through the lens of “data gravity” has become paramount. Facilities equipped with a dense mix of existing network and cloud providers create a compounding ecosystem advantage, allowing organizations to provision new cross-connects in hours rather than weeks. Because 1547 specifically acquired its facilities for their established carrier density, these environments are inherently designed to provide the seamless, scalable connectivity that hybrid and multi-cloud IT strategies demand.

To continue reading, please click here.

The post Beyond Basic Colocation: Building Hybrid and Multi-Cloud Hubs in Interconnection-Rich Data Centers appeared first on Data Center POST.

TL;DR

• The Shift to Inference: As AI transitions from model training to revenue-generating applications, inference systems must be managed as complex chains where a single weak link can bottleneck the entire system.
• Listening to System Diagnostics: Instead of treating inference as a black box, teams must learn to translate continuous performance signals, such as “Time to First Token” or latency spikes, to accurately diagnose issues like memory constraints or scheduler pressure.
• Workload-Specific Testing: Generic benchmarks fail to predict real-world performance because different AI applications (e.g., fraud detection vs. legal analysis) stress systems in unique ways; testing must model actual prompt shapes, multi-turn sessions, and traffic bursts.

# # #

Over the last several years of AI expansion, it’s felt like everyone’s been obsessed with scale. Train bigger models, build bigger clusters, and run bigger experiments. But scale only gets you so far. The next phase of AI will revolve around revenue, turning big models into big business.

For technology leaders, that shift revolves around inference. Training is where a model learns; inference is where it provides answers and where it delivers value. It’s how AI models handle customer complaints, flag fraudulent transactions, draft legal briefs, explain lab results, or answer a developer’s question at 2 a.m. However, it’s also where costs accumulate, latency becomes a UX problem, and operational blind spots get increasingly expensive.

In response, people tend to treat inference like any other GPU problem: buy more accelerators, tune the model, and let the tokens flow. But inference systems demand a different equation. Inference pipelines run across load balancers, security layers, retrieval systems, KV-caches, memory bandwidth, storage, network fabrics, orchestration software, and GPU compute. That means a bottleneck in any one place slows everything else downstream. Without proper tuning, costly infrastructure can sit idle while users wait for models to respond.

The problem stems from looking at stacks like they’re a single machine. Instead, we should consider inference as a chain; and chains can fail at their weakest link.

What the numbers say

It’s tempting to think of inference like a black box, an impenetrable fortress that leaves you at the mercy of lagging indicators like latency complaints 6 months after deployment. But that couldn’t be further from the truth. Inference systems are constantly generating performance signals. The problem is that most teams don’t speak their language enough to understand them in context.

Take Time to First Token, for example. If it’s climbing, that often means prefill compute is struggling with long prompts or retrieval-augmented context. If decode cadence is wobbling, memory bandwidth may be the constraint. Perhaps KV-cache is swelling? That means long sessions or agentic workflows are expanding the memory footprint in ways the architecture wasn’t designed for. Maybe P99 latency is spiking, even though median latency looks clean? That likely means that burst traffic is creating queue buildups, or the scheduler is under pressure.

These signals aren’t isolated issues. They’re diagnostics. But they show up scattered across different teams, different monitoring systems, and in different contexts. Connecting them demands a layer of telemetry that most organizations have yet to build.

Why workload type matters

One of the key challenges with inference systems is reliably predicting performance. Pre-deployment benchmarks and load tests pass, and the architecture diagram makes sense. Then the system goes live and users reveal something that synthetic tests failed to anticipate.

The problem is the inference workloads themselves. They behave in different ways based on their model’s industry application and use case. Simple “one-size-fits-all” tests cannot account for this kind of variability and oftentimes mask the specific problems that determine real-world performance.

For example, legal AI tools that read long contracts will stress context windows and memory. Fraud detection systems care about microseconds, not minutes. Healthcare workflows need sustained throughput and careful data handling. Ecommerce chatbots need to absorb traffic spikes that arrive without warning. Developer copilots accumulate state across long conversations, growing their memory footprint with every turn.

Each of these applications present different problems with different bottlenecks. If you were to benchmark any of these models against generic HTTP traffic or peak GPU throughput, you’ll get a clean number that tells you next to nothing about how the system will ultimately behave when real users show up with messy prompts, uneven sessions, and edge cases your test harness never covered.

Three things that need to change

1. Test against the actual workload. 

“How fast are our GPUs?” is the wrong question. The right one is more along the lines of: “How does our system perform when users, applications, security requirements, and traffic patterns hit it at the same time?” Answering that means modeling real prompt shapes, real response sizes, real bursts, multi-turn sessions, retrieval calls, and the adversarial inputs that show up in production.

2. Observe the system as a whole.

A clean metric in one layer doesn’t mean the system is healthy. Inference performance is determined by the weakest link across compute, memory, storage, networking, and orchestration. That means getting time-aligned telemetry that connects what went into the stack with how each layer responded. That way, it’s easier to distinguish between compute limits, memory constraints, network delays, retrieval bottlenecks, and guardrails that aren’t scaling under load.

3. Optimize for Cost Per Token, not just speed. 

Throwing more hardware at a bottleneck is often the most expensive way to fix it, if it’s not the wrong fix entirely. Instead, a more precise method of intervention is identifying weak links, such as subsystems and workload shapes. Sometimes the answer is more memory, other times it’s better batching, a redesigned storage layer, or a change in how orchestration handles concurrency. However, the goal is always the same: ensuring predictable latency, lower cost per token, and right-sized infrastructure deployments that aren’t overprovisioned as insurance against uncertainty.

The larger shift

Teams that understand inference well are moving from infrastructure-level confidence to system-level assurance. It’s no longer sufficient to know that each layer works. For CIOs and CTOs, this changes the questions worth asking. Try looking deeper under the hood by determining “Which workloads are driving our economics?” or “Can we show that a new architecture lowers cost per token without degrading the user experience?”

Inquiries like these turn inference from a black box into something measurable, auditable, and improvable over time. They also create a common tongue between technical and executive teams. In this language, performance depends on evidence, cost is something traceable, and risk is something testable rather than assumed.

Figuring this out means you can make better architecture decisions, avoid expensive over-investments, and know what to fix before users ever notice it needs fixing. After all, the infrastructure already has the answers. The challenge is learning how to translate.

# # #

About the Author

Mike Hodge is AI Solutions Lead at Keysight, where he drives global strategy and go-to-market execution across the company’s AI, network test, and security portfolios. He specializes in connecting innovation with real-world applications, helping organizations harness AI for smarter, more secure systems.

The post Inference, Decoded: Unlocking AI’s Potential by Learning the Language of Inference Systems appeared first on Data Center POST.

TL;DR

• Context Over Quantity: Finding more vulnerabilities no longer guarantees better security; relying solely on standard severity scores often wastes resources on low-impact alerts while delaying critical fixes.
• Prioritizing Business Risk: To properly manage threats, security teams must incorporate real-world business context to prioritize the most dangerous vulnerabilities.
• Automation is Essential: To keep pace with AI-driven discovery, organizations must move away from fragmented, manual tools and adopt unified security orchestration to automatically trigger, assign, and track remediation workflows.

# # #

For years, cybersecurity teams have operated under a shared assumption that finding more vulnerabilities will help reduce risk. That assumption no longer holds.

With the emergence and rapid advancement of cyber-capable frontier LLMs such as Claude Mythos and GPT 5.5 Cyber, discovery is not the limiting factor. Models are now capable of identifying vulnerabilities at a scale and speed that far exceeds what human teams can process. This includes not only known issues, but a potential slough of zero-day vulnerabilities surfacing in software and infrastructure previously believed secure.

As these capabilities become more widely available, organizations are entering an inflection point in security operations. Their challenge will be to sort through the noise and make sense of found issues and their potential business impact before attackers can exploit newly uncovered weaknesses.

The Vulnerability Tsunami is a Context Problem

The industry is being approached by what can truly be described as a vulnerability tsunami. As Mythos-class AI dramatically accelerate vulnerability discovery, the number of identified issues will mount to previously unseen levels. But more findings quickly become the enemy of an overburdened security team; and without equally accelerated triage & remediation, more findings ≠ better security.

In practice, most organizations are struggling to translate this influx of data into meaningful action. Vulnerabilities are flagged, categorized, and scored, but often without the context needed to determine their real impact on the business.

A critical vulnerability in isolation does not tell the full story. Is the affected asset exposed to the internet? Is it tied to sensitive customer data? Is there a compensating control already in place? Without answers to these questions, security teams must make judgment calls based on incomplete information.

This leads to inefficiencies across the board. Analysts spend large amounts of time triaging alerts that don’t matter. Development and operations teams receive remediation requests that may not align with actual risk. Important issues become delayed, while less impactful ones consume valuable resources.

There is a growing disconnect between what organizations know about their risk and what they can actually do about it.

Prioritizing by Business Risk

To move forward, enterprises need to rethink how they prioritize vulnerabilities. Traditional methods that rely primarily on severity scores are no longer sufficient in a high-volume environment.

What is needed is a more complete view of risk. We need to incorporate business context alongside technical findings. This includes understanding how vulnerabilities map to critical assets, how those assets support business operations and key data, and how likely a given issue is to be exploited in the real world.

When context is applied consistently, prioritization becomes more focused and defensible. Security teams can direct their attention to the vulnerabilities that pose the greatest risk to the organization, rather than trying to address everything at once.

This change also improves collaboration. When remediation requests are backed by clear business impact, they are easier for engineering and operations teams to understand and act on. Conversations move from abstract severity levels to concrete risks, which helps reduce friction and accelerate response.

Automation is Now a Requirement

Even with better prioritization, the scale of modern enterprise environments makes manual remediation unsustainable. As discovery continues to accelerate, the gap between identification and resolution will widen unless organizations adopt more automated approaches.

Automation plays a critical role in turning insight into action. When vulnerabilities meet defined risk criteria, response workflows should be triggered automatically. This can include creating and assigning tickets, enriching them with needed context, and tracking progress through to completion.

The goal is to ensure that routine actions happen consistently and without delay. Standardizing how vulnerabilities are remediated and managed helps organizations reduce bottlenecks and improve overall efficiency.

More importantly, automation helps establish accountability. Every identified risk should have a clear owner, a defined path to remediation, and visibility into its status. This level of coordination is difficult to achieve through manual processes alone, though human oversight is still very much needed in the process.

Orchestrated Security vs. Managing Fragmented Tools

Leading organizations are already adapting to this new reality. They are moving away from fragmented toolsets that generate isolated findings and toward taking more unified approaches that bring data together in one place.

