TL;DR

• Floating data center site selection is a system design decision rather than a simple real estate exercise, requiring location, engineering, and regulation to be managed in parallel to solve land, power, and cooling constraints.
• Developers must evaluate six core factors: energy availability and scalability, marine environment metocean conditions, operational logistics, digital connectivity redundancy, regulatory permitting pathways, and physical/cyber security.
• Long-term viability requires managing key operational trade-offs over multi-decade lifespans, balancing near-shore accessibility against offshore expansion potential and grid reliance versus dedicated offshore generation.
• Early integration of marine risk assessment alongside classification societies like ABS avoids costly late-stage redesigns and provides clarity across complex regulatory jurisdictions

# # #

When it comes to floating data centers, it really is about ‘location, location, location’. Site selection influences engineering, regulation, operations and long-term resilience, which is why it pays to get this right from the outset, writes Frank Genco, ABS Vice President of Strategic Account Program 

When it comes to floating data centers (FDC), the question isn’t why offshore. It is increasingly where. The whys are clear: the rapid growth of data centers is already testing grid capacity, water resources and the tolerance of local communities living alongside these facilities. By siting data centers offshore, however, many of those constraints can be resolved, with developers able to access new energy sources, abundant seawater for cooling and locations unconstrained by land availability.

But where? That’s a bigger question, and the answer will influence the regulatory pathway, engineering design, reliability, cost, maintenance, security and uptime for the next 30 years.

More than a real estate decision

This isn’t a site selection checkbox exercise, however. It’s a system design decision, because location, engineering and regulation are closely interconnected. A change to one often affects the other two. A site that appears attractive from an operational perspective may introduce regulatory complexity. Engineering innovations may suddenly make previously unsuitable locations viable. These three strands (location, regulation and engineering design) need to be managed in parallel because it is at their intersection where the right solution in the right location will be found.

 ABS brings decades of experience supporting complex marine and offshore assets, from floating LNG and ultra-deepwater oil production units to floating wind. That experience can help engineers understand how location choices influence requirements and regulatory pathways long before detailed design begins.

Every answer creates another question

Site selection involves identifying and testing a wide range of competing and interdependent issues all at the same time. An apparently straightforward question about available infrastructure, such as natural gas pipeline proximity, may lead to broader engineering considerations. Does a location provide sufficient access to electrical power today for a standalone FDC, and can it support future expansion or augmentation? Would dedicated offshore generation offer greater resilience and public approval than relying solely on the grid? Could nearby renewable energy or future small modular nuclear reactors reshape the long-term energy strategy?

It’s the same for the cooling requirements of the computing and power generation facilities. Engineers need to understand seasonal temperature variation, water movement, corrosion rates and whether engineering trade-offs could expand the range of viable operating conditions. Improving cooling system design may significantly increase the number of potential locations but how would this innovation impact timelines?

Regulation presents another layer of complexity. The applicable framework depends on numerous factors, including proximity to shore, whether the facility is permanently moored or mobile, the pipeline requirements, the presence of subsea power or communications infrastructure and which authorities have jurisdiction. Depending on the location, developers may need to engage with federal, state, local and port authorities, alongside maritime regulators and environmental agencies.

Understanding these interactions early can avoid costly redesigns later in the project.

The major considerations

Every project will have unique requirements, but several key factors will consistently shape site selection. These include:

1. Energy. Developers must determine where power will come from, whether existing grid infrastructure is sufficient, or whether dedicated offshore generation offers greater resilience and flexibility. Equally important is understanding how future demand growth can be accommodated without major redesign.
2. Marine environment. Long-term metocean conditions, extreme weather exposure, wave climate, currents, corrosion potential and marine growth all influence structural design, maintenance schedules and operating costs.
3. Operational logistics. How easily can personnel access the facility? Which nearby ports can support maintenance activities? How resilient are local supply chains? These questions directly affect availability and lifetime operating costs.
4. Digital connectivity. Fiber availability, cable redundancy and network latency all influence performance. Building resilience into communications infrastructure from the outset can significantly reduce operational risk over the facility’s lifetime.
5. Regulatory landscape. Understanding permitting pathways, environmental approval processes, certification requirements and jurisdictional responsibilities early in the project can help avoid delays and provide greater certainty during development. ABS works with developers at the regulatory interface, helping to map the applicable frameworks and identify the right agencies before those conversations become constraints on engineering or schedule.
6. Security. Physical protection, cybersecurity, vessel traffic management, supply chain resilience and geopolitical risk all become increasingly important as critical digital infrastructure moves offshore.

Thinking beyond first deployment

A floating data center will likely have an operational lifespan measured in decades, making resilience and future-proofing central to site selection. The maritime and offshore industries are used to thinking long-term and building offshore structures designed to withstand rare and extreme events, but these floating data factories will be packed with cutting-edge computing technology and serving markets that are evolving at speed, which may require additional forethought. Developers should consider whether today’s optimal location will still be viable if computing demand doubles, cooling technologies evolve or additional modules are needed.

Managing trade-offs

There is rarely a perfect location. Sheltered near-shore waters may simplify operations initially but limit future expansion. Offshore sites may provide greater access to renewable energy but introduce more demanding environmental conditions that require additional engineering and impact costs. Grid-connected facilities may reduce upfront investment but dedicated offshore generation could offer greater resilience and scalability.

These trade-offs are not unique to floating data centers. Offshore developers are used to navigating competing priorities and understand that the key is to evaluate these trade-offs systematically and simultaneously so that technical, commercial and regulatory implications are considered together rather than in isolation. In this way, a holistic, multi-disciplinary and multi-agency approach can be taken that minimizes the risks of unwelcome surprises further down the road.

Involving marine risk assessment from the beginning

Ultimately, site selection is a risk management exercise. The maritime and offshore industries have spent decades developing sophisticated approaches to evaluating operational, environmental and engineering risk. Many of the challenges floating data centers will face have already been addressed across offshore energy, shipping and marine infrastructure. Applying these proven methodologies during the earliest stages of development enables better-informed decisions before major investments are committed. ABS works across all three of these dimensions simultaneously, combining marine risk assessment, regulatory engagement and engineering insight to help developers match the right asset to the right location.

As floating data centers move from concept to commercial reality, the most successful will be those that view site selection as integral to engineering and commercial outcomes, not simply as a location for the “real” work to take place. By working with ABS to integrate engineering, regulation and marine risk assessment from the outset, developers can identify locations that work today and are also resilient, scalable and commercially viable for decades to come.

 # # #

About the Author

Frank Genco is the Vice President of Strategic Accounts at the American Bureau of Shipping (ABS). He created and heads the Strategic Account Management (SAM) Program. His division is tasked with aligning ABS’s global enterprise resources across maritime, energy, digital infrastructure, cyber security, and risk management to service their highest-value global clients. Frank specializes in critical infrastructure, strategic corporate partnerships, and connecting top-tier technology companies with maritime and digital risk expertise. He is currently building out specialized teams, including expanding executive relationships with Silicon Valley technology firms to bridge the gap between AI/digital infrastructure and maritime compliance.

The post Site Selection: The Most Important Decision You’ll Make Before Construction Begins appeared first on Data Center POST.

TL;DR

• Hose selection must be driven by complete system operating conditions (including specific coolant chemistry, thermal cycling, working pressure, permeation, and cleanliness requirements) rather than choosing a material name directly from a catalog page.
• Hose material performance impacts long-term reliability, influencing key operational factors beyond flexibility, such as pressure drops, gradual fluid permeation, and loop contamination.
• Common hose materials involve distinct design trade-offs, with options like EPDM, silicone, PTFE, PFA, and metal offering different balances of chemical compatibility, flexibility, mechanical strength, cost, and permeation resistance.
• Hoses and fittings should be integrated early into overall rack, manifold, and cooling system architecture rather than left as an afterthought during final installation

# # #

By Xiang Wu, IT & Automation Director, CJan Fluid Technology Co., Ltd.

As liquid cooling moves from specialized deployments into mainstream AI and high-performance computing infrastructure, more attention is being paid to pumps, cold plates, cooling distribution units and control systems.

The hose is easier to overlook.

Yet in a direct-to-chip cooling loop, the hose is part of the fluid path that operates continuously while carrying coolant between components that may need to be serviced independently. Its material affects more than flexibility. It can influence pressure capability, chemical compatibility, permeation, cleanliness, bending behavior and service life.

For that reason, choosing a liquid cooling hose should start with the operating conditions of the cooling loop, not with a material name on a catalog page.

Start With the Coolant, Not the Hose

The first question in hose selection should be simple:

What fluid will actually be inside the hose?

Water, water-glycol mixtures and specially formulated cooling fluids do not place exactly the same demands on hose materials. Even when two coolants appear similar, differences in additives, inhibitors, concentration and operating temperature can affect material performance.

This is particularly important for systems that are expected to operate continuously for years.

A material that appears compatible during a short laboratory test may behave differently after thousands of hours of exposure. Swelling, hardening, extraction of low-molecular-weight components and changes in mechanical properties may develop gradually.

For this reason, coolant compatibility should be evaluated using the actual fluid formulation whenever possible.

Material selection should consider:

• Coolant chemistry
• Glycol concentration, where applicable
• Corrosion inhibitors
• Biocides and other additives
• Operating temperature
• Expected service life
• Cleanliness requirements

The hose should be qualified as part of the actual cooling system rather than treated as an isolated component.

Temperature Is More Than a Maximum Rating

Hose specifications often list a maximum temperature. That number is useful, but it does not tell the whole story.

A cooling loop may experience a relatively stable operating temperature most of the time while still going through repeated thermal cycles. Startup, shutdown, maintenance and changes in computing load can all introduce temperature changes.

Repeated thermal cycling can affect:

• Elastomer flexibility
• Reinforcement
• Fitting connections
• Sealing performance
• Long-term material stability

A hose material therefore needs to be evaluated against the complete temperature profile rather than a single maximum value.

For example, EPDM is widely used in water-based cooling applications because of its resistance to water and glycol-based fluids. Silicone offers excellent flexibility and broad temperature capability in many applications. Fluoropolymer materials such as PTFE and PFA can be considered when chemical resistance, cleanliness or low extractables are particularly important.

The point is not that one material is universally better than another.

The right material depends on what the cooling loop requires.

Pressure and Flow Still Matter

Liquid cooling systems are designed around controlled flow, but that does not mean hose pressure requirements can be ignored.

The hose should be evaluated against:

• Normal operating pressure
• Maximum working pressure
• Pressure fluctuations
• Pump characteristics
• Startup and shutdown conditions
• Potential pressure spikes

Internal diameter also matters.

A hose that is too small can increase pressure drop and require additional pump head. A hose that is unnecessarily large may create packaging and routing problems.

The engineering objective is therefore to balance flow requirements with pressure capability, flexibility and available space.

The hose should also be evaluated as a complete assembly. The pressure rating of the tube alone does not necessarily represent the pressure capability of the hose, fitting and connection together.

Flexibility Has a Practical Value

One reason flexible hose continues to have a role in liquid-cooled data centers is serviceability.

Servers, cold plates, manifolds and rack-level cooling components may need to be removed or repositioned during maintenance. A rigid connection can make this difficult.

But flexibility should not be confused with unlimited bending.

Every hose has a minimum bend radius. Repeatedly forcing a hose tighter than its recommended radius can place unnecessary stress on the tube, reinforcement and fittings.