Security teams are moving towards improved security orchestration. They need the ability to correlate findings across different systems, apply consistent context, and drive coordinated responses.

When done effectively, this approach changes how security is managed and measured. Instead of focusing on the number of vulnerabilities discovered, organizations can track how quickly and effectively risk is reduced. Reporting becomes more aligned with business priorities, giving leadership a clearer view of where the organization stands.

The pace of vulnerability discovery will only increase from here. As models like Mythos and GPT 5.5 Cyber become widely adopted for vulnerability discovery, organizations will have access to more rapid insights about their security gaps than ever before–more than they can triage and resolve without help. And as attackers are already harnessing these same models to develop next-gen threats, the time to elevate vulnerability management programs to AI-scale is now. Investing in context-driven prioritization and automated orchestration will make the difference between floating above the tsunami, or drowning under it.

# # #

About the Author

Mark Lambert is the Chief Product Officer for ArmorCode, a leading application security posture management (ASPM) provider. Mark has built products for more than 20 years, and helped organizations streamline the delivery of secure, reliable and compliant software applications across the enterprise, embedded and IoT markets.

Prior to ArmorCode, he held product leadership positions with Parasoft, Advanced Visual Systems (AVS) and more. Mark holds a bachelor’s and master’s degree in computer science from Manchester University, UK.

The post When AI Finds Everything, Context Will be the Future of Vulnerability Management appeared first on Data Center POST.

TL;DR

• Transparency Builds Trust: While hyperscale AI data centers often face local pushback over resource usage, developers can earn community trust through responsible, transparent planning rather than just focusing on facility size.
• Water Conservation: Utilizing closed-loop cooling systems drastically reduces continuous water consumption, alleviating concerns about straining local water supplies.
• Economic Benefits: Beyond infrastructure, data centers drive local economic growth by creating jobs in both construction and long-term operations, while boosting nearby businesses.
• Mutual Success: Protecting a community and developing AI infrastructure are not mutually exclusive; responsible developers can provide necessary technological advancements while safeguarding local resources.

# # #

Around the country, AI data centers on a grand scale are increasingly a source of local controversy. In some cases, citizens have expressed concerns in “Save Our City” campaigns, worrying about excessive water consumption, grid strain, traffic and more from these massive facilities.

It is understandable that people would worry. For one thing, data centers are no longer seen as mere silent entities at the edge of the digital world. With the advent of artificial intelligence, data centers have become bigger, more energy-intensive, and more noticeable.

That doesn’t mean a valid concern equals actual harm.

The real question is whether the project has been conceived responsibly. The design for an advanced AI data center can take care of all of the key concerns – water conservation, grid safety, environmental sustainability, and economics.

Water Use Must Be Understood, Not Assumed

Water is typically the first question when discussing data centers. And that’s particularly true in areas where there is already an issue of water availability. In a closed-loop cooling approach, the cooling medium circulates through a contained system instead of being released through evaporation. In cases of high-density AI loads, this could take the form of a liquid-cooled system, a heat exchanger, a coolant distribution unit, a dry cooler, or air-cooled chiller. The critical community issue is not just whether water is used in the facility at all but, more importantly, how much water would be added after the initial filling, whether that water is potable, and what the facility’s water usage effectiveness (WUE) is.

Not all data center cooling systems are equal. Conventional evaporative cooling can mean continuous water consumption. A closed-loop system, by contrast, circulates cooling fluid within a sealed circuit, so it does not draw water continuously.

It makes a difference.

If the facility uses a closed loop system, then the discussion must revolve around real facts about water usage: how much water is needed; whether it is a one-off filling or regular; what the source of water is; how maintenance is done in the system, etc. Sometimes, water may be trucked in or sourced in a way that will not put strain on the community water supply or the city.

People have the right to information. But they also have a right to correct information.

Power Planning Is Part of Responsible Development

The second issue concerns electricity. Citizens are worried about whether the new facility will drive up prices, decrease reliability, or compete with residences and businesses for access to electricity. In terms of AI data centers, power planning goes beyond megawatt capacity; it starts with electrical design. Substations, grid interconnections, utility redundancy, UPS, battery storage, and on-site power generation are all parts of the solution that allows you to isolate the critical loads from local distribution. All of these elements allow you to achieve uptime, manage reliability risk, and define accountability for the necessary infrastructure. This is a legitimate concern. But here too, the answer lies in the infrastructure of the facility in question.

A proper data center is not a facility that suddenly appears and connects to the grid. A data center can have dedicated substations, improved interconnections, power reliability facilities, natural gas turbines, battery backup or other means of redundant power. This is all meant to increase reliability, while at the same time minimizing any strain on the current local infrastructure.

The public is entitled to raise questions such as who pays for the electrical facilities and how the facility will get its power, what are the facilities for backup power, how often the facility will generate power itself and what emissions controls are in place.

But these are very different questions than those that suggest that every data center imposes the same problem.

Economic Impact Should Be Part of the Conversation

Protecting the community matters, and so does recognizing what the community stands to gain.

Large-scale construction of a data center generates real economic activity in the area. The build itself puts electricians, mechanical contractors, civil contractors, truck drivers, safety experts, and engineers to work, and it drives business for nearby hotels, restaurants, suppliers, and service firms.

Once constructed, the data center might offer employment in operations, security, maintenance, facilities, landscaping, and other technical support. It can even assist with attracting workforce training programs and business investment.

This type of development can even become a part of an economic development plan for many communities, if done properly.

Responsible Development and Community Protection Can Work Together

This discussion is often reduced to a dichotomy of either protecting the community or opening it up for development. This is too narrow.

A community can both safeguard its future and provide the necessary infrastructure at the same time. A data center can consume substantial energy and build its own electrical infrastructure. A data center can require cooling and employ closed-loop technology to limit the amount of water consumed. It can be large and still provide local employment and business opportunities.

The competition in AI infrastructure is not only about who can build the biggest, most advanced facilities. The winner will be the responsible developer that earns public trust.

Data centers have become an essential part of America’s digital and economic infrastructure. The responsibility of the industry is to show the community not only what it plans to build but also how it plans to do that and whose interests will be served through this development.

Responsible data center development should not be seen as the opposite of protecting the community. Done right, it becomes one of the ways a community builds its future.

# # #

About the Author

Harshada Jadhao is a Senior Project Engineer at DPR Construction with experience in mission-critical and hyperscale data center construction. She holds a master’s degree in Construction Management & Technology from Arizona State University, along with graduate and undergraduate degrees in engineering from Nagpur University. Her work focuses on schedule and procurement coordination, cross-trade integration, quality control, turnover readiness, and constructability in complex project environments. She has contributed to major data center projects, including the Meta Hyperscale Data Center and the Abilene Data Center, and brings a practical field-based perspective to the evolving role of project management in AI-driven infrastructure.

The post When “Save Our City” Meets the AI Data Center: How Responsible Developers Earn Community Trust appeared first on Data Center POST.

Originally posted on Datalec Ltd.

TL;DR

• AI-Driven Density Surge: AI workloads are pushing data centre rack densities to extreme levels (projected to reach 600kW by late 2027), accelerating a critical industry transition from traditional air cooling to advanced liquid-cooling technologies.
• Localized Infrastructure Manufacturing: To support these complex thermal and structural demands, DPI has expanded its bespoke manufacturing in the UAE to produce liquid-ready whitespace solutions, including heavy-duty Hyperion ceiling systems, modular pods, and dedicated stainless-steel technology cooling loops.

# # #

Artificial intelligence is fundamentally reshaping data centre design by driving unprecedented increases in power density, cooling requirements, and infrastructure complexity. Traditional data centre racks that once operated between 5kW and 15kW are now being replaced by high-performance deployments routinely reaching 227kW, with projections suggesting demands could hit an extraordinary 600kW by late 2027. Because air can no longer efficiently remove heat at these extreme densities, the industry is rapidly transitioning to advanced liquid-cooling technologies, like direct-to-chip and immersion systems, which significantly alters how whitespace environments must be designed and built.

To meet the demands of liquid-ready infrastructure, Datalec Precision Installations (DPI) has significantly expanded its localized engineering and manufacturing capabilities in the UAE. This includes the local bespoke production of freestanding structural Hot Aisle Containment (HAC) units, standardized modular HPC pods, and the Datalec Hyperion Ceiling System, which is uniquely designed to support heavy mechanical and electrical loads without relying on HAC supports. Furthermore, to support the growing need for specialized process water infrastructure, DPI has established a dedicated stainless-steel fabrication workshop in the region to manufacture precision technology cooling loops under rigorous global quality controls.

Beyond structural fabrication, DPI provides complete end-to-end integration and global lifecycle support, working alongside leading Coolant Distribution Unit (CDU) and glycol manufacturers to offer everything from system flushing and commissioning to 24/7 rapid response maintenance. Shifting to an onshore, import-independent supply chain and championing the “Made in the Emirates” mark provides distinct commercial and environmental benefits for these deployments. This localised approach drastically reduces embodied carbon by eliminating international freight, compresses project delivery timelines from months to weeks, and mitigates supply chain risks while actively supporting regional economic growth.

To continue reading, please click here.

The post Engineering the Future: How DPI is Evolving for AI Infrastructure Data Centre Deployments appeared first on Data Center POST.

TL;DR

• AI capacity is set to surge: Structure Research projects AI-focused data center capacity will grow from 2.3 GW today to 150 GW by 2030, reflecting unprecedented infrastructure expansion.
• New research brings clarity: The report provides a bottom-up view of who is funding, building, and consuming AI infrastructure, offering a more complete picture of the rapidly evolving ecosystem.
• Power and strategy will shape the future: Findings highlight power availability as a key constraint while examining how hyperscalers, AI labs, neocloud providers, and sovereign AI initiatives are driving the next phase of AI infrastructure growth.

# # #

Artificial intelligence is driving one of the fastest infrastructure expansions the data center industry has ever experienced, but understanding where capacity is being built and who ultimately consumes it has become increasingly complex.

To provide greater visibility into the evolving market, Structure Research has released its new AI Infrastructure Report, offering a bottom-up analysis of the organizations funding, building, and utilizing AI infrastructure worldwide. According to the report, AI-focused data center capacity is projected to grow from approximately 2.3 gigawatts (GW) today to 150 GW by 2030, representing a 66-fold increase over the next several years.

Built using Structure Research’s proprietary datasets, the report combines company-level infrastructure tracking with analysis of neocloud providers and sovereign AI initiatives to create a unified view of the AI infrastructure ecosystem. The research reconciles both infrastructure ownership and compute consumption, helping organizations better understand where capital is flowing, where capacity is being deployed, and how market dynamics are changing.

Among its key findings, the report identifies power availability as one of the largest constraints to future AI infrastructure growth and highlights significant differences in how organizations convert infrastructure investment into available compute capacity. It also examines how hyperscalers, AI labs, neocloud providers, and sovereign AI programs are pursuing increasingly distinct infrastructure strategies that will shape the market through 2030.

The AI Infrastructure Report is designed for hyperscalers, cloud providers, data center operators, investors, policymakers, and enterprise technology leaders seeking a deeper understanding of the forces reshaping AI infrastructure deployment and the future of global compute capacity. Executive summaries are now available here from Structure Research.

The post Structure Research Forecasts AI Data Center Capacity Will Reach 150 GW by 203 appeared first on Data Center POST.

Data Center Post had the opportunity to connect with Andy Fenton, an accomplished telecom and data centre executive with over 20 years of experience in the industry. Currently serving as the VP of Sales and Marketing for Telehouse Canada, Andy oversees the company’s sales strategy to drive long-term growth and manages relationships with major telecom and big tech clients. Beginning his career at PSINet, he managed sales, business development, and the national sales team, where he gained valuable expertise in the colocation business. He then transitioned to Bell Canada as the General Manager for Bell’s Systems Integrator Group, managing relationships with key partners, including Accenture, Wipro and IBM Canada.

Prior to joining Telehouse, Andy spent 14 years at Cogeco as head of the Carrier and Strategic Accounts Team playing a crucial role in the company’s growth from a single location to over 15 data centres across Canada, the U.S. and the U.K. Over his career Andy developed and managed relationships with major telecom and technology companies including IBM, Accenture, Microsoft, Rogers, Telus, Hydro One and Tencent.

What does your company do?

As a subsidiary of KDDI Corporation, a global leader in telecommunications, Telehouse brings 35 years of global expertise in providing reliable, secure, and flexible colocation services. Our family of data centres are all strategically placed in more than 40 locations around the world. From Paris to Shanghai, we maintain optimum SLA uptime standards of 99.999%.

In Canada, Telehouse operates three interconnected data centres in downtown Toronto, forming one of the country’s most established connectivity hubs. Telehouse Canada’s facilities provide direct access to a dense ecosystem of carriers, cloud providers, ASPs, and ISPs, enabling fast, low‑latency connectivity to key North American markets. Telehouse Canada supports organizations across diverse industries with secure, resilient, and highly interconnected infrastructure that drives growth and enables global reach.

We offer both shared and dedicated colocation space in all our data centres. Each data centre is connected through our dark fibre network, providing rapid access to a range of 200 + connectivity providers, including leading cloud service providers.

Our data centres act as the nerve centre of Canadian telecommunications, serving as a meeting point for internet service providers (ISP), application service providers (ASP) and Canada’s largest carrier networks to connect and exchange data.

Telehouse also supports high-density AI workloads while enabling edge deployment closer to end users through robust connectivity and interconnection.

What problems does your company solve in the market?

Telehouse Canada helps organizations address critical infrastructure challenges, including the need for diverse, reliable low- latency connectivity, scalable capacity, and secure, high‑performance environments to support digital growth and AI workloads.

As part of a global platform of data centres, Telehouse enables direct access to leading cloud platforms, helping businesses to seamlessly connect to the services they need to optimize their infrastructure, support hybrid and multi-cloud strategies, and expand internationally.

Our carrier‑neutral colocation and interconnection solutions provide the flexibility to scale infrastructure on demand, optimize network performance, and improve business continuity. By tailoring solutions to each customer’s technical and business requirements, Telehouse Canada enables organizations to accelerate digital transformation, improve resilience, and compete effectively in the global digital economy.

This is particularly important as organizations deploy increasingly data-intensive applications, including AI workloads, which require greater infrastructure density, scalable capacity, and reliable connectivity.

What are your company’s core products or services?

Telehouse Canada provides a comprehensive portfolio of colocation, interconnection, and connectivity services from Canada’s most connected carrier-neutral data centre campus in downtown Toronto. Our colocation services range from single cabinets in shared environments to customized cages and dedicated suites, supporting everything from enterprise IT to high-density AI workloads.

We deliver seamless interconnection services through seven on-site Meet-Me Rooms, enabling customers to connect directly to a wide ecosystem. Our fully managed, network-neutral environment simplifies network expansion and accelerates deployment, while diverse pathways and secure infrastructure ensure resilience, redundancy, and route diversity.

Connectivity services provide low-latency, flexible solutions designed to support cloud on-ramps, hybrid and multi-cloud architectures, and high-performance data exchange. These services together enable businesses to support latency-sensitive applications and scale efficiently within a highly interconnected digital ecosystem.

What markets do you serve?

We serve a variety of markets, including enterprises and small to medium businesses, cloud services providers, tech startups and scale-ups, financial institutions, healthcare organizations, AI developers, e-commerce platforms, content providers, government agencies, and educational institutions.

What challenges does the global digital infrastructure industry face today?

As AI adoption accelerates, the demands placed on digital infrastructure are rapidly evolving. Organizations are seeking data centre environments capable of supporting performance-intensive workloads at scale, driving the need for higher-density architectures and more advanced, efficient cooling solutions.

At the same time, the focus is shifting toward infrastructure that can balance performance with resilience and sustainability. Businesses increasingly require low-latency, energy-efficient environments that can handle growing volumes of data exchange while supporting broader environmental and operational goals.

These challenges are further compounded by constraints around power availability and energy capacity, which are reshaping how infrastructure is designed, deployed, and scaled. In parallel, the importance of sovereign infrastructure continues to grow, as organizations look to ensure data residency, meet regulatory requirements, and maintain greater control over critical digital assets.

How is your company adapting to these challenges?

Telehouse Canada is actively advancing its infrastructure to meet the growing demands of AI, high‑density deployments, and increased connectivity requirements. We are investing in infrastructure designed to support performance-intensive AI workloads, incorporating advanced technologies such as direct‑to‑chip liquid cooling to efficiently support performance‑intensive workloads. Liquid cooling is more thermally conductive than air, allowing Telehouse Canada to remove up to 80 per cent of heat directly from high-power server components. As a result, reliance on power-intensive computer room air conditioners and server fans is reduced, lowering overall energy consumption while delivering a more sustainable and efficient cooling model. Combined with our dense interconnection ecosystem and carrier-neutral environment, these upgrades help customers scale AI workloads while maintaining fast, reliable access to networks, cloud providers and end users.

To address rising power and efficiency challenges, Telehouse Canada focuses on optimizing energy use through innovative cooling approaches and a high‑efficiency facility design. Our Toronto data centres leverage renewable energy sources and integrate sustainable cooling solutions, including Enwave’s Deep Lake Water Cooling system, reducing environmental impact while supporting continued capacity growth.

Additionally, in response to growing demand for sovereign infrastructure, Telehouse Canada offers secure, Canadian‑based data centre solutions that support local data residency, regulatory compliance, and greater control over critical digital assets. This is particularly important for organizations in regulated industries and those prioritizing national data sovereignty.

Through a combination of advanced cooling technologies, energy‑efficient design, rich interconnection, and locally operated infrastructure, Telehouse Canada enables organizations to scale AI workloads, improve performance, and build resilient, future‑ready digital platforms.

What are your company’s key differentiators?

More than half (50 per cent) of all Canada’s carriers, service providers (xSPs) and content providers have a presence at Telehouse Canada’s data centres, making this interconnected hub the main gateway for Canada’s internet traffic flow, helping Canadians connect digitally with each other and the rest of the world. By connecting different networks and providers together, our data centres play an important role in facilitating Canadians’ experience of the digital world as vast, seamless, fast, and interconnected.

For Canadian consumers, data centres act as conduits for the flow of all our important data. Whether it’s the content we stream on our devices or the emails we send to colleagues, most of that content likely passes through our data centres as it moves between networks.

For businesses and public sector organizations, 151 Front Street West and its sister data centres are the primary gateway through which all commerce, content and communications data is shared both domestically and globally.

As part of a global platform of data centres, Telehouse enables customers to seamlessly extend their infrastructure internationally, supporting global expansion with consistent interconnection and colocation services.

What can we expect to see/hear from your company in the future?

Canada is known around the world for its technology leadership, including advanced research, top-tier talent and the growth of innovative companies. We’re focused on expanding the capacity of our facilities and advancing their capabilities to meet the growing demand for connectivity services.

We also remain focused on enhancing customer experience and delivering greater value— staying ahead by fostering partnerships, innovating our own operations, and aligning with global best practices.

What upcoming industry events will you be attending?

We will be attending RAISE Paris 2026 (July 8-9), ALL IN 2026 (Sept. 16-17), and AI Infra Summit (Sept. 15-17).

Do you have any recent news you would like us to highlight?

Telehouse Canada has completed a major infrastructure upgrade featuring a first-of-its-kind deployment of direct liquid cooling within an interconnection hub in Canada, enabling high-density AI workloads and cabinet densities of up to 120 kW per rack within its downtown Toronto data centre campus.

The investment strengthens Telehouse Canada’s ability to support AI-driven operations by combining advanced cooling technology with resilient, low-latency, interconnection-rich infrastructure designed for performance-intensive workloads at scale.

The upgrade also advances sustainability efforts by reducing overall energy consumption and repurposing captured heat through Enwave’s district energy system, while supporting Canada’s broader digital infrastructure and AI innovation goals.

Readers can learn more about the upgrade here.

Is there anything else you would like our readers to know about your company and capabilities?

Given our established connectivity hubs and our major infrastructure upgrade, Telehouse Canada continues to prioritize a strong commitment to supporting Canadian businesses and the economy.

As content distribution and edge computing continue to evolve, neocloud providers and AI companies are increasingly looking for power and capacity closer to carrier hotels and major connectivity hubs. While early AI factories were often built in remote locations focused on training workloads, there is now a growing trend toward inference workloads being deployed closer to end users and closer to the edge.

At the same time, rack density requirements are increasing significantly, with demand moving from roughly 40kW to 120kW per rack. While many new AI-focused data centres are being built, they still require fibre connectivity and interconnection back to carrier hotels like Telehouse Canada. The industry expected this shift toward highly interconnected AI infrastructure to happen eventually, but it appears to be occurring sooner than anticipated. As a result, facilities located adjacent to carrier hotels and internet exchanges are becoming increasingly important, and Telehouse is well positioned to support this evolution by providing both high-density infrastructure and the connectivity required to bring these environments together.