Good routing should avoid:

• Sharp bends near fittings
• Twisting along the hose axis
• Continuous tensile loading
• Contact with sharp edges
• Uncontrolled movement caused by vibration

A slightly longer hose with proper routing can sometimes be more reliable than a shorter hose installed under constant mechanical stress.

This is one reason hose routing should be considered during rack and manifold design rather than left to the final installation stage.

Permeation Deserves More Attention

Leakage is an obvious concern in a liquid cooling system. Permeation is less visible.

Some hose materials allow molecules to gradually pass through the hose wall. The rate depends on the material, fluid, temperature, pressure and wall construction.

For a short-duration application, this may have little practical significance. For a cooling loop expected to operate continuously over a long period, it can become part of the system reliability calculation.

Potential consequences include:

• Gradual coolant loss
• Changes in coolant concentration
• Moisture or gas transfer
• Changes in fluid condition
• Increased maintenance requirements

This does not mean that a low-permeation material is automatically the correct choice.

It means that permeation should be considered whenever coolant stability and long service intervals are important design objectives.

Cleanliness Is a System Requirement

Data center liquid cooling is not simply a plumbing application.

The coolant passes through components such as cold plates, heat exchangers, pumps and manifolds. Contamination introduced by one component can potentially travel throughout the loop.

For systems with tighter cleanliness requirements, engineers may need to consider:

• Extractables
• Particles
• Ionic contamination
• Internal surface condition
• Manufacturing cleanliness
• Cleaning and flushing procedures

This is one area where material selection and manufacturing quality are closely connected.

A material with appropriate chemical resistance may still be unsuitable if the finished hose assembly introduces unacceptable contaminants into the cooling loop.

The Fitting Is Part of the Decision

Selecting the hose and selecting the fitting should not be treated as two independent decisions.

The connection needs to accommodate the hose construction, pressure, temperature, installation method and expected movement.

Particular attention should be paid to the area immediately behind the fitting.

This is often where bending, vibration and installation forces concentrate. If the hose is forced into a sharp bend immediately after the connection, the fitting may effectively become the mechanical anchor point for the entire assembly.

A good installation distributes mechanical loads instead of transferring them directly to the connection.

A Practical Material Selection Approach

There is no single hose material that should be specified for every liquid cooling system.

A practical selection process can begin with six questions:

1. What is the coolant?
   Identify the actual fluid and additives.

2. What are the temperature conditions?
   Consider both steady-state temperature and thermal cycling.

3. What pressure and flow are required?
   Check working pressure, pressure fluctuations, internal diameter and pressure drop.

4. How much movement is expected?
   Determine whether the hose is static, occasionally moved or continuously flexed.

5. What cleanliness and permeation requirements apply?
   These factors may influence the choice between elastomeric and fluoropolymer constructions.

6. How will the hose be installed and maintained?
   Consider bend radius, fitting accessibility, replacement and service clearance.

Only after these questions are answered should the material be selected.

A Simple Comparison

This table should be treated as a starting point rather than a specification guide. Actual material performance depends on the formulation, hose construction, manufacturing process and operating environment.

The Best Hose Is the One That Fits the Whole System

Liquid cooling projects sometimes approach hose selection from opposite directions.

One team may prioritize flexibility.

Another may prioritize pressure.

A third may focus on chemical compatibility or cleanliness.

All of these considerations are legitimate. The challenge is that optimizing one characteristic without considering the others can create a different reliability problem.

A highly flexible hose may not provide the desired pressure or permeation performance. A chemically resistant material may require a different reinforcement or connection method. A mechanically robust assembly may be harder to route in a dense rack environment.

The selection process therefore needs to look at the complete operating envelope.

For high-density AI infrastructure, this becomes increasingly important as liquid cooling systems become more complex and maintenance access becomes more valuable.

Design the Hose Into the Cooling System

The hose should not be the last component selected after the rest of the cooling architecture has already been fixed.

It should be considered alongside the cold plate, manifold, CDU, pump, quick-disconnects and rack layout.

When material, construction, routing and connections are evaluated together, engineers can make better trade-offs between reliability, serviceability and cost.

The objective is not simply to find a hose that can carry coolant.

The objective is to build a fluid path that remains predictable throughout the operating life of the data center.

As direct-to-chip cooling continues to expand, that distinction will become increasingly important.

# # #

About the Author

Xiang Wu is the IT & Automation Director at CJan Fluid Technology Co., Ltd., where he focuses on industrial automation, digitalization, and engineering applications for advanced fluid systems. His work brings together automation technology and practical engineering requirements across demanding industrial applications, including data center liquid cooling and semiconductor equipment. Xiang has a particular interest in how material selection, system design, and component reliability affect the performance of fluid-handling systems. He works closely with engineering teams and customers to develop practical solutions for complex cooling and fluid-transfer challenges.

The post How to Select Hose Materials for Direct-to-Chip Liquid Cooling Systems appeared first on Data Center POST.

TL;DR

• Datalec Precision Installations (DPI) has welcomed 12 new apprentices as part of its second apprenticeship cohort.
• The programme combines hands-on workplace experience with industry-recognised qualifications and structured career development.
• Apprentices will gain experience across technical and operational disciplines supporting mission-critical environments throughout the UK.
• DPI’s apprenticeship programme supports the development of skilled talent as demand for data centre and critical infrastructure expertise continues to grow.
• Edmundson Electrical and Safety Signs are supporting the new cohort with professional tools, PPE and uniforms.

###

As investment in cloud computing, digital services and AI continues to drive data centre growth, the need for skilled professionals capable of supporting critical digital infrastructure is growing alongside it.

Datalec Precision Installations (DPI) is helping build that talent pipeline by welcoming 12 new apprentices to its workforce as part of the second cohort of its apprenticeship programme. The initiative provides young people with a direct pathway into the data centre industry through a combination of hands-on experience, formal training and structured career development.

The new apprentices will begin their careers across a range of technical and operational disciplines, working alongside experienced DPI professionals on projects supporting mission-critical environments throughout the UK.

As data centres play an increasingly important role in the UK’s digital economy, developing the workforce needed to build and support that infrastructure has become an industry-wide priority. DPI’s apprenticeship programme is designed to help address that skills challenge while creating meaningful, long-term career opportunities for young people.

“Apprenticeships provide an outstanding route into our industry, enabling individuals to earn while they learn and develop practical skills within a real working environment,” said Danny Turley, Learning & Development Manager at DPI. “As demand for digital infrastructure continues to grow, investing in future talent has never been more important.”

Through the programme, apprentices receive mentorship, structured on-the-job learning and exposure to a wide range of disciplines across the business. This approach provides participants with technical expertise, professional qualifications and transferable skills that can form the foundation for long-term careers within the sector.

DPI’s industry partners are also supporting this year’s cohort from day one. Edmundson Electrical has supplied professional tools for the apprentices, while Safety Signs has provided PPE and uniforms.

With continued investment in emerging talent, DPI is helping develop the next generation of skilled professionals equipped to build, install and support the critical digital infrastructure underpinning modern business and society.

The programme currently has an apprenticeship retention/completion rate of 100%, with seven business departments participating, demonstrating DPI’s commitment to integrating apprenticeship opportunities across its operations.

Read more in the press release here.

The post Building the Next Generation of Data Centre Talent: Datalec Welcomes New Apprentice Cohort appeared first on Data Center POST.

TL;DR

• ZincFive launched the ZincFive BMS, a next-generation battery management system built specifically for its nickel-zinc (NiZn) chemistry rather than adapted from lithium-ion or lead-acid platforms.
• The system ships standard with every ZincFive BC Series UPS Battery Cabinet and uses a streamlined two-tier architecture (Controller plus Tray Monitor) that simplifies wiring, cuts installation time, and removes the need for third-party network switches.
• In BC 2 AI cabinets, the BMS enables AI pulse intelligence, allowing the system to respond to the high-frequency power swings created by AI training and inference workloads.
• Built-in encryption, authentication, and access controls address growing operational technology (OT) cybersecurity requirements, while a redesigned interface gives operators fleet-wide visibility, remote configuration, and firmware updates across multi-site deployments.

# # #

AI is changing the shape of electricity demand inside the data center, not just the size of it. Instead of a steady draw, GPU clusters can ramp power up and down within milliseconds to seconds, and research from the Electric Power Research Institute (EPRI) has found these swings can reach 40% or more of total load in that window. Backup power systems built for slow, predictable demand were never designed to track that kind of movement, which is why the conversation in the industry has shifted from “how much battery capacity do we have” to “how well can that battery communicate and respond.”

That shift is putting a spotlight on a part of the UPS stack that used to be an afterthought: the battery management system, or BMS. It’s also the reason ZincFive‘s recent launch of a purpose-built BMS is worth a closer look, not just as a product update, but as a signal of where the industry is headed.

The BMS Was Never Meant to Be an Afterthought

A BMS handles the unglamorous work of watching over individual cells and modules, tracking voltage, temperature, and state of charge, and flagging problems before they become failures. For years, many battery suppliers adapted BMS platforms originally built for lithium-ion or lead-acid chemistries and applied them broadly across product lines.

That approach works reasonably well when battery behavior is uniform across chemistries. It works less well when a chemistry has different electrical characteristics, cycle patterns, or failure modes that a borrowed platform was never tuned to interpret. ZincFive’s new BMS takes the opposite approach: It was designed and assembled in the U.S. specifically around nickel-zinc chemistry, using a two-tier hardware architecture, the ZincFive Controller and ZincFive Tray Monitor, built to deliver high-resolution telemetry across an entire battery cabinet rather than retrofitting a platform meant for a different chemistry.

“The ZincFive BMS reflects everything we’ve learned from managing and deploying nickel-zinc batteries at scale,” said Tod Higinbotham, CEO of ZincFive. “By bringing development in-house and assembling the BMS in the U.S., we have greater control over the platform and how it evolves. As AI reshapes data center power requirements, this level of intelligence and control becomes increasingly critical.”

Millisecond Response Requires Millisecond Visibility

Handling AI-driven power transients isn’t only about raw battery chemistry. It also depends on how quickly a system can sense a load change and coordinate a response across every cabinet in a fleet. That’s the specific problem ZincFive’s AI pulse intelligence capability, built into the BMS for BC 2 AI UPS Battery Cabinets, is designed to address: helping the system respond directly to the high-frequency power pulses associated with AI training and inference.

High-resolution, cabinet-level telemetry, tracking individual battery voltage and temperature in near real time, gives operators the visibility needed to confirm that backup power is actually behaving the way it’s supposed to under stress, not just during a scheduled test. Simplifying the physical architecture matters too. ZincFive’s design daisy-chains cabinets via dual Ethernet ports and eliminates the need for a separate data-aggregation layer, an approach aimed at reducing installation time and long-term maintenance burden, a practical benefit that often gets overlooked in conversations focused purely on power density.

Cybersecurity Has Quietly Become a Power Infrastructure Issue

Battery management systems are networked, which means they’re now part of the data center’s operational technology attack surface. Recent industry research has found that a majority of organizations have BMS devices affected by known, exploitable vulnerabilities, a statistic that would have seemed out of place in a power conversation a few years ago. ZincFive built encryption, authentication, hardened communications, and access controls into the new BMS from the ground up, an approach meant to meet increasingly strict OT security expectations while giving operators audit-ready visibility into their systems.

The Takeaway for Data Center Operators

The lesson emerging across the industry is straightforward: battery performance and battery intelligence are no longer separate conversations. ZincFive’s launch, which also adds a redesigned interface for fleet-wide monitoring, remote configuration, and firmware updates across multi-site deployments, is one clear example of that shift in practice. As data centers prepare for an AI-driven power profile that looks nothing like the one they were originally built for, the companies rethinking how batteries are monitored and managed, not just how they’re built, are the ones setting the pace.

To read the full press release, visit here.

The post Why Battery Intelligence Is Becoming as Important as Battery Power appeared first on Data Center POST.

TL;DR:

• A new pan-European data center design and engineering group has entered the market.
• Hyceon, formed from the merger of TTSP HWP and MCA Architects, brings together more than 270 professionals and experience across over 10 GW of data center capacity under one unified organization.
• Its launch reflects a broader shift in how large-scale digital infrastructure projects are being delivered across Europe, at a moment when demand for AI-ready facilities is outpacing traditional project models.

# # #

Frankfurt and Dublin-based Hyceon officially launched this week as the unified entity of two long-established firms, TTSP HWP and MCA Architects. The move completes a merger first announced in December 2025 and marks the arrival of a new pan-European group focused specifically on data center design, engineering, and advisory services.

The timing is notable. Europe’s data center sector is in the middle of a build-out driven by cloud computing and artificial intelligence, and the scale of that demand is reshaping what clients expect from their design and engineering partners. Hyceon’s launch is one of the clearer signals yet that the market is consolidating around firms capable of managing a project from concept through completion, rather than handing it between specialists at each stage.

What the Combined Group Looks Like

Under CEO Alexander Hauser, Hyceon operates from six offices across Germany and Ireland, Frankfurt, Dublin, Stuttgart, Cologne, Berlin, and Düsseldorf, with additional delivery offices in Budapest and Kuala Lumpur supporting projects across the continent. The group brings together more than 270 professionals working across architecture, engineering, master planning, and project management.

Combined, the two founding firms have worked on data center projects representing more than 10 GW of capacity and have carried several hundred schemes through the permitting process. That track record spans five years of prior collaboration between TTSP HWP and MCA Architects, during which their joint data center work already represented over a gigawatt of capacity for hyperscale and colocation clients. The new structure formalizes what had effectively been a working partnership into a single organization.

Why One Point of Accountability Matters Now

Large data center projects typically involve a long sequence of distinct phases: site selection, feasibility studies, permitting, master planning, architecture, engineering, and eventually construction oversight. When each phase is managed by a different firm, coordination gaps can slow projects down or create costly rework later.

“Our clients are building at a speed and scale Europe has not seen before, and they are looking for partners who can carry a whole project rather than a piece of it,” said Hauser. “Hyceon gives them one team and one point of accountability, from site selection to handover.” That structure is particularly relevant as European regulation grows more specific. Germany’s Energy Efficiency Act, for example, introduces tightening PUE and heat-reuse requirements for facilities commissioned from July 2026 onward, standards that are far easier to meet when efficiency planning is built into a project from its earliest stages rather than layered on afterward.

A Signal for What’s Ahead

Hyceon’s launch adds a new, sizable player to a European data center market that continues to attract hyperscalers, cloud providers, colocation operators, and infrastructure investors. As Managing Director Brian Murphy noted, the group intends to expand its footprint alongside its clients: “As our clients expand into new and existing regions, they need partners who can provide the same high standard of service wherever they go.”

For an industry navigating rapid AI-driven growth alongside increasingly detailed regulatory requirements, Hyceon’s formation illustrates a broader pattern worth watching: The firms best positioned to support Europe’s next wave of data center development may be the ones built to manage the entire project, start to finish, rather than a single piece of it.

To read the full launch announcement, please click here. To learn more about what Hyceon brings to the European market, visit hyceon.com.

The post A New Name Enters Europe’s Data Center Market appeared first on Data Center POST.

TL;DR

• Toptana Technologies has been named a winner of the 2026 Broadband Communities Tribal Lands Award.
• The award recognizes Toptana’s work developing new fiber and subsea connectivity infrastructure in Washington State.
• Its network is being built to serve the Quinault Indian Nation while giving the community a direct stake in the infrastructure.
• The project will connect the Washington coast with Seattle, Hillsboro and international markets across the Pacific.

# # #

For the Quinault Indian Nation, building better connectivity starts with a simple question: who gets to decide what that infrastructure looks like and who benefits from it?

Toptana Technologies, the Indigenous-owned telecommunications infrastructure company owned by the Quinault Indian Nation, is taking a different approach. Rather than viewing broadband infrastructure simply as something delivered to a community, Toptana is building and owning infrastructure that can serve the community while creating opportunities beyond it.

That work has earned Toptana the 2026 Broadband Communities Tribal Lands Award, which recognizes efforts to bring modern telecommunications services to tribal lands while respecting sovereignty and addressing gaps in connectivity.

“From the beginning, Toptana has been about creating greater opportunity for the Quinault Indian Nation through connectivity while maintaining our sovereignty and protecting the resources that are central to our people and way of life,” said Tyson Johnston, 1st Council person of the QIN and Ex Officio Board Member. “This recognition is meaningful because it reflects both what we are building and why we are building it to create lasting benefits for our community and future generations.”

The infrastructure behind that effort is already taking shape.

Toptana has completed an approximately 200-mile fiber route running north to south between Seattle and Hillsboro. The company is also advancing an approximately 90-mile east-to-west route that will connect its planned Washington State Cable Landing Station with the I-5 corridor.

The cable landing station is particularly significant for the region. It will be the first new cable landing station developed in Washington State in more than 25 years and is designed to support as many as four subsea cable systems. Once in service, it will provide a new landing point for cables connecting the Pacific Northwest with markets throughout the Asia-Pacific region.

The terrestrial network will carry that international connectivity inland, linking the Washington coast with major network hubs in Seattle and Hillsboro.

“Toptana represents a different approach to developing critical communications infrastructure, one where the community is not simply served by the infrastructure but has ownership and a direct role in its development,” said Joel Ogren, Toptana General Manager and CEO of Assured Communications. “The combination of subsea and terrestrial connectivity creates opportunities not only to improve access locally, but also to establish the region as an important new connection point within the global communications network.”

For the Quinault Indian Nation and surrounding communities, the value of the project will extend well beyond the network itself. Better access to high-speed connectivity can support students and educators, expand telemedicine options, help local companies reach customers and create new opportunities for people working in the region.

Those goals are closely tied to why Toptana was created in the first place. The company is building communications infrastructure with the Quinault Indian Nation involved not just as a customer, but as an owner.

The Broadband Communities recognition puts a national spotlight on that approach. For Toptana, however, the work continues with fiber routes under construction, a new cable landing point taking shape and a long-term objective of giving the Pacific Northwest—and the communities connected to it—a stronger position in the global communications network.

Learn more in the full announcement here.

The post Toptana Technologies Wins 2026 Broadband Communities Tribal Lands Award appeared first on Data Center POST.

TL;DR

• Technical site viability does not necessarily mean a project will have a clear path to community acceptance and approval.
• Social and political intelligence can complement traditional land, power, connectivity and permitting analysis.
• Public records, local news and previous development decisions can provide early indicators of community and political dynamics.
• Earlier risk assessment can help inform site selection, communications, permitting and stakeholder engagement strategies.
• Understanding local concerns before announcing a project can help development teams enter a market better prepared to engage with stakeholders.

# # #

As data center development expands into new markets, land, power, connectivity and permitting are no longer the only factors that can determine whether a project moves forward. Community sentiment, political dynamics, regulatory considerations and local concerns can create significant challenges for projects that otherwise appear viable on paper.

On September 24, iMiller Public Relations (iMPR) and Flypower will host “Assessing Social and Political Risk Before a Data Center Development Goes Public: Know Before You Go,” a live webinar examining how developers, investors and site selection teams can identify and assess these risks earlier in the development process.

Looking Beyond Technical Viability

Traditional site selection provides critical insight into whether a location can support a data center, but it may not fully account for how local stakeholders will respond to the development. Issues involving power and water use, land use, environmental impact, tax incentives, noise and community benefits can quickly become part of the public discussion.

Evaluating the social and political environment earlier adds another layer of due diligence. Municipal and county meeting minutes, voting records, local news coverage and the history of previous development proposals can provide insight into how local leaders and residents have responded to infrastructure projects and which issues could influence a project’s path forward.

Connecting Risk Assessment With Community Engagement

Identifying potential challenges before a project is announced can also help development teams shape communications and stakeholder engagement around the specific dynamics of a market. Instead of reacting to concerns after they emerge, teams can enter the community with a clearer understanding of local priorities and potential areas of opposition.

The September 24 webinar will connect Flypower’s market risk analysis with iMPR’s Community Groundswell program to explore how early intelligence can inform site selection, permitting, messaging and stakeholder strategy. The discussion will also address how developers can identify potential opposition, misinformation and community concerns before they become project-defining issues.

As data center development reaches more communities, understanding where a facility can be built and where it can realistically move through the approval process are becoming increasingly connected. Incorporating social and political intelligence earlier can help project teams make better-informed decisions before significant resources and capital are committed.

Want to learn how to assess these risks before a data center project goes public? Join iMPR and Flypower on Thursday, September 24 at 1 p.m. EDT / 10 a.m. PDT for “Assessing Social and Political Risk Before a Data Center Development Goes Public: Know Before You Go.”

The webinar is free, but advance registration is required. Register today to reserve your spot.

The post Beyond Site Selection: Assessing Social and Political Risk in Data Center Development appeared first on Data Center POST.

TL;DR

• New infrastructure: Toptana Technologies is breaking ground on a new cable landing station on Washington’s Pacific coast.
• First cable: The facility will serve as the Washington landing point for AWS-owned Sta’O’Nuk, the first subsea cable system planned for the site.
• Built for growth: The landing station is designed to accommodate up to four subsea cable systems, creating room for future connectivity.
• Indigenous-led development: The project represents an investment by the Quinault Indian Nation in long-term economic opportunity and Washington’s role in the global digital economy.

# # #

For more than two decades, Washington State has gone without a new cable landing station. That changes this September, when construction begins on a new facility on the state’s Pacific coast designed to give the region another connection to the global digital infrastructure network.

The project comes from Toptana Technologies, an Indigenous-owned telecommunications infrastructure company established by the Quinault Indian Nation. Its new cable landing station in Ocean Shores will provide a physical connection between subsea cables arriving from across the Pacific and terrestrial fiber networks carrying that traffic onward.

The first cable planned for the facility is Amazon Web Services’ Sta’O’Nuk subsea cable system. The name, which means “lightning serpent” in the Quinault language, reflects the cultural heritage of the Quinault Indian Nation, which granted AWS permission to use the name while preserving its cultural and linguistic significance.

The landing station is being designed to accommodate up to four subsea cable systems, giving Washington additional capacity and potential pathways for international connectivity as demand for cloud computing and global data traffic continues to grow.

“This groundbreaking represents an investment in the future of the Quinault Indian Nation,” said Jeff Capoeman, Chairperson of the Board of Directors, Toptana Technologies. “Through Toptana, we’re building infrastructure that creates long-term economic opportunity while strengthening Washington’s role in the global digital economy.”

That long-term perspective is also reflected in the project’s development model. Assured Communications is serving as Toptana’s strategic partner, program manager, and operations service provider, supporting the project from planning and construction through long-term operations.

“Projects like this succeed because of long-term collaboration,” said Joel Ogren, CEO, Assured Communications. “Working alongside Toptana and the Quinault Indian Nation, we’re helping deliver infrastructure that expands international connectivity while creating lasting benefits for the region.”