Where can our readers learn more about your company?

Readers can visit telehouse.ca to learn more about our company and follow us on our social platforms, like LinkedIn, X, Facebook, and YouTube, to stay up to date on announcements and events we’re attending.

How can our readers contact your company?

Readers can contact us via our website online, linked here.

The post Corporate Profile: A Conversation with Andy Fenton, VP of Sales and Marketing at Telehouse Canada appeared first on Data Center POST.

TL;DR

• Shift to Edge Automation: Data centers are moving away from inefficient, manually monitored Security Operations Centers (SOCs) toward automated, decentralized security located directly at the “edge” (the entrance of the data hall).
• Operational Efficiency: Implementing automated entry and screening lowers operational expenses (OpEx) by using AI to handle routine access, freeing up human staff for strategic oversight, and logging all events centrally for audit compliance.
• The Orion DataGuard™ Solution: To support this new “Lean SOC” model, Orion DataGuard™ offers a single modular unit that integrates AI/LiDAR tailgating prevention, biometric authentication, and digital-media detection to provide highly secure and frictionless access.

# # #

The traditional Security Operations Center (SOC) is under siege; not by threats, but by its own inefficiency. The 2026 SIA Security Megatrends explicitly notes that SOCs and monitoring will be disrupted and automated. For data center operators, this disruption is the key to unlocking massive ROI.

For years, physical security at data centers relied on a “manual triage” model: cameras and sensors fed data to an SOC, where security monitoring staff spent most of their time on repetitive tasks. In 2026, that model is turning obsolete. The goal now is the reduction or elimination of the SOC as we know it, moving security intelligence directly to the “edge”, the actual entrance of the data hall. This reflects a shift toward more decentralized, real-time security – positioned closer to the physical infrastructure.

Turning Security into a Value Chain

Data centers are moving away from security staff and implementing fewer SOC’s, towards remotely monitored/controlled automated security systems; especially for security entrance control, visitor management, and surveillance. By implementing self-service, automated security screening, data centers can finally bridge the gap between high security and low OpEx. This is where the ROI becomes undeniable.

Moving authentication and screening to an automated portal can:

• Eliminate Tier 1 Bottlenecks: AI handles the “noise” of daily access.
• Optimize Staffing: Shift human capital from “watching doors” to high-level strategic oversight.
• Ensure Compliance: Every event is centrally logged via AI based event management systems, such as Orion Infinity™ for audit-ready reporting.

Introducing Orion DataGuard™: The Modular Solution for 2026

At Orion Entrance Control, Inc., we’ve listened to the pressure data center developers face; balancing many conflicting priorities like real estate and energy costs with the absolute need for data center server security. Orion DataGuard™ is our answer.

DataGuard™ integrates AI/LiDAR-based tailgating prevention through Orion DoorGuard™, identity and biometric authentication, digital-media detection and secure asset management into a single modular unit. It is a highly secure, ADA-accessible screening environment that satisfies stringent AHJ fire codes while providing a frictionless experience. Whether for new builds or retrofits, DataGuard™ enables a “Lean SOC” model by automating all the processes involved in providing secure entry. By partnering with best-of-breed technology providers, Orion ensures that DataGuard™ fits into your existing tech stack while preparing you for a future where security is automated, autonomous, and incredibly efficient.

# # #

About the Author

Mike McGovern, VP Data Center Security at Orion Entrance Control, Inc.

For over 26 years Mike has helped develop, design and integrate pedestrian entrance control systems (security revolving doors, turnstiles, speed gates, security portals) working with end users, architects, integrators and security engineer consultants. His portfolio includes architectural optical turnstile, revolving door, and portals systems at many of the USA’s most iconic and recognizable buildings and campuses. He is a frequent speaker and panel moderator on integrated security and data center security topics at leading data center trade shows and seminars. McGovern is an active member of the ASIS SAEC (Security Architecture and Engineering Community) Steering Committee and the SIA (Security Industry Association) Data Center Advisory Panel. As VP Business Development at Orion Entrance Control, Inc., he leads their data center security entrance control business and product development strategy.

About Orion Entrance Control, Inc.

Orion Entrance Control, Inc. is an American technology company that specializes in speed lanes, optical turnstiles, full height turnstiles, security tailgating solutions, and occupancy sensor solutions. Serving high-rise buildings, government and corporate offices, and other facilities, Orion delivers secure, visually appealing, and easily integrated entry solutions. With decades of experience working with security integrators and architectural planners, Orion provides the expertise and tools to streamline lobby security projects while delivering actionable data and metrics that enhance operational efficiency and safety. For more information, please visit www.orioneci.com or follow us on LinkedIn.

The post Data Center Security Enhanced with DataGuard™: The Disruption and Automation of Security Operations Centers (SOCs) appeared first on Data Center POST.

TL;DR

• Massive Market Growth: The global data center refrigerant market is projected to reach $1.62 billion by 2035 (up from $633 million in 2025), driven by the rapid expansion of artificial intelligence, cloud platforms, and high-density computing.
• Shift to Sustainable Solutions: Regulatory pressures and sustainability goals are forcing a transition to low-emission alternatives, with Hydrofluoroolefins (HFOs) capturing a dominant 67.7% market share in 2025 due to their low global warming potential and high cooling efficiency.
• U.S. Market Leadership: The United States accounted for a massive 79% of the market in 2025, heavily supported by continuous investments in hyperscale expansions and strict energy efficiency mandates.

# # #

The global data center refrigerant market was valued at USD 633 million in 2025 and is estimated to grow at a CAGR of 9.6% to reach USD 1.62 billion by 2035, according to recent report by Global Market Insights Inc.

The market is undergoing a major transformation driven by the rapid expansion of digital infrastructure, including hyperscale facilities, cloud platforms, and accelerating global data consumption. Increasing computational workloads from artificial intelligence, big data analytics, and high-density computing environments are significantly raising thermal management requirements across modern data centers. As a result, demand for advanced refrigerants that ensure efficient heat control, operational stability, and optimized energy performance is increasing steadily.

The industry is also experiencing a structural shift away from traditional high-global-warming-potential refrigerants toward low-emission and energy-efficient alternatives. Regulatory pressure from global climate agreements and regional environmental policies is accelerating the adoption of next-generation refrigerants. Operators are increasingly focusing on reducing emissions, improving leakage control, and enhancing lifecycle management practices to comply with evolving sustainability requirements. This transition is further reinforced by rising emphasis on lowering operating costs and improving energy efficiency across large-scale digital infrastructure.

The data center refrigerant market from HFOs segment held a 67.7% share in 2025 and is projected to grow at a CAGR of 10.2% through 2035. HFO refrigerants continue to dominate due to tightening environmental regulations and the global push toward decarbonization. Their significantly lower global warming potential, combined with strong cooling efficiency, makes them a preferred solution for next-generation data center cooling systems. The ongoing phase-out of high-GWP refrigerants is further accelerating their adoption across the industry.

The hyperscale segment accounted for 40% share in 2025 and is expected to grow at a CAGR of 10.4% from 2026 to 2035. This segment leads the market due to extremely high computing densities and large-scale infrastructure operated by major cloud and technology providers. These facilities generate substantial heat loads from artificial intelligence workloads, cloud computing, and advanced digital services, requiring highly efficient refrigerant-based cooling systems to maintain continuous operations and system stability.

U.S. data center refrigerant market held a 79% share in 2025, generating USD 176.3 million. Growth in the country is strongly supported by rapid hyperscale data center expansion driven by artificial intelligence, cloud computing, and data-intensive applications. Continuous investments in new facilities by leading technology firms are increasing demand for advanced cooling systems. Rising adoption of high-density server infrastructure is further intensifying thermal loads, strengthening the need for high-performance refrigerants. In addition, strict energy efficiency regulations and sustainability mandates are encouraging the shift toward low-GWP refrigerant solutions and modern thermal management technologies.

Key companies operating in the global data center refrigerant industry include Daikin Industries, Linde plc, Honeywell, Chemours, AGC, Dongyue, Zhejiang Juhua, Arkea and Sinochem. Companies in the market are actively focusing on developing low-global warming potential refrigerants that align with tightening environmental regulations and sustainability targets.

A major strategy involves increasing investment in research and development to enhance refrigerant efficiency, thermal stability, and compatibility with high-density computing environments. Market players are also strengthening partnerships with data center operators and cooling system manufacturers to integrate advanced refrigerant solutions into next-generation infrastructure. Expansion of production capabilities in high-demand regions is being prioritized to ensure supply chain stability and faster delivery. Additionally, firms are emphasizing compliance-driven innovation, improving leakage detection systems, and enhancing lifecycle management solutions.

The post Data Center Refrigerant Market to Surpass USD 1.62 Billion by 2035 appeared first on Data Center POST.

TL;DR

• The Hyperscale Mismatch: Large wholesale data centers in Miami cater to massive enterprise volumes, forcing small and medium-sized businesses (SMBs) into expensive minimum commitments that far exceed their actual requirements.
• The Cost of Unused Capacity: Under these enterprise models, SMBs end up paying for stranded power and empty rack space, making professional colocation financially unviable and leaving them vulnerable to risks like housing critical systems in office server closets during hurricane season.
• Fractional Colocation Solutions: Instead of accepting preset sales minimums, businesses can utilize right-sized options like quarter-rack (10U) or half-rack (20U) setups, which provide the same enterprise-grade facility standards but scale power and space commitments to exact hardware loads.
• Accessible Resilience: Local, right-sized colocation allows South Florida businesses to affordably access vital infrastructure, such as robust hurricane resilience, physical security, and low-latency connectivity, while preserving the flexibility to scale up seamlessly as demand grows.

# # #

The South Florida business landscape is booming, cementing its status as a critical gateway for national and international commerce.

As a growing business or enterprise IT leader in this region, migrating your core infrastructure to a Miami data center can solve on-premises space limitations, remove cooling headaches, and provide real physical security against extreme weather events. But when evaluating local options, smaller businesses quickly hit a commercial wall with the industry’s massive wholesale giants.

Search for Miami colocation today, and the biggest names dominate the results: global hyperscale and wholesale providers with massive campuses, enterprise sales teams, and pricing models built for customers leasing entire cages and data halls. What those results don’t show is the fine print that matters most to a small or medium-sized business: the minimum commitment.