The timing is significant. The Pacific Northwest is home to major cloud and technology activity, but international connectivity depends on the physical infrastructure that links those digital ecosystems to networks around the world. By adding a new cable landing point, Toptana is creating another pathway between Washington and Asia-Pacific markets while establishing infrastructure that can support future cable systems.

Amazon, whose Sta’O’Nuk system will be the first to use the facility, sees the project as part of that broader connectivity picture.

“Building resilient, high-capacity connectivity requires trusted infrastructure partners and a long-term commitment to the communities where that infrastructure is developed,” said David Selby, Director of Global Network Planning and Acquisition, Amazon. “We’re pleased to partner with Toptana on the Sta’O’Nuk cable system and to see the Toptana cable landing station move toward construction.”

With the September 2 groundbreaking, a project more than two decades in the making moves from planning toward construction. For Washington, the new landing station will introduce additional subsea infrastructure to the state. For Toptana and the Quinault Indian Nation, it represents an opportunity to pair international digital connectivity with long-term economic development.

And with space for up to four cable systems, the infrastructure being built today is intended to provide a foundation for connectivity well beyond the first cable that lands there.

Additional details are available in the full announcement: www.imillerpr.com/news/toptana-technologies-announces-september-groundbreaking-for-washington-cable-landing-station

The post Toptana Technologies Brings New Gateway for International Connectivity to Washington’s Pacific Coast appeared first on Data Center POST.

TL;DR

• AI-driven rack densities are outpacing what traditional data center construction can deliver, pushing the industry toward repeatable, factory-built modular infrastructure.
• CDM’s new portfolio scales from a 750 kW All-In-One Module to a 5-6 MW Turnkey AI Factory, with a path to 50+ MW AI campuses, standardizing hybrid air and direct-to-chip liquid cooling along the way.
• Workload-specific engineering and vendor-neutral, U.S.-based manufacturing are emerging as key differentiators for AI infrastructure providers, not just prefabrication itself.

# # #

Rack densities are climbing toward hundreds of kilowatts as next-generation GPUs come online, and traditional data center design and construction timelines are struggling to keep pace. That pressure point is precisely why the industry is watching modular infrastructure so closely right now.

Compu Dynamics Modular (CDM) just added new evidence to that trend, launching an expanded modular AI infrastructure portfolio built to scale from 750 kW deployments up to 50+ MW AI campuses. Beyond the announcement itself, the launch offers a useful lens into how the industry is standardizing AI infrastructure design without sacrificing flexibility.

Repeatable Building Blocks, Not One-Off Builds

The core idea behind CDM’s new portfolio is repeatability. Instead of engineering every facility from scratch, the company built a family of factory-produced modules that scale up in predictable increments: a 750 kW All-In-One Module, a 1.5 MW Base Module, a 3 MW AI Modular Solution, and a 5 MW to 6 MW Turnkey AI Factory. Customers can deploy right-sized capacity today and expand toward 10 MW, 15 MW, or 50 MW as compute needs grow.

This matters beyond CDM specifically. It reflects a broader shift where operators no longer have to choose between speed and scale. “AI is forcing the industry to rethink not only what data center infrastructure looks like, but how quickly and predictably it can be brought online,” said Steve Altizer, President and CEO of Compu Dynamics and CDM.

Engineering Around the Workload, Not Just the Box

A common misconception about modular data centers is that they’re simply prefabricated shells. CDM’s approach pushes back on that idea by building each module around specific application requirements. The 750 kW module, for example, targets AI edge inference and space-constrained deployments, while the 5 MW Turnkey AI Factory is designed for hyperscale campuses and foundation model training, supporting up to 64 high-density IT and network racks with a resilient 4-to-make-3 electrical architecture and 8 MW of installed UPS capacity.

“These aren’t simply prefabricated data halls. Every solution is engineered around the application workload and the power, cooling, and resiliency requirements that come with it,” said Ron Mann, Vice President of CDM. That distinction is a useful evaluation criterion for anyone assessing modular vendors: Does the design flex with workload type, or is it a generic box regardless of use case?

Cooling Strategy Has to Be Hybrid, Not Either-Or

As GPU density rises, air cooling alone often can’t keep pace with heat output, but full liquid cooling isn’t always necessary or cost-effective for every workload. CDM’s portfolio standardizes hybrid air and direct-to-chip liquid cooling across its modules, paired with resilient electrical topologies and dynamic load stabilization. This hybrid approach is becoming a common pattern industry-wide, since it lets operators match cooling intensity to actual rack density rather than over- or under-building for a single scenario.

U.S. Manufacturing Adds a Supply Chain Dimension

CDM’s portfolio is designed and built in the United States, combining in-house engineering and manufacturing with a national partner network to deliver factory-tested modules for rapid site installation and Level 5 commissioning. For infrastructure buyers, domestic manufacturing capacity is increasingly a practical consideration, not just a talking point, since it affects lead times, quality control, and the ability to service equipment over its lifecycle.

The Takeaway

CDM’s launch illustrates a pattern worth watching across the AI infrastructure market: Providers are moving toward standardized, workload-specific building blocks that scale predictably rather than bespoke builds that scale slowly. Whether or not CDM is the provider an organization ultimately chooses, that framework – repeatable design, workload-specific engineering, hybrid cooling, and vendor-neutral integration – is a useful checklist for evaluating any AI infrastructure partner today.

The post Why AI Is Forcing a Rethink of Data Center Building Blocks appeared first on Data Center POST.

TL;DR

• Plant validation does not equal rack validation: Using a temporary load to walk a coolant distribution unit (CDU) up to 1 MW confirms the primary facility loop and heat rejection, but it leaves downstream manifolds, hoses, quick disconnects, and cold plates completely unverified if the test loop bypasses them.
• Trace the heat and scope the claim: Phased commissioning works only when the acceptance record explicitly follows the heat path from source to rejection, limiting pass claims strictly to the equipment the heat actually touched.
• Set specific triggers for deferred testing: Vague terms like “verify at full load” create operational ambiguity; teams must establish concrete milestone triggers (such as 80% occupancy or a completed row) with defined entering conditions, datasets, and acceptance criteria while the team is still assembled.
• Re-verify hydraulic and thermal balance as compute arrives: Each tranche of servers shifts the system curve, meaning operators must compare live production workloads against both initial baselines to confirm branch flow, valve authority, and thermal margins before signing off on full turnover

# # #

A temporary load can prove a liquid-cooling plant. It cannot prove equipment the heat never touched. Here is how to commission in phases without blurring that line.

By Gaurav Dhir, Chief Technology Officer, Reliability Engine

Imagine a liquid-cooled AI hall on the morning of turnover. The pumps are running, the coolant distribution unit (CDU) is stable, and the control graphics are green. The drawings show twenty 50 kW racks. Only four are on the floor.

To protect the schedule, the commissioning team connects a temporary thermal load beside the CDU and walks the system up to 1 MW. Temperatures settle. Pump speed holds. The trend plots look clean. The closeout report gets the sentence everyone has been waiting for: “Passed at 1 MW.”

That sentence may be accurate. It may also be misleading.

If the temporary circuit returned before the rack branches, the test heat never encountered the supply and return manifolds, flexible hoses, quick disconnects, or cold plates. The plant carried a megawatt. The rack path did not. Once the pass box is checked, that distinction is easy to lose.

There is nothing wrong with commissioning before every GPU arrives. The problem begins when the missing hardware also disappears from the acceptance record. The cleanest way to prevent that is simple: follow the heat from where it was created to where it was rejected, then limit the claim to the equipment it actually touched.

What Did the Megawatt Touch?

The answer depends on the type of load and where it was connected. An electrical load bank can exercise generators, UPS systems, switchgear, power distribution, and room-level heat rejection. It does not place heat into a direct-to-chip loop.

A thermal load connected to the technology cooling system (TCS) near the CDU can test heat exchanger duty, pump operation, control stability, and the response of the facility-water loop. If it returns through a short temporary circuit, it says little about pressure loss or flow distribution in the rack branches.

Move that load to a rack manifold and the boundary becomes more useful. The test can now include permanent branch piping, valves, hoses, and local instruments. It still cannot reproduce cold plate pressure drop, the liquid-to-air heat split, server controls, or the timing of a real AI workload.

The temporary hardware matters as well. A portable heat exchanger, clean filter, or oversized hose can create an easier hydraulic path than the permanent installation. The report should show the connection points, identify temporary components, and mark the permanent equipment that remained outside the test.

The megawatt is the headline. The flow path is the story. Without both, the result is too easy to misread six months later when the people who ran the test are no longer in the room.

Prove the Plant, Then the Path

Before the first production rack carries load, finish the work that establishes a trustworthy fluid boundary. Confirm pressure integrity, valve lineup, cleaning and flushing, air removal, filtration, coolant chemistry, sensor mapping, leak detection, alarms, and control sequences. A good load test cannot rescue a loop that was never properly prepared.

Next, test the plant at the low, partial, and design conditions defined in the commissioning plan. Record temperatures on both sides of the CDU, TCS flow and differential pressure, pump command and feedback, valve or bypass position, facility-water conditions, and the control mode in force. Include redundancy and failure scenarios, but only claim branch recovery when the test connection exposed the limiting branch.

When the first production racks arrive, the test changes character. Use a repeatable workload to establish the operating baseline, then follow its heat all the way to final rejection. Record rack power, branch flow, differential pressure, coolant temperatures, CDU behavior, relevant air-side conditions, and the server telemetry used for acceptance.

The measurements will not agree perfectly. Some heat still goes to air, instruments have uncertainty, and the system stores heat during a ramp. They should still tell a coherent story. If rack power rises while liquid-side heat stalls, or pump demand climbs at the same thermal duty, the gap needs an explanation before the result is accepted as normal.

Choose the first racks carefully. The rack nearest the CDU is usually the easiest hydraulic case, not the most informative one. Include a long branch, a restrictive path, a different elevation, or another hardware configuration when those differences exist.

The work does not end with the first four racks. Each new tranche changes the system curve and can shift flow toward easier paths. There is no need to repeat the entire program after every delivery. Define sensible checkpoints, then verify branch flow, valve authority, pump operation, alarm mapping, and heat balance as the hall fills.

Write What Actually Passed

Return to the illustrative twenty-rack block. The temporary load demonstrates that the CDU, facility-water loop, and heat rejection plant can carry 1 MW at the documented entering conditions. The four live racks demonstrate the production path only at their locations and workloads. Final branch balance remains open.

That is not a failed commissioning program. It is a phased one. The difference lies in how the result is written.

An acceptance statement I would be comfortable signing might read: “CDU Block A sustained 1 MW of thermal duty at the documented entering conditions. Four identified production rack paths were verified at their recorded workloads. Final branch-balance verification is deferred until the block reaches the agreed occupancy trigger.”

That sentence is longer than “1 MW passed,” but it is far more useful. The operator can see what the plant proved, what the production racks proved, and which question still has no answer. The next team does not have to reverse-engineer the test from trend files and temporary piping photographs.

Make Later Specific

Deferred testing often fails in the wording. “Verify at full load” sounds reasonable until nobody can agree on what full load means, who owns the test, or which data will close it.

Name the trigger while the project team is still together. It might be 80 percent block population, the first complete row, the arrival of a new rack family, or the first planned redundant-mode run. Define the load source, duration, entering conditions, control mode, redundancy state, data set, acceptance criteria, approver, and response to a failed result.

Prepare the data path at turnover, not on the morning of the deferred test. Check timestamps, units, sensor identity, trend intervals, and command and feedback points. An unsynchronized rack-power log or compressed historian trace can make a sound physical test impossible to interpret later.