For a South Florida business that needs to house a handful of servers, a storage appliance, and a firewall, that floor can make professional colocation an impossible infrastructure decision because of the budget problem. The business ends up paying for empty cabinet space and stranded power capacity, or it walks away from colocation entirely and accepts the risk of an office server closet during hurricane season.

This guide explains why hyperscale minimums fail smaller deployments, what a right-sized data center engagement looks like with quarter-rack, half-rack, and full-rack options, and why power should be provisioned to your actual requirements rather than the maximum a cabinet can support.

The Enterprise-Focused Structure of the Miami Colocation Market

Miami is one of the most strategically important data center markets in the Western Hemisphere. It is the primary connectivity gateway between North America, Latin America, and the Caribbean, home to dense subsea cable landings and carrier ecosystems that route an enormous share of intercontinental traffic. That strategic value has attracted the largest global colocation operators, and their presence shapes how Miami colocation is sold.

Hyperscale and wholesale providers optimize for volume. Their ideal customer leases hundreds of kilowatts across cages and suites, signs multi-year enterprise agreements, and consumes interconnection services at scale. Everything about their operation, from sales process to contract structure, reflects that buyer. The smallest unit they are willing to sell is set accordingly: one full cabinet, a 5 kW committed power floor, and terms designed for enterprise procurement.

None of that is wrong as a business strategy, it just means that the most visible Miami data center options are built around enterprise buyers, without consideration for smaller businesses that make up a significant part of the South Florida market.

What Large Provider Minimums Mean for Small Businesses

At the largest Miami colocation providers, retail colocation often starts at one full rack paired with a 5 kW minimum power commitment. For a multinational enterprise, that may be a small line item. For an SMB, it can define the entire economics of the deal.

Consider a typical small business deployment: 8 to 12 rack units of equipment drawing 1 to 2 kW under normal load. Under a hyperscale minimum, that business may pay for:

• A 42U or 45U cabinet where most of the space sits empty
• 5 kW of committed power when actual draw peaks closer to 2 kW
• Cross-connect fees, remote hands rates, and setup charges scaled to enterprise budgets
• Multi-year terms with escalators built around enterprise renewal cycles

Power is usually where the mismatch hurts most: if a business draws 1.8 kW but commits to 5 kW, the unused 3.2 kW still shows up on the bill every month, and over a multi-year term, that stranded capacity can become a serious cost. It can seriously raise the bill, and push smaller businesses into awkward infrastructure choices, from delaying hardware refreshes to keeping critical systems in office server rooms or moving steady workloads into cloud environments that may cost more over time.

What Right-Sized Data Center Colocation Looks Like

Right-sized Miami data center colocation starts with the actual deployment, not a preset sales minimum. The amount of space and power should reflect what the business is really putting in the rack.

Smaller environments may only need a quarter rack or half rack, while larger or denser deployments may require a full rack with a higher committed power level. The footprint follows the equipment, not the other way around.

Power should follow the same logic as space, with the commitment based on the actual hardware load rather than the maximum theoretical capacity of the rack. A deployment drawing 1.5 kW should be provisioned differently from a high-density setup that needs 8 kW. The hardware profile sets the number, with enough headroom to grow without overbuying from day one.

Half-Rack and Quarter-Rack Miami Colocation Options Exist for Small Businesses

Many SMBs don’t realize these options exist because many Miami colocation options are marketed around full racks and enterprise deployments. In reality, though, smaller footprints like quarter racks and half racks can be a much better fit for businesses that only need a modest amount of equipment.

A quarter rack provides roughly 10U of secure, individually lockable space, which is often enough for a basic production environment with a few servers, storage, a firewall, and a switch. The smaller footprint doesn’t mean a lower facility standard. Quarter rack deployments can still use core data center infrastructure like redundant power paths, N+1 cooling, physical security, fire suppression, and 24/7 monitoring.

A half rack, around 20U, gives growing businesses more room for virtualization clusters, shared storage, network equipment, and backup infrastructure. It is also useful for companies consolidating servers from multiple offices or closets into one professional environment. In both cases, power should scale to the deployment itself: a half rack drawing 2.5 kW commits to 2.5 kW, not an arbitrary floor.

Scaling From Quarter Rack to Full Rack Without Overcommitting

Businesses that genuinely need a full cabinet still benefit from the right-sizing principle. The goal is not to stay small, but to match the deployment to current requirements.

Because the model is fractional by design, scaling is built in. A company can start in a quarter rack, move into a half rack, and later expand to a full rack within the same facility. That kind of path helps avoid buying more space and power than the business is ready to use.

Why Location Is Still Part of the Equation

Right-sizing addresses the economics. Location addresses operations. For businesses across Miami-Dade, Broward, and Palm Beach counties, local colocation can offer advantages that are difficult to replicate from another market.

For South Florida businesses, hurricane resilience is often the biggest driver behind moving infrastructure out of an office server room and into a professional facility. Proximity makes everyday operations easier, too. When hardware needs to be installed, replaced, or physically inspected, engineers can drive to the facility instead of coordinating logistics across the country. Miami’s position as a connectivity hub also benefits organizations serving Latin America and the Caribbean, serving a core need: lower latency.

The Bottom Line: Miami Colocation Should Fit Your Business

Miami colocation doesn’t have to start with a full rack, an excessive power commitment, or more infrastructure than the business currently needs. The better approach is to look at the actual hardware first, then choose the footprint and power profile that fits it. That sounds obvious, but it’s often missing from the way colocation is sold. In many cases, the better fit already exists; businesses simply have to know where to look.

# # #

About the Author

Michael Zrihen is the Senior Director of Marketing & Internal Operations Manager at Volico Data Centers

The post Why Hyperscale Data Center Pricing Fails Small Businesses Seeking Miami Colocation appeared first on Data Center POST.

TL;DR

• AI Workloads Overwhelming Traditional Cooling: AI’s demand for concentrated compute power is driving rack densities from 5-10 kW to beyond 50-100 kW, creating extreme heat that traditional, water-intensive air cooling systems can no longer handle.
• Infrastructure Must Evolve: Thermal management, power delivery, and resource efficiency are now co-dependent, requiring developers to treat data center infrastructure as an integrated system rather than sacrificing sustainability for performance.
• The Immersion Cooling Advantage: Fully submerging IT components in thermally conductive fluid provides superior, uniform heat removal, enabling higher rack densities while significantly lowering both water and energy consumption compared to traditional methods.
• Fluid Selection is the Linchpin: The ultimate success and sustainability of immersion cooling rely on selecting the right engineered fluids, which dictate heat transfer efficiency, hardware lifespan, and overall environmental impact.

# # #

The Current State of Data Center Cooling

In 2026, data centers are caught between surging demand and tightening constraints. AI workloads are driving the need for higher compute density, while developers face limits related to power availability, water usage, and permitting timelines. Policymakers and communities are scrutinizing new projects, slowing approvals due to various concerns ranging from grid strain to land and water use. Meanwhile, hyperscalers and enterprise operators must meet aggressive sustainability targets even as their infrastructure footprints expand.

These pressures are exposing fundamental structural challenges in how data centers are built. Traditional air cooling with water-intensive systems cannot scale to meet AI’s mounting demands. Developers must now tackle multiple variables at once: more compute per square foot, lower energy and water intensity, and alignment with evolving regulatory frameworks. Meeting these challenges requires rethinking infrastructure as an integrated system where performance, efficiency, and sustainability are engineered together, not sacrificed for each other.

AI as the Catalyst for Convergence

Artificial intelligence has fundamentally changed infrastructure development strategies. Unlike traditional workloads, AI training and inference require highly concentrated compute power, delivered through GPUs and accelerators operating at significantly higher thermal design power (TDP). Rack densities that once averaged 5–10 kW are now pushing beyond 50–100 kW, with next-generation deployments expected to drive densities even higher.

The implications for power infrastructure are already appearing. According to Dean Nelson, founder and chairman of Infrastructure Masons, the industry is building as much data center capacity in the next three years as was built in the past thirty. The infrastructure required to support that pace of growth must align with these operational realities. After all, the cascading effects are significant. Higher compute density drives more heat, which demands more cooling. Cooling systems consume more power and, in many cases, more water, putting compounding pressure on resources already under strain. Power availability is becoming a gating factor, with utilities struggling to keep pace with the speed and scale of data center growth.

As a result, AI performance is determined by the infrastructure that supports it, not just by model architecture or silicon. Thermal management, power delivery, and resource efficiency are now co-dependent variables. As such, decisions from the chip to the facility cooling plant must be evaluated through a single, system-wide lens that takes these factors into account.

Immersion Cooling as an Infrastructure Strategy

Cooling is now a critical lever for enhancing performance and efficiency. As TDPs rise across the technology stack, liquid cooling is becoming a requirement at leading AI facilities. Immersion cooling, in particular, is uniquely positioned to address the three constraints underpinning modern data center development: heat management, water use, and power consumption.

Compared to traditional air cooling and direct-to-chip liquid cooling, immersion cooling delivers significant improvements in thermal management efficiency. By fully submerging IT components in a thermally conductive fluid, heat is removed directly and uniformly at the source, eliminating the limitations of air cooling and the complexity of localized cold plates and plumbing. Direct-to-chip cooling targets high-heat components like GPUs and CPUs but still relies on air or secondary systems to cool the rest of the hardware. Immersion cooling, by contrast, manages the entire system within a single thermal environment.

The Operational Benefits of Immersion Cooling

The superior heat capacity and thermal conductivity of immersion fluids enable consistent, high-efficiency heat transfer across all components, even at extreme power densities. This helps data centers support significantly higher rack densities without complex airflow management, extensive piping networks, or oversized mechanical cooling systems. Benefits such as better performance, longer hardware lifespan, and greater operational reliability are proving critical for high-density AI workloads.

Immersion cooling also reduces reliance on air handling and evaporative systems, significantly lowering water consumption. It mitigates overall energy demand for cooling as well, easing pressure on constrained power infrastructure. The net effect is not only thermal efficiency, but a more balanced and sustainable resource profile that regulators, investors, and communities increasingly prioritize.

Fluid Selection: The Linchpin of Data Center Infrastructure Sustainability

However, realizing these benefits depends on fluid selection. Fluid qualities like thermal stability, heat transfer efficiency, material compatibility, and environmental profile all determine system performance and lifecycle outcomes. The right fluid enhances heat transfer, extends system longevity, and reduces maintenance requirements. It also shapes sustainability metrics, from lifecycle emissions to end-of-life handling.

As the demands of AI workloads intensify, these decisions carry increasing weight. Now, when selecting a cooling technology, data center developers are making long-term infrastructure commitments that will determine performance, scalability, and environmental impact for years to come. Immersion cooling, when paired with thoughtfully engineered fluids and integrated system architectures, can help developers align all three qualities to improve operational efficiency while addressing common challenges.

# # #

About Robert Schuetzle

Robert Schuetzle is President and CEO of Infinium, a technology company transforming how the world powers, moves, and computes. Backed by strategic investors including Amazon, MHI, NextEra Energy, and AP Ventures, Infinium is best known for its leadership in ultra-low carbon electrofuels and is now expanding that expertise into digital infrastructure through Infinium Edge™, its platform focused on energy efficiency and thermal management for power- and heat-intensive systems such as data centres.

The post Fueling Performance at the Edge: Why Cooling, Power, and Sustainability Are Intertwined in the AI Data Center Era appeared first on Data Center POST.

TL;DR

• A Labor Agreement Built Around People: Quantica Infrastructure and the Southeastern Montana Building and Construction Trades Council have signed a Memorandum of Understanding (MOU) to guide union construction for the Big Sky Campus, a large-scale data center and energy project.
• Workforce Development as a Core Commitment: The agreement establishes a framework focused on apprenticeship opportunities, training pathways, workforce development, and qualified local labor.
• Long-Term Benefits for Montana: The partnership is designed to support job creation, career development, and economic activity while helping deliver critical infrastructure for the state’s future.

# # #

As discussions around data centers often focus on power, technology, and infrastructure, one critical component can sometimes get overlooked: the people who build them.

For the Big Sky Campus, Quantica Infrastructure, LLC (“Quantica”) and the Southeastern Montana Building and Construction Trades Council (SEMTBCTC) are working to make workforce development part of the project from the beginning.

The two organizations recently announced a labor Memorandum of Understanding (MOU) that will help guide construction of the Big Sky Campus, a large-scale data center and energy project being developed by Big Sky Digital Infrastructure (BSDI), a Quantica company.

The agreement establishes a cooperative framework focused on workforce development, apprenticeship and training pathways, project stability, and qualified local labor. It also creates opportunities to expand participation in the skilled trades while supporting construction of the infrastructure needed to power the next generation of digital growth.

For Montana’s construction workforce, that means more than jobs on a single project. It means opportunities for training, career development, and long-term participation in an industry that continues to grow across the country.

“The Big Sky Campus represents a significant opportunity for Montana’s skilled trades workforce,” said Clint A. McCulloch, President of SEMTBCTC and Business Manager of UA Plumbers and Pipefitters Local Union 30. “We are pleased to work alongside Quantica to help create family-supporting careers, expand apprenticeship opportunities, and ensure Montana workers have a role in building the infrastructure that will support the state’s future economy.”

The partnership also reflects a broader goal shared by both organizations: ensuring that major infrastructure investment creates lasting local benefits.

Through apprenticeship programs, training opportunities, and a focus on qualified local labor, the agreement is intended to help build a highly skilled workforce while creating pathways into the construction trades for future generations of Montana workers.

“Major infrastructure projects like the Big Sky Campus have the potential to deliver meaningful economic benefits when they prioritize skilled labor and long-term workforce development,” said Riley McCauley, Vice President of SEMTBCTC and Representative of Western States Regional Council of Carpenters. “We see a real opportunity to support a strong, skilled workforce in Montana by partnering with Quantica on apprenticeship programs, training pipelines, and local hiring efforts, helping create durable career pathways while supporting responsible infrastructure development in the region.”

For Quantica, the agreement aligns with the company’s vision for how large-scale infrastructure projects should be developed.

“Our vision for the Big Sky Campus extends beyond digital infrastructure,” said John Chesser, Chief Executive Officer of Quantica. “We are committed to creating high-quality jobs, supporting workforce development, and building lasting opportunities for Montanans. This agreement reflects our belief that strong partnerships with Montana workers can help deliver world-class infrastructure safely, efficiently, and with meaningful benefits for our communities.”

The Big Sky Campus is expected to bring substantial investment, long-term employment, and economic activity to the region. BSDI has also emphasized its commitment to funding the new generation and infrastructure required for the campus without increasing electricity costs for Montana ratepayers.

As development moves forward, the agreement provides a framework for ensuring that the benefits of the project extend beyond the infrastructure itself and into the workforce, businesses, and communities that will help build it.

The post Union Partnership Powers the Big Sky Campus: Quantica and Montana Trades Council Forge Labor Framework appeared first on Data Center POST.

Originally posted on Mod42.

TL;DR

• Modular is a manufacturing methodology, not a shipping container.
• Mod42 manufactures complete data center buildings off-site using volumetric modular construction.
• Factory-built modules enable parallel construction, reducing deployment timelines by up to 50%.
• Purpose-built modular facilities deliver the same performance, quality, and serviceability as traditional data centers.
• The modular approach provides greater flexibility to adapt to evolving AI workloads, cooling technologies, and power requirements.
• Mod42 helps organizations deploy AI infrastructure faster while maintaining long-term scalability and future readiness.

# # #

A common misconception in the digital infrastructure industry is equating modular data centers with shipping containers filled with servers. In reality, modular construction is a manufacturing methodology that shifts work into a controlled factory environment to reduce risk and accelerate schedules. Moving beyond simple containerized solutions, Mod42 utilizes volumetric modular construction to manufacture entire building sections off-site. These purpose-built modules incorporate the exact same structural, electrical, mechanical, and cooling systems found in traditional facilities, resulting in a fully operational, enterprise-grade data center rather than a temporary deployment.

The rapid deployment speed of this methodology comes from parallel manufacturing rather than the physical shape of a container. While site preparations, foundations, and utilities are being completed on location, the actual data center is built simultaneously in the factory. By overlapping these critical phases instead of executing them sequentially, organizations can bypass traditional construction bottlenecks and bring complete, factory-built AI capacity online in under seven months.

Beyond speed, this volumetric approach provides the critical adaptability required by rapidly evolving artificial intelligence workloads. Because these facilities are engineered as actual buildings rather than restricted transportation packages, they offer the necessary flexibility to accommodate soaring rack densities, changing cooling strategies, and future hardware upgrades. This ensures that operators can efficiently scale their operations, easily add modules as demand grows, and future-proof their long-term infrastructure investments.

To continue reading, please click here.

The post A Modular Data Center Is Not a Container: Why Mod42 Manufactures Complete AI Data Centers appeared first on Data Center POST.

TL;DR

• Tournament proceeds will fund workforce development initiatives focused on digital infrastructure education and career access.
• Supported programs include scholarships, mentorship, internships, educational tours, conference participation, and industry engagement.
• The event aims to unite executives, investors, and technology leaders around a shared commitment to developing future industry talent.
• Sponsorship opportunities enable organizations to support educational initiatives while participating in a high-level networking experience.

# # #

As the digital infrastructure industry continues to expand to support AI, cloud, and connectivity demands, workforce development remains a critical priority. To help address this need, the Nomad Futurist Foundation has announced the inaugural Nomad Ho’ohui Invitational, an invitation-only charity golf tournament that will take place January 15-16, 2027, on Oahu, Hawaii, immediately preceding PTC’27.

Hosted at the Kapolei Golf Club, the two-day Ryder Cup-style competition is designed to bring together executives, investors, technology leaders, and innovators from across the digital infrastructure ecosystem. Beyond the competition itself, the event will feature networking receptions, a gala dinner, and opportunities for industry participants to build relationships while supporting a larger philanthropic mission.

Unlike many industry gatherings that focus primarily on business development, the Nomad Ho’ohui Invitational is centered on investing in the future workforce. Proceeds from the event will directly benefit the Nomad Futurist Foundation’s initiatives to increase awareness of careers in digital infrastructure and provide students with meaningful educational and professional opportunities. Funding will support scholarships, educational resources, data center tours, mentorship programs, internships, conference participation, and direct engagement with industry professionals.

The name “Hoʻohui,” a Hawaiian term meaning to join, unite, or connect, reflects the event’s broader purpose of bringing together the industry to create lasting impact. By aligning with PTC’27, one of the sector’s premier annual gatherings, the Invitational provides participants with an opportunity to strengthen professional relationships while contributing to programs that help prepare the next generation of talent.

Industry leaders have emphasized that the future of digital infrastructure depends not only on continued investment in technology, but also on sustained investment in people. As AI infrastructure, cloud computing, telecommunications, and data center development accelerate worldwide, organizations increasingly recognize that cultivating skilled professionals will be essential to supporting long-term growth and innovation.

The Nomad Futurist Foundation is inviting organizations to participate as founding partners through sponsorship opportunities that include player participation, event branding, guest access, and advisory committee involvement. Every contribution supports programs intended to expand career pathways and increase access for students and underserved communities interested in pursuing opportunities within digital infrastructure.

As the industry prepares to gather in Hawaii for PTC’27, the Nomad Ho’ohui Invitational offers a unique opportunity to combine networking, collaboration, and philanthropy in support of workforce development. By investing in education, mentorship, and career access today, participants can help strengthen the talent pipeline that will power tomorrow’s digital economy. To learn more about the Nomad Ho’ohui Invitational, please click here.

The post Nomad Ho’ohui Invitational Combines Industry Collaboration and Workforce Development Ahead of PTC’27 appeared first on Data Center POST.

TL;DR

• From Policy to Infrastructure: AI governance is no longer just a high-level policy issue; because AI is increasingly embedded directly into operational platforms and automation pathways, it must be managed as a core infrastructure strategy.
• Visibility Precedes Governance: AI capabilities are quietly accumulating within everyday SaaS apps and monitoring tools, meaning organizations must first establish a clear inventory of where AI exists, what data it processes, and what decisions it influences to prevent systems from gaining unchecked operational authority.
• Transparency as a Competitive Edge: With customers increasingly demanding to know how AI interacts with their environments and data, service providers that proactively communicate their governance frameworks, vendor controls, and safeguards will differentiate themselves as trusted operational partners.

# # #

For many organizations, AI governance still sounds like a policy discussion. In reality, it is rapidly becoming an infrastructure discussion.

AI systems are increasingly embedded into the platforms organizations use to manage infrastructure, monitor environments, automate workflows, support users, and analyze operational data. In some cases, AI systems directly influence operational changes across environments. As adoption accelerates, organizations are facing a new challenge: governing the AI models themselves, and also the infrastructure access, automation pathways, and operational influence those systems inherit inside the environment.