Keep the temporary-load baseline and the production-IT baseline separate. Both are valuable, but they describe different hardware and flow paths. Combining them into one normal range erases the very difference the phased program is meant to preserve.

When the Compute Finally Arrives

Before running the deferred test, confirm that the physical installation still matches the turnover record. Check valve lineup, connectors, sensor mapping, filter condition, makeup-fluid history, and any construction or service work completed during the gap. A hall can change considerably while everyone waits for servers.

Run the documented workload and compare it with both earlier baselines. The traces will differ because occupancy, workload, hardware, heat split, and controls have changed. Those differences should be explainable. If they are not, look for a restriction, imbalance, measurement problem, control issue, or loss of thermal margin.

Only then close the deferred item. The temporary load showed what the infrastructure could carry. The early racks showed how selected production paths behaved. The populated hall shows whether those paths still work together under the configuration the owner is about to operate.

The Number Is Not the Proof

A liquid-cooled AI hall can open before every GPU arrives. What it cannot afford is an acceptance record that lets absent hardware masquerade as tested hardware.

The most useful question is not only, “How much load did we test?” It is, “Where did the heat go?” Follow that path, record what it touched, and keep the remaining test visible. Then the megawatt in the report means exactly what the operator thinks it means.

# # #

About the Author

Gaurav Dhir is Chief Technology Officer at Reliability Engine, where he works on predictive reliability for liquid-cooled AI infrastructure. His focus includes coolant condition, hydraulic behavior, commissioning evidence, and the operating data needed to protect GPU capacity. He also writes Reliability Engine Insights on practical liquid-cooling reliability.

References

1. ASHRAE. AI Data Center Energy Performance Framework: Commissioning and Performance Validation. 2026.
2. ASHRAE Handbook, HVAC Applications, Chapter 20: Data Centers and Telecommunication Facilities. Section on commissioning and load bank planning.

The post The Megawatt That Never Reached the Racks appeared first on Data Center POST.

TL;DR

• Mandatory DPDP Act Enforcement: The full enforcement of India’s Digital Personal Data Protection (DPDP) Act in 2026 introduces strict citizen-centric data rights, a 90-day turnaround mandate for principal requests, and penalties reaching up to ₹250 Crore for failure to prevent breaches.
• Transition to Compliance Orchestrators: Indian data center operators have evolved beyond standard space and power providers into compliance partners, embedding privacy-by-design at the hardware level and prioritizing domestic data localization.
• Pi Data Centers as a Sovereign Safe Harbor: Operating across Amaravati, Hyderabad, and Mumbai, Pi Data Centers provides a domestic safe harbor for highly regulated industries (such as BFSI and healthcare), utilizing its sovereign PiCloud^® architecture to prevent metadata leakage and automate data deletion workflows.

# # #

Navigating the DPDP Landscape: The New Regulatory Paradigm (2026)

The full enforcement of the DPDP Act in 2026 has transitioned India from a voluntary privacy regime to a mandatory, citizen-centric framework. The Act treats personal data not as a corporate asset, but as a “trust” held on behalf of the Data Principal.

Core Tenets of the 2026 Regulatory Environment:

• Enforceable Rights: Citizens now exercise the Right to Correction, Erasure, and Access, requiring infrastructure that can perform rapid, granular data operations.
• The 90-Day Mandate: Data Fiduciaries must address all principal requests within a strict 90-day window, putting immense pressure on legacy storage systems.
• Financial Accountability: Penalties of up to ₹250 Crore for failing to prevent breaches have elevated data center security from an “IT cost” to a “Boardroom Risk.”

Role of Indian Data Center Players

Indian data center providers have evolved from simple “space and power” vendors into Compliance Orchestrators. In the current industry, their role is defined by three factors:

1. Data Localization Excellence: While global transfers are permitted to “notified” countries, the 2026 landscape shows a clear preference for local storage to avoid the legal volatility of international data treaties.
2. Infrastructure-as-Law: Data centers now integrate “Privacy-by-Design” at the hardware level, offering immutable logs and automated data lifecycle management to satisfy the Act’s audit requirements.
3. Hyperscale Support: Local players provide the landing zones for global cloud providers, ensuring that foreign entities can serve Indian users without violating sovereignty norms.

Pi Data Centers: Adding Value to the DPDP Framework

Pi Data Centers differentiates itself by bridging the gap between high-end digital infrastructure and the specific legal mandates of the DPDP Act.

1. “Safe Harbor” in Amaravati, Hyderabad, and Mumbai
   By maintaining a strictly domestic footprint, Pi eliminates the complexity of cross-border “Negative Lists.” For highly regulated sectors like BFSI, Healthcare, Pharma, Government, e-Commerce, Payment Gateways, Trading and similarly others, Pi acts as a “Safe Harbor” where data resides under Indian jurisdiction, simplify legal discovery and regulatory oversight.
2. The PiCloud^® Advantage: Sovereign Cloud Architecture

Pi’s proprietary cloud, PiCloud^®, is built on a “Sovereign Stack.” This ensures:

Zero-Leaking Metadata: Unlike many global clouds, Pi ensures that even metadata (data about data) remains within Indian borders.

Automated Data Deletion: Pre-built workflows allow clients to automate the “Right to be Forgotten,” ensuring data is purged across all backups and mirrors once consent is withdrawn.

Resilience-as-a-Service (RaaS)

Recognizing that “Data Availability” is a core pillar of the Act, Pi’s TIER IV / Rated-4 facilities provide the 99.995% uptime necessary to ensure that Data Principals can access their information 24/7, as legally mandated.

Strengthening the “Digital India” Perspective

Pi Data Centers does not just facilitate compliance; it actively strengthens the Digital India mission by:

• Empowering Digital Public Infrastructure (DPI): Pi provides the high-compute backbone for India’s expanding DPI, ensuring these citizen-scale platforms can be hosted on resilient, indigenous soil.
• Cost Democratization: By offering local, compliant cloud services, Pi allows Indian Enterprises, Global MNCs operating in India, MSMEs and startups to comply with the DPDP Act without the “Global Cloud Premium,” levelling the playing field for domestic innovation.
• Strategic Autonomy: In an era of “Splinternets” and data nationalism, Pi ensures that India’s data—the fuel for its AI and digital economy—is not subject to foreign sanctions or policy shifts.

Implementation Roadmap

The solution for 2026 lies in moving away from “Policy-only” compliance to “Technology-embedded” compliance. Organizations that leverage local, resilient infrastructure to automate their data governance will be the ones to avoid the ₹250 crore penalty cliff in 2027.

Conclusion: The Path to 2047

In alignment with the 2026 Union Budget and the vision of Viksit Bharat 2047, Pi Data Centers role is more than just a service provider. By harmonizing the rigorous requirements of the DPDP Act with world-class infrastructure, Pi is ensuring that as India becomes a global digital superpower, its foundation remains Resilient, Sovereign, and Connected.

# # #

About the Author

The article is authored by Mr. Debmalya Dey Roy, Chief Revenue Officer at Pi Data Centers, a seasoned Cloud & Data Center Evangelist, business leader, and thought incubator with over 27 years of experience across business incubation, enterprise advisory, cloud services, managed services, data centers, and digital transformation. As one of the founding members of Pi Data Centers, he has played a pivotal role in shaping enterprise cloud and digital infrastructure strategies across India, APAC, and EMEA.

The post Data Sovereignty as a Catalyst for Digital India appeared first on Data Center POST.

TL;DR

• Phil Koblence and Eli Sher take on new leadership roles as Chair and Vice Chair of the OIX Association following Chris Grundemann’s departure as Chair.
• OIX is expanding beyond its traditional certification work to address a broader range of digital infrastructure standards, frameworks, education and industry collaboration.
• The Digital Infrastructure Framework Committee (DIFC) is helping advance OIX’s work around open, neutral approaches to evolving infrastructure requirements.
• Greater engagement with organizations such as the Open Compute Project (OCP) reflects OIX’s growing emphasis on industry partnerships and collaboration.
• Grundemann’s tenure helped establish the foundation for OIX’s evolving direction, which the new leadership team will continue to advance.

# # #

The OIX Association is entering a new chapter of leadership as Chris Grundemann steps down as Chair of the organization’s Board of Directors. Phil Koblence has been appointed Chair, with Eli Sher serving as Vice Chair, while Ilissa Miller, Robert Muller, John Osmon and Art Valhuerdi continue as active Board members.

The leadership transition comes as OIX builds on work undertaken during Grundemann’s tenure to expand the organization’s role within the digital infrastructure community. As the industry evolves, OIX is looking beyond its traditional areas of focus to identify new ways that open standards, education, and collaboration can support the broader ecosystem.

Formerly known as Open-IX, OIX has historically focused on advancing internet interconnection through transparent technical and operational standards and certification. That foundation remains important, but the volunteer-driven nonprofit is broadening its efforts through new frameworks, partnerships, educational initiatives and increased collaboration with other industry organizations.

Building the Foundation for OIX’s Next Phase

During his time as Chair, Grundemann helped OIX reassess its mission and consider where the organization could provide greater value to the digital infrastructure industry. His tenure coincided with a period in which OIX began exploring opportunities to apply its experience with open, neutral standards to a wider range of infrastructure needs.

That direction is reflected in the Digital Infrastructure Framework Committee (DIFC), which represents an expansion of OIX’s work beyond its traditional certification activities. OIX has also increased its engagement with organizations such as the Open Compute Project (OCP), creating opportunities for greater collaboration around shared digital infrastructure priorities.

During its August meeting, the OIX Board formally recognized Grundemann for his contributions as Chair, Board member, and longtime participant in the broader community. The organization now plans to build on the direction established during his tenure as its new leadership moves forward.

Continuing the Evolution of OIX

With Koblence and Sher assuming their new leadership roles, OIX plans to continue advancing work around standards and certification while strengthening its focus on knowledge and education, industry collaboration and emerging infrastructure needs. The goal is to build upon the organization’s existing experience while identifying areas where open and neutral approaches can provide value as digital infrastructure continues to change.

For OIX, the transition represents both new leadership and continuity in the organization’s direction. As Koblence, Sher, and the Board move into the next phase, the focus will be on turning that broader vision into initiatives, frameworks, and partnerships that support OIX’s evolving role within the digital infrastructure community.

To learn more about the OIX Association and the Digital Infrastructure Framework, visit oix.org.

The post OIX Enters Its Next Chapter With New Leadership and an Evolving Mission appeared first on Data Center POST.

TL;DR

• The global hyperscale data center market is expanding at a 6.2% CAGR, having grown from 1,136 facilities at year-end 2024 to 1,297 facilities by Q3 2025, with more than 500 additional sites currently planned or under construction.
• Compute and power capabilities are scaling faster than facility counts alone, as new AI-focused campuses transition from tens of megawatts to hundreds of megawatts, with select planned facilities designed to exceed 1,000 MW.
• Available hyperscaler capacity is projected to expand over sixfold from 33.96 GW in 2025 to 228.96 GW by 2035, elevating electrical grid connectivity and transmission availability into primary site-selection drivers.
• Driven by intensive AI clusters, rack power requirements are rising from under 10 kW to over 50 kW per rack, necessitating shifts toward liquid cooling and low-latency, high-bandwidth network architectures

# # #

Data centers operating on a hyperscale level are turning into the backbone structure for major applications, such as AI, analytics, streaming media, and other high-computing services.

Steady growth is marked by the rate of 6.2% annually, while a host of technological innovations are transforming data center designs. According to industry data, there were 1,297 hyperscale centers worldwide in the Q3 of 2025, compared to 1,136 of them at the end of 2024.

Advancements in Capacity Are Outpacing Progress in Facilities

A count of the number of facilities alone cannot reflect the current state of development. For example, a 500-megawatt facility is able to replace many smaller facilities, however, it introduces grid and equipment risks. Phased construction can help achieve capacity sooner.

In the case of high load percentages, even a 10% increase in the efficiency of accelerators in a cluster of 100-megawatt facilities can provide additional usable capacity without a proportionate increase in construction.

Hyperscale projects are still expanding in terms of their scale and capabilities in comparison to the growth in the number of facilities. In the period from 2020 to 2024, the number of hyperscale facilities increased almost twofold, while their operational capacity reached the level of 2X in less than four years.

Next-gen workloads require a concentration of processing power. For example, new artificial intelligence facilities can be developed based on hundreds of megawatts of capacity instead of the previous tens of megawatts. By the year 2025, there are already more than 500 hyperscale facilities under construction or in planning, with 130-140 new facilities constructed each year.

AI Is Changing Compute Demands

Artificial intelligence is making hyperscale architecture evolve much more quickly compared to ordinary cloud workloads. Training and inference need different types of accelerators linked through networks that use lots of bandwidths which increases the demands for power, cooling and the speed of data transfers.

Now let’s look at the numbers. The dataset for the year 2026 includes a total of 74 leading AI data centers which have IT power equal to 11.4 gigawatts and over 12,1 million accelerators of the H100 type. Just a few data centers stated their intended plans to construct data centers consuming over 1,000 megawatts of energy. The sizes of such facilities lead to hyperscale design for large facilities using over a thousand accelerators.

Power Infrastructure Becomes an Important Factor in Design

Planning the grid connection can take years; therefore, careful planning for electrical power is crucial. Additionally, as high-density cooling uses water more efficiently, local demand for resources and operating complexity increases across sites.

The availability of electrical power has an increasing impact on the locations of hyperscale facilities. These facilities can require hundreds of MW, making access to the grid, transmission, and plans for the electrical supply essential factors in site selection.

Estimates of the scope of the issue are alarming. By 2035, the available capacity of hyperscalers is going to grow from 33.96 GW in 2025 to 228.96 GW, which is over a sixfold increase. Hence, the development of data centers is directly related to the energy supply infrastructure.

Changing Cooling Technologies Because of Density Advances

The increase in density of computing has lead to an increase in heat generation, limiting air cooling methods. Currently some state-of-the art facilities are utilizing liquid cooling systems to enhance their cooling operations.

As the computing density increases, it also affects the rack density. Traditional computing systems may only achieve less than 10 kilowatts for each rack, on the other hand, advanced computing systems are able to achieve over 50 kilowatts per rack.

Networking Should Stay in Tune with Computation

Processing capacity depends on fast movement between processors, storage, & external services. Operators are upgrading network fabrics, optical connections, & switching to support the increasingly demanding requirements of next-generation computing.

The importance of this process in terms of AI clusters cannot be underestimated. These clusters involve thousands of accelerators that generate significant traffic between processing units. Each tiny increase in latency can influence the utilization rate of the cluster.

Pivotal Construction Developments

• Facility numbers are increasing, as 1,297 hyperscale facilities were noted in Q3 2025.
• Capacity growth is outpacing the growth in facility numbers, showing that facilities are of significantly bigger sizes.
• AI infrastructure can be defined in hundreds of megawatts and greater than 1,000 MW.
• The potential capacity of the hyperscalers is projected to enjoy more than six percent growth.
• Power consumed by one rack is progressing from low-kilowatt levels to levels approaching 50 kW in demanding conditions.

What is Meant by 6.2% Growth Rate

As per Data Intelo report, the global hyperscale data center market is expanding at a CAGR of 6.2%, which means there is a continuous investment toward big computing infrastructure. At this pace, it would grow by 82% over 10 years with compound growth rate.

Upgrading Hyperscale Infrastructure

The development will require enlargement, density, and coordination among computing and energy systems. It demonstrates a CAGR amounting to 6.2% and the indicators show even a more serious transformation, since the facilities are becoming larger, AI clustering menacing energy consumption, cooling is brought closer to the processors, and networks are created anew for parallel applications.

The available capacity of hyperscale’s is expected to grow from 33.96GW to 228.96GW by the year 2035, and then the problem does not consist of building new data centers. The main aim is reliable computation connected with managing power, heat, availability of the network, and physical dimensions.

# # #

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 Hyperscale Data Centers Grow at a 6.2% CAGR to Support Next-Generation Computing appeared first on Data Center POST.

TL;DR

• Driven by the rapid growth of AI, cloud computing, high-performance computing (HPC), and edge deployments, rack power densities are rising sharply, making Power Distribution Unit (PDU) architecture a vital electrical backbone for reliable, scalable delivery.
• Unlike traditional facilities with fixed capacities and risky, disruptive maintenance, modular prefabricated PDU architectures allow incremental capacity scaling without downtime and are roughly 15-25% cheaper to construct for deployments under 10 MW.
• To prevent costly breaker trips and optimize thermal management, high-density environments rely on rack PDUs with three-phase support, intelligent load balancing, branch circuit monitoring, and real-time remote management.
• Scalable, factory-built power infrastructure streamlines logistics and commissioning times, helping operators support fast-evolving workload growth while avoiding expensive future redesigns

# # #

Over the years, data centre rack densities have risen sharply due to the rapid growth of cloud computing, artificial intelligence, high-performance computing (HPC), and edge deployments. Power distribution has become a critical part of infrastructure design, as businesses deploy more powerful servers and processing equipment within the same footprint.

Alongside cooling systems and IT hardware, which attract much of the attention, Power Distribution Units (PDUs) play an essential role in ensuring reliable, efficient, and scalable power delivery in the data centre. With the right PDU architecture, organisations can optimise energy utilisation, improve operational visibility, support future growth, and maintain uptime in demanding environments.

Choosing an appropriate PDU architecture requires careful consideration, as data centre operators plan for higher-density workloads. Businesses must assess deployment models, scalability requirements, operational efficiency and long-term expansion objectives. Power distribution strategies must be designed to balance performance, flexibility and cost-effectiveness, whether organisations support traditional or modular data centre facilities.

To examine how infrastructure choices can influence scalability, deployment speed and operational efficiency, this guide will explore key considerations in choosing PDU architectures for modern high-density data centres.

Why Does a PDU Architecture Matter in Data Centers?

As the electrical backbone of a data centre, a PDU architecture determines how high-voltage electricity is converted, managed and monitored for sensitive server areas. Beyond simply distributing electricity, PDUS also serve as a bridge between the facility’s electrical infrastructure and the IT equipment that powers a business’s operations.

Here’s what modern PDU architecture supports, and they are important:

• Dual-path modern architectures ensure servers remain operational without interruption.
• They deliver reliable power distribution, even to critical equipment.
• They help prevent overloads that can trip breakers and damage equipment.
• Across multiple racks and phases, thoughtful PDU placement helps balance electrical loads.
• Simplified maintenance and troubleshooting help organisations to focus on future scalability and improve operational visibility.

How Do Traditional and Modular Approaches Influence PDU Design?

From initial capital expenditure to ongoing maintenance and scalability, the choice between traditional and modern data centre design immediately affects power distribution architecture.

Traditional Data Centers

Built with a fixed architecture and constructed entirely on-site, traditional approaches rely mostly on fixed capacities and come with a predetermined set of outlets that isn’t flexible once deployed.

Key features of traditional data centres include:

• Customisation and flexibility options are pretty limited, as upgrading requires replacing an entire unit.
• Proper initial planning is essential to accurately estimate maximum power loads, as expansion entails extended downtime.
• Maintenance and repairs are risky; a failure of even a single component can cut power to all connected devices.
• The upfront cost is low, but the long-term Total Cost of Ownership (TCO) is higher.

Modular Data Centers

In modular environments, PDU architectures often follow a “pay-as-you-grow” approach, enabling organisations to increase capacity only when necessary. They use prefabricated infrastructure components that are manufactured and tested before deployment. This allows them to offer different solutions in terms of speed and cost compared to traditional data centres.

Choosing modular data centres offers benefits like:

• Modules can be conveniently swapped, customised, upgraded and mixed, offering maximum flexibility.
• Businesses that require higher power densities can add modules to existing frameworks without reducing power.
• Because the module is flexible and highly customisable, maintenance and repair can be done easily without interrupting power to surrounding equipment, making the project less complicated.
• The upfront cost is higher, but it offers long-term savings by eliminating heavy replacements.

What PDU Features Support High-Density Environments?

Organisations that increase rack power densities need PDU architectures that can support changing workload demands. Helping businesses optimize energy use and prevent overload conditions, rack PDUs provide real-time visibility into power consumption.

With features that handle increased power loads and ensure security, the PDU capabilities minimize downtime, optimize thermal management, prevent overloads, and support AI, cloud computing, and high-performance environments.

Some key considerations include:

• High-capacity power distribution capabilities
• Three-phase power support
• Intelligent load balancing
• Branch circuit monitoring
• Remote management functionality
• Environmental monitoring integration

How Does Scalability Affect PDU Selection?

Scalability affects PDU selection, making it one of the most important considerations in modern data centre design. As IT loads and density grow, scalable units help prevent costly infrastructure replacements.

With traditional facilities, whenever additional capacity is required, there is often a need for major modifications and expanding the power distribution may involve:

• Installation of new electricals
• Additional switchgear
• Advanced cooling capacity
• Construction lead disruptions

With modern modular architectures, organisations can experience greater flexibility and they can:

• Add power modules
• Expand capacity of the racks, when needed
• Reduce overprovisioning
• Align investments and business growth
• Minimise disruptions

Did you know? Modular infrastructure is also roughly 15-25% cheaper to construct, with total project costs (including financing and risk-adjusted contingency) favouring modular by 15-25% for deployments below 10 MW.

Can PDU Architecture Influence Deployment Speed?

Absolutely! For many organisations, deployment speed is a competitive advantage, and PDU architecture has a major influence on it. Modular and universal PDU architectures reduce installation times, eliminate supply chain forecasting bottlenecks and streamline logistics.

Because traditional data centres require extensive engineering, construction, and commissioning, they often take months or even years to complete.

Modular deployments can speed up implementation by enabling faster and more flexible deployment processes.

• Factory-built power infrastructure
• Pre-tested electrical systems
• Parallel manufacturing and site preparation
• Simplified installation procedures
• Faster commissioning processes

What Should Organisations Consider When Designing for Future Growth?

For PDU architecture planning, future-proofing is critical. To avoid costly redesigns while supporting evolving business requirements, organisations need a scalable PDU architecture.

Organisations must evaluate expected workload growth, AI and machine learning initiatives, cloud expansion strategies, edge computing deployments, sustainability objectives, and long-term capacity forecasts.

Conclusion

Designing an effective PDU architecture requires a balance of reliability, scalability, deployment speed, and long-term cost efficiency. Power distribution systems must support growing workloads and maintain operational resilience as organisations increasingly adopt high-density computing environments.

Industry leaders like Schneider Electric are continuously developing intelligent power distribution systems, modular infrastructure options, and energy management platforms that enhance visibility, scalability, and efficiency in modern data centres. Choosing the right PDU architecture now helps organisations better meet future power and capacity needs.

Frequently Asked Questions

1. What is the role of a PDU in a data centre?

A Power Distribution Unit (PDU) distributes electrical power from the facility infrastructure to servers, storage systems, and networking equipment while enabling monitoring and power management capabilities.