For MSPs and SMBs, this shift carries significant implications. AI capabilities are quietly appearing inside monitoring systems, endpoint operations, automation engines, service management platforms, and collaboration tools. In many environments, organizations may already be relying on AI-assisted capabilities without formally recognizing them as AI systems.

Visibility Must Come Before Governance

One of the biggest AI governance mistakes organizations can make is assuming AI systems are easy to identify. AI is increasingly embedded across SaaS applications and operational tooling in ways that may bypass centralized oversight. A workflow platform introduces automatic AI-assisted script creation, a monitoring tool adds AI-generated recommendations, or a collaboration platform that begins summarizing operational data automatically. In each case, the AI recommendation itself acts as an anchor: the first number, option or plan presented becomes the reference point around which human judgement is adjusted. Over time, organizations accumulate a growing number of AI-enabled capabilities without a centralized understanding of what systems exist, what they can access, or what operational decisions they can influence.

The first step toward effective governance is visibility. Organizations need a clear operational inventory of where AI exists across the environment, what infrastructure it connects to, what data it processes, what decisions it influences, and what level of operational authority it possesses. Without centralized tracking and accountability, businesses risk creating environments where AI systems quietly gain operational influence without sufficient oversight.

AI Is Changing the Meaning of Change Management

Traditional change management processes were designed around human-driven activity. AI introduces a more dynamic operating model, where systems can generate recommendations, accelerate workflows, and potentially influence operational decisions at machine speed. Even when humans remain involved, the pace and scale of operational activity can change dramatically.

Organizations should begin treating AI-enabled systems with the same operational discipline they apply to automation platforms and critical infrastructure tooling. That includes understanding whether AI systems can modify infrastructure directly, trigger automation workflows, or influence operational decisions without clear approval boundaries. It also means ensuring rollback mechanisms exist, maintaining strong audit logging, and validating whether human oversight remains in place before operational changes occur.

The operational failures associated with AI may not resemble traditional outages. In many cases, organizations are more likely to encounter subtle configuration drift, unintended automation behavior, or operational decisions made from incomplete context. These issues can spread quietly across environments before teams fully understand what changed or why.

Transparency Is Becoming a Competitive Requirement

Organizations are also facing growing pressure from customers who increasingly want transparency into how AI systems interact with their environments and data. Customers want to understand whether AI systems process operational information, whether external vendors are involved, whether customer data contributes to model training, and what safeguards exist if systems behave unexpectedly.

For MSPs and service providers, this creates both risk and opportunity. Organizations that proactively communicate governance practices, including oversight models, logging standards, vendor controls, and data handling policies, will increasingly differentiate themselves as trusted operational partners rather than reactive adopters of emerging technology.

Operational Maturity Will Define AI Governance Success

The organizations making the most progress today are not necessarily the ones building the most complex governance frameworks. They are the ones building practical governance models grounded in operational reality. The goal is to establish enough visibility, accountability, and operational control to scale AI safely while maintaining resilience and customer trust.

Organizations that move early on operational AI governance will be better positioned to scale automation safely, respond to customer concerns confidently, and adapt as regulatory expectations continue evolving.

Ultimately, AI governance is also a conversation about infrastructure access, operational influence, automation boundaries, and accountability. As AI systems become more deeply integrated into operational environments, governance will increasingly determine whether organizations can scale AI safely and confidently over the long term. That makes AI governance more than a compliance issue – it should be a core infrastructure strategy.

# # #

About the Author

Nicole Reineke is a technology executive and AI strategist currently serving as a Distinguished Product Manager and Director of AI Strategy at N-able. She previously served as Senior Vice President of Innovation at Iron Mountain and Senior Distinguished Engineer at Dell Technologies, and brings more than 25 years of experience leading high-tech ventures and driving enterprise innovation. In addition to holding dozens of granted patents, she teaches AI and innovation at Georgetown University and co-authors publications focused on breakthrough success and applied AI strategy.

The post Why AI Governance Must Become an Infrastructure Strategy appeared first on Data Center POST.

TL;DR

• The Rise of Bare Metal: Bare metal is now the dominant model for renting GPU clusters, sitting directly between traditional colocation and public cloud by allowing customers to rent entire, provider-owned servers without an intervening virtualization layer.
• A Distinct New Asset Class: GPUs have separated from data center real estate to become a standalone asset class because they possess all four necessary traits: identifiable assets, observable pricing, active resale markets, and lenders willing to accept them as collateral.
• The Portability Advantage: Unlike traditional real estate, GPU servers are portable and serial-numbered, meaning lenders can physically repossess and re-market the hardware in the event of a loan default.

# # #

Bare metal has become the dominant way GPU clusters are rented. The customer’s software runs directly on the hardware, with no hypervisor, no virtualization layer, and no managed services in between.

Since 2023, it has pulled the server into a separate asset class of its own from Data Centers. GPU clusters are financed in the capital markets, where it is now priced, traded, and borrowed against.

Where bare metal sits between colocation and cloud

Bare metal sits between the two models the industry already knows. Colocation has the tenant owning the servers and renting space, power, and cooling. Public cloud has the provider owning everything and selling virtual slices with managed services on top.

Bare metal is in the middle. The provider owns the servers, and the customer rents them whole, by the hour or on multi-year contracts. Neoclouds such as CoreWeave, Lambda, and Crusoe, cloud providers built specifically around GPU capacity, run their entire business on this model.

And now, traditional colocation operators are looking to do it too. They’re adding bare metal and GPU-as-a-service lines to capture the same demand, taking on hardware ownership their business model never required before.

The hardware became an asset class

An asset class needs four things: identifiable assets, observable pricing, a resale market, and lenders willing to take the asset as collateral. GPU servers now have all four.

The GPU can leave the building. When a data center loan defaults, the lender ends up owning real estate in one location. But when a GPU loan defaults, the lender holds serial-numbered servers that can be repossessed and re-marketed. That portability is what allows the hardware to be financed separately from the building it sits in.

In March 2026, CoreWeave closed an $8.5 billion facility rated A3 by Moody’s, the first investment-grade rating ever assigned to GPU-backed debt, supported in part by Meta contracts worth at least $19 billion.

The pricing of that debt tells the story faster than the headlines. CoreWeave’s earliest GPU-backed borrowing reportedly priced in the low teens. The floating-rate portion of the March 2026 facility priced at 2.25 percentage points over SOFR, the overnight benchmark rate. Lenders re-rated the asset from exotic to ordinary in under three years.

The data center industry spent three decades learning to finance buildings. Now, racks inside carry capital markets of their own.

# # #

About the Author

Bernie Margulies is the founder and CEO of American Compute, which helps GPU buyers get access to better GPU financing, and structures residual value insurance for lenders/lessors.

The post The New GPU Asset Class appeared first on Data Center POST.

TL;DR

• Grain Management announced plans to combine Ritter Communications and Great Plains Communications under a new organization and brand, Rightfiber.
• The combined company will serve more than 400 communities across 20 states with a 28,000-mile fiber network connecting 300,000 homes and businesses.
• Rightfiber aims to pair expanded scale and investment capacity with the local service, community focus, and customer relationships that have defined both organizations.

# # #

The broadband industry is often described in terms of speed, scale, and coverage. But the reason this deal matters now is simpler: Communities, businesses, and service providers all need reliable fiber networks that can grow with demand, not just keep up with it.

That is the context behind Grain Management’s announced combination of Ritter Communications and Great Plains Communications into a new organization called Rightfiber. The move brings together two established regional providers with long operating histories and a shared focus on fiber infrastructure, customer service, and community investment.

Two Companies, One Direction

Ritter Communications and Great Plains Communications each built their reputations in different parts of the country, but their business philosophies are closely aligned. Both companies have spent decades expanding fiber access, serving local markets, and investing in the kind of infrastructure that supports everyday life as well as business growth.

According to the announcement, the new company will be led by Heath Simpson as CEO, with Todd Foje serving as Executive Chairman. The leadership structure brings together executives from both organizations as the combined company moves forward under the Rightfiber brand.

What Rightfiber Brings

The new organization will serve more than 400 communities across 20 states and connect 300,000 homes and businesses through a 28,000-mile regional fiber network. Those numbers matter because they show how much the company’s footprint expands with this combination.

But the story is not just about size. It is also about what scale can make possible. Rightfiber is expected to have more resources to support network expansion, continued investment, and growth through both organic buildout and acquisitions. For customers, the promise is that local service and personalized support remain part of the model even as the company gets bigger.

Why Local Still Matters

In telecom, bigger can sometimes feel distant. This announcement makes a point of saying the opposite: The combined company will continue to operate with a strong local focus, preserve customer relationships, and keep investing in the communities it serves.

That matters because fiber is not just a technical upgrade. It is the backbone for remote work, education, health care, business operations, and the digital tools people now expect to use every day. A stronger network can change what is possible for a town, a business district, or a region that needs better connectivity to compete.

The Takeaway

Rightfiber brings together two established fiber providers under a single organization designed to expand scale, resources, and investment while maintaining a local operating approach.

The transaction still needs customary approvals and regulatory review, but the larger message is already clear: The future of broadband will be shaped by companies that can combine reach, investment, and local trust. That is what makes this combination worth watching.

For full details, read the press release here.

The post One Fiber Story, Bigger Reach: Strategic Combination Forms Rightfiber appeared first on Data Center POST.

TL;DR

• AI demand is accelerating infrastructure investment and driving major changes in facility design, deployment strategies, and capacity planning.
• Power availability has emerged as one of the most significant constraints on future data center growth.
• Investors continue to support digital infrastructure expansion, but financing structures are becoming more sophisticated and risk-conscious.
• Workforce development remains essential as operators seek the talent needed to support long-term industry growth.

# # #

The rapid expansion of artificial intelligence is creating unprecedented demand for digital infrastructure, forcing the industry to rethink everything from power procurement and cooling strategies to financing models and workforce development. Those themes dominated conversations at Datacloud Global Congress 2026, held June 1-4 in Cannes, France, where industry leaders examined how to scale capacity while navigating increasingly complex operational, regulatory, and economic challenges.

Across four days of sessions, networking events, and the  Datacloud Awards, one theme remained constant: AI is accelerating infrastructure demand at a pace that is forcing the industry to rethink how future capacity will be planned and deployed. As AI workloads continue to grow, the industry is facing a reality where traditional development approaches are no longer sufficient. Operators are being asked to deploy larger facilities faster, secure power in constrained markets, attract investment, and build infrastructure that can remain relevant despite rapidly evolving technology requirements.