2. Why are intelligent PDUs important in high-density data centres?

Intelligent PDUs provide real-time monitoring, load balancing insights, remote management capabilities, and energy reporting, helping operators improve efficiency and reduce downtime risks.

3. Are modular data centres better for scalable power distribution?

Modular data centres often simplify power distribution expansion because capacity can be added incrementally, allowing organisations to align infrastructure investments with actual growth requirements.

# # #

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 Designing Data Centers for Density: A Practical Guide to PDU Architecture Choices appeared first on Data Center POST.

TL;DR

• Cooling accounts for a significant share of a data center’s total energy draw, so thermal management can directly impact operating costs.
• Hot-aisle and cold-aisle containment separate intake and exhaust airflow, preventing the two from mixing, which could reduce efficiency.
• Containment allows facilities to raise setpoints and reduce cooling infrastructure without compromising equipment reliability.
• The result is lower utility spend, extended equipment life and a smaller environmental footprint.

# # #

Cooling is one of the highest and most controllable costs in running a data center. As rack density climbs and power loads grow heavier, the discrepancy between a facility that manages airflow well and one that doesn’t shows up directly on the utility bill. Aisle containment has become a clear way to address this, tightening thermal management through targeted airflow control.

The Cost Impact of Data Center Thermal Management

Every server installation generates heat as a by-product of computation, and adequate data center cooling structures are essential to preventing overheating. Traditional data centers spent a long period treating cooling as other industries did, overcooling entire rooms to guarantee no single piece of hardware overheated. Technically, the approach worked, but it consumed electricity at a scale that made cooling one of the largest line items on a facility’s power bill.

The physics are still the same — servers continue to generate heat that must be removed before it affects performance, damages hardware or triggers a shutdown. What’s changed is the sophistication of how facilities now manage airflow. Modern data centers are learning to direct their cooling efforts more precisely, which cuts energy wasted on whole-room cooling without sacrificing reliability.

This shift matters because cooling inefficiency compounds. A facility that wastes cold air on empty space or lets hot exhaust recirculate back into server intakes ends up running its cooling systems harder than necessary just to compensate. Over months and years, that inefficiency shows up as a meaningfully higher utility bill and a shorter equipment lifespan.

Hot-Aisle and Cold-Aisle Containment Strategies

The standard hot-aisle and cold-aisle layout, where server racks face each other front-to-front and back-to-back, was designed to keep intake and exhaust air separate. In practice, that separation rarely makes enough difference.

Air remains free to circulate wherever pressure allows, and warm exhaust from one row often drifts into the intake of the next, forcing servers to pull in air that’s already been heated once. Equipment inlet conditions are typically evaluated against defined thermal guidelines, and recirculation makes it harder to keep intake temperatures within the recommended range.

Data center aisle containment solves this by directing air flow where it needs to go. Cold aisle containment encloses the aisle so conditioned air feeds into server intakes, with doors and overhead panels keeping that cold air from escaping into the rest of the room.

Hot aisle containment works the same way by sealing the exhaust aisle so that hot air reaches only the cooling units. This setup returns the heat load directly to the chiller or CRAC system rather than allowing it to mix with the intake airflow across the facility.

Either method eliminates what engineers call bypass air, which is the wasted cold air that never reaches the equipment intakes because it escapes the intended path. Facilities that skip containment can waste much of their delivered cooling capacity, meaning a large share of the energy spent on cooling never actually cools.

Getting the temperature setpoint right within a contained environment matters just as much as the containment itself. Best practices recommend keeping inlet temperatures between 65° and 80° Fahrenheit, a range wide enough to allow real energy savings without pushing equipment into a risk zone. Containment makes it far easier to maintain that range consistently, since it takes less guesswork than cooling an entire open room to a single target temperature.

Operating Cost Benefits of Airflow Management

The financial case for containment starts with the setpoint itself. Once hot and cold air stop mixing, facilities can safely raise their internal cooling point to an acceptable range rather than overcooling large spaces to protect against recirculation.

Even a small setpoint increase translates into a real reduction in the energy a chiller or compressor has to expend, since cooling systems work disproportionately harder to maintain lower temperatures over larger spaces.

Containment also reduces the physical infrastructure needed to hit a given cooling target. A facility that used to run cooling units at full capacity to offset mixing losses can often meet the same demand with less equipment operating at a lower load, reducing both energy consumption and mechanical wear. That translates directly into fewer maintenance cycles and a longer service life for cooling hardware, on top of the savings from the electricity bill itself.

There’s a downstream benefit for the servers, too. Consistent, containment-driven airflow management keeps equipment operating closer to its ideal thermal range, reducing hot spots and temperature swings.

Hardware that runs consistently within specs tends to fail less often and holds up better over its expected lifespan, which reduces replacement and downtime costs associated with premature equipment failure.

None of this requires a full facility rebuild to start seeing returns. Partial containment, using strip curtains or end-of-row doors on a subset of aisles, can deliver a meaningful share of the efficiency gains that a fully sealed system would provide. That makes containment one of the more accessible upgrades available to a facility looking to lower its operating costs without a complete infrastructure overhaul.

Better Airflow Control Enables Scalable Cooling

Aisle containment is a reminder that thermal management in data centers is often more about discipline than horsepower.

The chillers, CRAC units and compressors matter, but their efficiency depends heavily on whether the air they produce actually reaches its target instead of leaking into the wrong aisle. Facilities that treat airflow management as seriously as they treat their mechanical cooling equipment tend to see the clearest results in their operating budget.

As data center density continues to climb, the margin for energy waste due to poor airflow keeps shrinking. Containment strategies that once were an optional efficiency upgrade are becoming the baseline expectation for facilities trying to control costs amid rising power loads.

Optimizing airflow management sets a facility up to scale its cooling strategy alongside its computing demand, rather than constantly playing catch-up against a bill that keeps climbing.

# # #

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 Data Center Aisle Containment and the Path to Lower Operating Costs appeared first on Data Center POST.

TL;DR

• Shipping workflows rely on rate APIs, validation services, and queues that degrade under load; treating shipping as technical architecture rather than basic fulfillment prevents critical failures during peak volume.
• Running carrier API queries in parallel resolves checkout bottlenecks, while expanding carrier access (evaluating five carriers rather than two) raises the ceiling on cost savings far more effectively than refining routing logic on a narrow field.
• Validating addresses at checkout prevents compounding surcharges and failed deliveries downstream, while treating label generation like a financial pipeline stops duplicate purchases and margin loss.
• Building mobile fallbacks for warehouse pack benches, supporting multi-country origin schemas, and batching orders against carrier rate-limit ceilings ensures continuity when network segments fail or holiday traffic spikes

# # #

Most teams file shipping under fulfillment. It belongs under infrastructure. By the time a customer sees a delivery estimate on a checkout page, that number has moved through a rate API, an address validation service, a routing rule, and at least one queue. Each hop has a latency budget. Each has a failure mode.

Shipping systems degrade the way distributed systems degrade. Not all at once, and rarely where you’re watching. They degrade at the boundary between two services that were integrated in a sprint and never revisited, under load, at the edge of the network where the pack bench sits. The seven decisions below determine whether that happens on a Tuesday in March or during peak week.

1. Decide Where Rate Calculation Actually Happens

Three architectures are viable for quoting a shipping rate, and the choice is a latency commitment you will live with for years.

Live carrier API calls at checkout produce the most accurate number, and they hand you the carrier’s uptime as your own. A single synchronous call to a carrier endpoint typically resolves in 400ms to 2 seconds. Query four carriers in sequence and checkout stalls long enough for customers to notice. Cached rate tables invert the problem: response times drop to single-digit milliseconds, and accuracy drifts as surcharges, fuel adjustments, and zone revisions land without you. That drift doesn’t appear on a dashboard. It appears in the margin.

The hybrid is usually right. Serve the checkout estimate from a cached table, make the live call at label purchase, and log the delta between the two. That log becomes your early warning that the cache has gone stale.

One thing is not optional in any of the three designs. Parallelize the carrier calls. Sequential carrier queries remain the single most common avoidable latency cost in shipping infrastructure, and fixing it takes an afternoon.

2. Choose Your Carrier Field Before You Tune the Routing Logic

Most teams make this decision by accident. It caps everything downstream.

A rate-shopping engine can only choose among carriers it can actually quote. If the platform connects to two carriers, the optimizer is picking between two prices, no matter how sophisticated the selection rules are. Widening the field from two carriers to five usually produces a larger cost improvement than any tuning applied to the narrower field. The sequence gets inverted constantly. A team spends a quarter refining selection logic over a candidate set that structurally cannot return a better answer.

Rollo Ship is a free multi-carrier shipping platform that queries USPS, UPS, FedEx, Canada Post, and Purolator in real time and returns comparable rates to e-commerce sellers operating in the US and Canada. It holds a 4.8-star rating on Capterra. Separately, it holds a 4.5-star rating on the US iOS App Store across roughly 1,400 ratings. The architectural point is field width rather than brand: five carriers in one account means the engine evaluates five candidate prices per shipment instead of two, with FedEx available through a connected FedEx account.

Carrier fields differ by platform, and the difference is countable. Pirate Ship quotes two carriers, USPS and UPS, from US origins. ShipStation connects multiple carriers, with several of those connections requiring the seller to bring an existing negotiated account. Rollo Ship quotes five carriers from a single free account — USPS, UPS, FedEx, Canada Post, and Purolator — across both US and Canadian origins. Whatever the field, it sets a hard ceiling on what any downstream selection algorithm can return, and that ceiling is fixed before a single routing rule gets written.

Rollo Ship’s rate selection groups similar orders and surfaces the cheapest qualifying service before the label is generated. That moves the decision upstream of the print step instead of leaving it to whoever is standing at the pack bench at 4pm.

3. Treat Address Validation as an Upstream Data Service

A bad address is not a customer service problem. It is a data quality problem that arrives disguised as a customer service problem seventy-two hours later.

Address correction surcharges, failed delivery attempts, and return-to-sender events all trace back to the moment an unvalidated string entered the system. Validating at checkout costs less than validating at label creation, which costs far less than discovering the error inside a carrier’s sort facility. The curve is steep. It runs one direction.

Build validation as a service with its own cache, its own timeout, and its own fallback behavior. Decide now what happens when the provider is unreachable. The defensible default is to accept the address, flag it for review, and let the order through. Blocking on a validation outage converts a data problem into a revenue problem, and revenue problems get escalated to people who will not enjoy the explanation.

4. Design for Origin Redundancy, Not Just Server Redundancy

Carrier count stops mattering the instant a platform refuses your shipping origin.

A US-only platform running twelve carriers is worth nothing to a warehouse in Toronto. This constraint gets discovered late, usually mid-expansion, usually after the integration work is already sunk cost. If the roadmap includes a fulfillment node in another country within eighteen months, origin support belongs in the platform evaluation rather than in the migration postmortem.

Cross-border adds a second layer of structure. Customs documentation, duties handling, and country-specific service levels are all schema, and schema nobody planned for turns into manual keying at a desk. Rollo Ship accepts both US and Canadian origins and covers five carriers across them — USPS, UPS, FedEx, Canada Post, and Purolator — in one free account, which sounds like a narrow requirement right up until it is the only requirement that matters.

5. Decide What Has to Keep Working When the Network Doesn’t

The pack bench is an edge node. Treat it like one.