AI Demand Is Reshaping Infrastructure Strategy

Artificial intelligence remained the dominant topic throughout the event as operators, technology providers, and investors assessed how the market is evolving to support next-generation compute requirements. Conversations focused on how the growing share of AI workloads is changing infrastructure priorities, from facility design and cooling architectures to deployment models and regional investment strategies.

During the keynote panel examining how the data centre ecosystem is keeping pace with AI demand, Olivier Micheli, CEO & President of DATA4, joined fellow industry leaders to explore how rapidly changing workload requirements are influencing infrastructure strategies. Panelists discussed the projected growth of AI workloads, the increasing importance of inference-driven deployments, and how neocloud providers are emerging as a disruptive force with infrastructure requirements that differ from traditional hyperscale and enterprise environments. The conversation also examined how enterprise AI adoption is reshaping demand patterns and what Europe must do to remain competitive as global investment in AI infrastructure accelerates.

Many sessions also explored Europe’s efforts to strengthen its position in the global AI market. Discussions surrounding AI gigafactories, sovereign cloud initiatives, and large-scale infrastructure investments reflected growing momentum across the region as governments and private-sector organizations work to expand digital capacity.

Power Has Become the Industry’s Most Valuable Resource

While AI is driving demand, power availability is increasingly determining where and how projects move forward. Multiple sessions focused on the challenge of securing reliable energy in markets facing grid constraints, permitting delays, and growing competition for available capacity.

One of the recurring themes was the need to diversify power strategies. Industry leaders explored a range of solutions including battery energy storage systems, behind-the-meter generation, renewable energy partnerships, microgrids, and emerging nuclear technologies. Discussions also addressed the growing importance of flexibility, as operators seek to build facilities capable of adapting to changing energy markets over the coming decades.

The industry’s focus is shifting from simply obtaining power to managing energy risk. As data centers become larger and more energy-intensive, operators are evaluating how to improve resiliency while balancing sustainability goals, operating costs, and long-term growth plans.

New Financing Models Are Emerging to Support AI Expansion

The scale of AI-driven infrastructure investment is creating demand for more sophisticated funding approaches. Traditional financing structures are being supplemented by joint ventures, alternative debt vehicles, portfolio financing strategies, and new forms of risk sharing between investors, operators, and customers.

During the keynote panel, “How are DC Projects Being Funded?”, Alex Hernandez, CEO of PowerBridge, joined industry leaders to examine how capital is being deployed across traditional and AI-focused data center developments. The discussion explored the quality of customer offtake agreements, the long-term profitability of AI investments, how neocloud providers are securing financing, and the increasing use of joint venture structures to fund large-scale infrastructure projects.

Several investment-focused sessions also explored consolidation across the sector, off-balance-sheet financing models, and evolving approaches to underwriting AI infrastructure projects. While capital remains available, investors are placing greater emphasis on execution capability, power availability, and the long-term viability of infrastructure assets.

Risk Allocation Continues to Evolve

As projects grow larger and more complex, operators and customers are reassessing how risk is allocated throughout the development lifecycle. Contracting structures that worked for traditional data center projects are being reexamined as AI facilities require significantly larger investments and longer planning horizons.

Atif Ansar, Executive Chairman and Co-Founder of Foresight, participated in a panel examining how contracting models are evolving as AI gigafactories increase the scale and complexity of infrastructure projects. Discussions focused on balancing risk between tenants and operators, including how power failures, supply chain disruptions, early termination rights, and warranty and indemnity insurance are being addressed in a new generation of AI-focused agreements.

The conversation reflected a broader trend across the industry. As infrastructure projects become larger and more expensive, contractual frameworks are evolving to provide greater certainty for both developers and customers while supporting continued investment in new capacity.

Developing the Workforce Needed for Future Growth

The industry’s expansion is creating significant demand for skilled workers across engineering, construction, operations, sustainability, and energy management disciplines. While AI is driving much of the growth, workforce development remains a critical factor in determining how quickly new capacity can be delivered.

During a panel focused on creating a resilient workforce pipeline, Susanna Kass, Senior Advisor to the Board and Operating Partner at Digital Gravity Infrastructure Partners, joined industry leaders to discuss how workforce development programs can help address ongoing labor shortages across the sector. The conversation explored the evolving skill sets needed to support data center growth and strategies for expanding the talent pool across engineering, construction, operations, and energy disciplines.

Workforce challenges continue to affect projects across multiple regions, particularly as operators pursue larger campuses and increasingly complex infrastructure deployments. Industry leaders emphasized that long-term growth will depend on building sustainable talent pipelines capable of supporting the next generation of digital infrastructure.

Preparing for the Next Phase of Growth

Throughout the conference, discussions repeatedly returned to a common challenge: how to scale infrastructure quickly without sacrificing resiliency, efficiency, or long-term value. AI may be the catalyst driving today’s investment cycle, but the decisions being made around power, financing, workforce development, and facility design will influence the industry for years to come.

The path forward will require collaboration across the entire digital infrastructure ecosystem. As operators pursue larger campuses, higher-density deployments, and more ambitious growth targets, success will increasingly depend on the industry’s ability to align capital, energy, technology, and talent at unprecedented scale.

To learn more about Datacloud Global Congress and future Datacloud events, visit the official Datacloud website.

The post AI Infrastructure Growth Tests the Limits of Power, Capital and Scale  appeared first on Data Center POST.

TL;DR

• The Midwest AI Boom: A massive 100-Terabit connectivity corridor is being built in the Midwest, which is rapidly becoming a strategic hub for AI infrastructure due to its availability of land, power, and buildable space.
• Symphonic Collaboration: Building such large-scale network infrastructure requires immense trust and deep coordination across service providers, equipment vendors, and operational teams to achieve shared goals.
• Supply Chain and Efficiency: High global demand is creating supply chain bottlenecks for critical networking gear, pushing the industry to focus on smarter transport models that prioritize power and spectral efficiency over pure speed.
• Future Constraints: Moving forward, massive fiber and multi-gigawatt power requirements will be the industry’s biggest hurdles, alongside community permitting resistance. Furthermore, the extreme cost of downtime makes network diversity and resilience an absolute business imperative.

# # #

In Episode 66 of the NEDAS Live! Podcast, host Ilissa Miller sits down with Priscilla Favors of Aureon, Chris Crowe of T3 Broadband, Kurt Raaflaub of Nokia, and Jeff Sanders of Midco for a timely conversation about one of the industry’s most coordinated transport builds to date. Centered on the theme “Delivering the AI Era: Building a 100TB Midwest Connectivity Corridor,” the discussion explores why the Midwest is becoming increasingly important to AI infrastructure, and what it takes to bring large-scale network projects like this to life.

As Miller notes, the conversation goes beyond data centers and compute to spotlight the transport layer that makes AI possible. The panel makes clear that none of the AI buildout works without the underlying network infrastructure, or the “nervous system,” as one speaker describes it, that moves massive amounts of data between facilities, regions, and end users.

Why the Midwest, Why Now

One of the most compelling themes in the episode is the Midwest’s rising role in the digital infrastructure landscape. Priscilla Favors explains that the region is benefiting from the growing need for land, power, and buildable space, especially as hyperscale and AI deployments expand beyond traditional coastal markets. What once may have been viewed as secondary territory is now proving to be strategically positioned for the next wave of infrastructure growth.

Chris Crowe adds that the industry is in a period of unprecedented adoption, with demand rising faster than many expected. As AI accelerates, so does the need for connectivity that can support new workloads, new architectures, and new expectations around speed and resilience.

Collaboration at Scale

A major takeaway from the conversation is that projects of this magnitude cannot be executed by one company alone. Jeff Sanders describes the deployment as a symphony, with each participant contributing specialized expertise while staying tightly aligned on the shared goal. From design and fulfillment to construction and interconnection, the build required deep coordination across service providers, equipment vendors, and operational teams.

Favors reinforces that point by emphasizing how critical trust was throughout the process. The work moved quickly, required executive approval at unusual hours, and demanded collaboration across technical, financial, and leadership levels. In her view, the project succeeded because the partners were willing to move together, adapt quickly, and rely on one another’s expertise.

Supply Chain Pressure

The episode also highlights a growing operational challenge across the industry: supply chain constraints. Crowe explains that major and minor components alike can become bottlenecks when global demand spikes. From cabling and shelves to amplifiers and other critical gear, large projects are increasingly impacted by component availability and timing.

Kurt Raaflaub expands on the larger market backdrop, pointing to dramatic increases in data center capital expenditures and the pressure that AI and cloud providers are placing on the broader ecosystem. He notes that regional service providers and integrators are essential to keeping projects moving, especially as the market decentralizes and more connectivity is needed between secondary markets.

Networks Need to Evolve

If the first phase of digital infrastructure was about more bandwidth, the next phase is about efficiency, flexibility, and smarter transport. Raaflaub explains that optical networking is moving beyond pure speed and toward spectral efficiency, lower power consumption, and better cost per bit. In his view, the infrastructure that worked a decade ago is no longer enough for today’s AI-driven requirements.

Sanders echoes that shift, pointing to solutions such as 800-gig wavelength services, managed optical fiber network services, and more adaptable transport models. The era of simply “throwing bandwidth at it” is giving way to a more nuanced approach built around customer needs, geography, diversity, and timing.

Looking Ahead

When asked what the industry may underestimate over the next five years, the panel repeatedly returns to two words: fiber and power. Sanders predicts that intra-building fiber requirements inside AI campuses may become even more massive than many expect, while power demands could escalate toward multi-gigawatt clusters.

Favors adds that land and power availability will remain significant constraints, while also warning that downtime is becoming far too expensive to ignore. She shares that even a single day of outage can cost millions, reinforcing why diversity and resilience are now business imperatives, not just technical preferences.

A Race Defined by Partnerships

The episode closes on a bigger industry truth: AI infrastructure is not just a technical race, but a partnership race. The panel agrees that success will depend on collaboration across suppliers, integrators, service providers, and customers, as well as the ability to educate stakeholders and adapt to changing political, social, and operational realities. Sanders, in particular, warns that community resistance and permitting hurdles may become defining constraints on where AI data centers are built.

As Miller concludes, the industry is at the tip of a much larger transformation, and this Midwest deployment may be just the beginning. With AI demand accelerating and infrastructure expectations changing rapidly, the message from Episode 66 is clear: the next era will be built by those who can move fast, think collaboratively, and deliver at scale.

To continue the conversation, listen to the full episode here.

The post Delivering the AI Era: Building a 100-Terabit Midwest Connectivity Corridor appeared first on Data Center POST.