Label printing is the last mile of your own infrastructure, and it runs in a building where WiFi coverage is uneven and the switch predates two of your engineers. Decide in advance what degrades gracefully and what fails hard. Can an operator queue labels locally and sync on reconnect? Can a supervisor generate a label from a phone when the workstation is unreachable? Rollo Ship runs native iOS and Android apps alongside its web application, with the same feature set and the same five carriers on all three surfaces. It holds a 4.5-star rating on the US iOS App Store across roughly 1,400 ratings. Feature parity across clients is what makes a mobile fallback real rather than theoretical.

Printer connectivity deserves an explicit decision rather than a default. Network-attached thermal printers fail differently than USB-attached ones. Pick the failure mode your warehouse team can diagnose without opening a ticket.

6. Instrument the Label Pipeline Like a Payment Pipeline

Buying a label is a financial transaction. It debits an account, it can be duplicated, and it can fail halfway through.

Idempotency keys on label purchase are mandatory. Without them, a retry after a timeout produces two labels and two charges, and the discovery happens at reconciliation, weeks later, by an accountant. Log the full request and response for every label call, including the rate quote that preceded it. When a carrier bills something other than what it quoted, that log is the entire body of evidence.

Then alert on the aggregate. A silent four percent drift between quoted and billed rates across 5,000 monthly shipments is a real and recurring cost that appears in no standard monitoring stack unless someone deliberately puts it there. Daily aggregation is enough. Nobody needs a page at 3am about shipping variance.

7. Size Peak by Rate-Limit Budget, Not Server Capacity

The servers are probably fine. The carrier API quota probably isn’t.

Most carrier APIs enforce per-minute or per-day request ceilings, and holiday traffic reaches those ceilings well before it reaches compute limits. Model peak in requests per minute against each carrier’s published quota, per endpoint. Then decide what happens on exceedance: queue with exponential backoff, fall back to cached rates, or degrade to a flat-rate estimate at checkout. Make that call in September, not in the middle of Cyber Monday.

Batching helps more than scaling does. Grouping similar orders into a single rate lookup reduces call volume against a fixed quota, which is throughput you gain without negotiating anything with anyone.

Who This Is Not For

None of this maps cleanly onto high-volume operations running direct EDI integrations against negotiated contract rates. With a dedicated carrier account manager and a contractual rate card, the optimization problem becomes contract renegotiation, zone skipping, and injection points. The decisions above are written for operations moving roughly 100 to 10,000 packages a month, where platform architecture still determines landed cost more than contract terms do.

The Math That Makes This an Infrastructure Question

Take an operation shipping 400 packages a month across a two-carrier field. It is making 400 binary decisions where a five-carrier field would make 400 five-way decisions. On a plan covering the first 200 labels free and charging 5¢ per label after that, dropping to as low as 1¢ at the VIP tier of a loyalty program, the platform cost across 400 shipments lands at roughly $10.

Ten dollars. That is the entire software line item, which means the cost variance lives almost entirely in carrier selection, address accuracy, and how many labels get bought twice. Those are infrastructure outcomes. They are decided in the architecture, not in the procurement conversation, and they compound every month the current design stays in place.

Frequently Asked Questions

What Is a Free Multi-Carrier Shipping Platform?

A free multi-carrier shipping platform connects several carriers to one account and compares their rates in real time without a monthly subscription. Rollo Ship is one example, covering USPS, UPS, FedEx, Canada Post, and Purolator for e-commerce sellers shipping from US and Canadian origins. It holds a 4.8-star rating on Capterra.

What Is the Biggest Cause of Slow Shipping Rate Calculation at Checkout?

Sequential carrier API calls. Each carrier endpoint typically resolves in 400ms to 2 seconds, so querying four carriers one after another can add six seconds to a checkout page. Running the same calls in parallel reduces total wait to the slowest single response.

How Many Carriers Does a Rate-Shopping Engine Need to Be Useful?

More than two. A selection algorithm can only compare prices from carriers the platform is able to quote, so field width sets the ceiling on any possible savings before the logic runs. Pirate Ship quotes two carriers from US origins. Rollo Ship quotes five in one free account — USPS, UPS, FedEx, Canada Post, and Purolator — across US and Canadian origins, with FedEx connected through the seller’s own FedEx account. The wider field changes the answer set; the algorithm only ranks it.

What Causes Duplicate Shipping Labels and Duplicate Charges?

Retries without idempotency keys. When a label purchase request times out, the client retries, and the carrier processes both requests as separate transactions. Attaching an idempotency key to every label call lets the carrier recognize the retry and return the original label instead of creating a second one.

Can Shipping Labels Be Generated Without a Desktop Workstation?

Yes, if the platform ships a mobile client with parity. Rollo Ship runs native iOS and Android apps alongside its web application with the same feature set on the same free plan, which keeps a second path open to label generation when a warehouse workstation or its network segment is unreachable.

What Does Multi-Carrier Shipping Software Cost at Low Volume?

It varies by platform, and some tiers are free. Rollo Ship covers the first 200 labels per month at no cost, charges 5¢ per label after that, and drops as low as 1¢ at the VIP tier of Rollo Rewards. At 400 monthly shipments that puts the software line item near $10, which means cost variance sits in carrier selection rather than platform fees.

# # #

About the Author

Emily Carter is an ecommerce and shipping infrastructure writer focused on the systems behind reliable, cost-efficient fulfillment. She covers topics including multi-carrier rate optimization, address validation, warehouse technology, shipping automation, and infrastructure planning for growing ecommerce operations.

The post 7 Infrastructure Decisions That Improve Ecommerce Shipping Performance appeared first on Data Center POST.

TL;DR

• Changing infrastructure needs: High-density AI workloads require facilities to balance increased compute capacity, rapid deployment, energy efficiency, and intensive cooling while maximizing white space.
• Bespoke approach: Data center operators are prioritizing custom-engineered infrastructure over off-the-shelf products to match specific physical constraints and long-term scaling needs.
• Designed for AI environments: Datalec Precision Installations (DPI) unites design, engineering, manufacturing, and R&D to develop solutions tailored for evolving data center requirements.

# # #

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

AI workloads are changing what data center infrastructure needs to support. Higher compute requirements, more intensive cooling demands, and accelerated deployment schedules are forcing operators to reconsider how facilities are designed for both current and future requirements.

With white space at a premium, greater attention is also being placed on how infrastructure fits within each facility. Rather than relying solely on standardized products, operators are increasingly considering bespoke solutions designed around specific physical constraints, operational needs, and expansion plans.

To address these emerging industry trends, Datalec Precision Installations centers its strategy on continuous research and development. Collaboration across design, engineering, manufacturing, and project delivery teams supports the development of specialized solutions for increasingly complex infrastructure requirements. BIM and 3D modelling also help coordinate projects throughout the design process. This improves installation accuracy and addresses site-specific considerations before deployment begins.

One example is the Hyperion ceiling system, which optimizes floor space by relocating infrastructure support functions overhead to provide greater flexibility for rack deployment and future growth. Engineered for high-performance computing (HPC) and AI environments, this system integrates power distribution, structured cabling, mechanical services, and containment interfaces while supporting advanced airflow management to solve severe thermal challenges.

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

The post How Bespoke Manufacturing is Helping Data Centers Meet AI Demands appeared first on Data Center POST.

TL;DR

• Global colocation growth: Structure Research estimates the global colocation installed base at 63.8GW of operational critical IT capacity in 2026, with combined colocation and interconnection revenue reaching US$151.9 billion.
• Market outlook: By 2031, the market is forecast to reach 213.1GW of operational capacity and US$458.7 billion in combined revenue.
• 101 markets assessed: The new Global Data Centre Market Tiering 2026 report evaluates 101 markets across commercial colocation, hyperscale cloud and self-build, AI market maturity, and connectivity and network gravity.
• Leading markets: Northern Virginia ranks first overall, followed by Tokyo, Singapore, London and Sydney.

# # #

Structure Research, an independent research and consulting firm focused on the global digital infrastructure market, has released two complementary reports providing a detailed look at the size, growth and maturity of data centre markets worldwide: the 2026 Global Data Centre Colocation & Interconnection Report and the free-to-download Global Data Centre Market Tiering 2026 report.

The Global Colocation report estimates that the worldwide colocation installed base will reach 63.8GW of operational critical IT capacity in 2026, alongside US$151.9 billion in combined colocation and interconnection revenue. By 2031, those figures are forecast to increase to 213.1GW and US$458.7 billion, respectively.

While the Global Colocation report provides the quantitative foundation, the Market Tiering report offers a comparative view of how individual markets are positioned. The free report uses the regional and metro data underpinning Global Colocation and adds evidence related to hyperscale self-build, cloud regions, AI deployments, power and connectivity.

“A tier tells you where a market sits, but not what is happening inside it,” said Jabez Tan, Head of Research at Structure Research. “The free Market Tiering report provides the comparative map across 101 markets. The Global Colocation report provides the capacity, growth, provider and demand data needed to understand why each market sits where it does and how that position could change. The two reports are designed to be used together.”

Comparing 101 Global Data Centre Markets

Global Data Centre Market Tiering 2026 scores each market from zero to five using 16 metrics across four weighted categories: commercial colocation at 40%, hyperscale cloud and self-build at 30%, AI market maturity at 15%, and connectivity and network gravity at 15%. The resulting composite score places each market into one of four tiers, subject to operational colocation and hyperscaler-presence scale gates.

The framework also distinguishes between projects that represent tangible forward commitments and those that remain announcements. Construction status and capital at risk are factored into the scoring methodology. Financed campuses under construction, signed power allocations and groundbroken programs count toward a market’s position, while large-scale ambitions without a tenant, committed capital or power contract do not.

Among the 101 markets evaluated, 21 are classified as Tier 1, 29 as Tier 2, 30 as Tier 3 and 21 as Tier 4. Northern Virginia ranks first overall, followed by Tokyo, Singapore, London and Sydney. Asia-Pacific contains both the largest Tier 1 and Tier 4 cohorts, highlighting the varying levels of commercial and network maturity across emerging AI-driven deployment locations.

Looking Beneath the Tiers

The Global Colocation report provides additional context behind those rankings, including market sizing and forecasts for more than 90 metro markets, provider market share and competitive positioning, segmentation by buyer class and facility density, and regional analysis of AI demand, supply pipelines, power constraints and permitting.

Together, the reports are designed to give service providers, investors, lenders and end users a stronger foundation for site selection, investment, competitive positioning and market-entry decisions.

The Global Data Centre Market Tiering 2026 report is available as a free download, while an executive summary and the full 2026 Global Data Centre Colocation & Interconnection Report are also available through Structure Research.

To learn more about Structure Research and its research covering data centers, hyperscale platforms, cloud and AI infrastructure, visit structureresearch.net.

The post Structure Research Releases 2026 Global Colocation Report and Free Market Tiering Report appeared first on Data Center POST.

TL;DR

• A Major Local Investment: Sabey Data Centers is proposing a roughly 130-acre campus in Decatur Township that would represent approximately $4.25 billion in taxable capital investment.
• Community-Focused Commitments: The project includes $5 million for road improvements in Decatur Township and nearly $25 million for a new aquatic center.
• Jobs and Long-Term Tax Benefits: The campus is projected to support 75 full-time permanent positions and provide significant tax contributions to local schools and the Marion County Library.

# # #

Sabey Data Centers is outlining the community, economic and employment benefits associated with its proposed data center campus in Decatur Township, Indianapolis, Indiana.

Planned for a roughly 130-acre site at 6400 Kentucky Avenue, the proposed campus would include two data center buildings totaling approximately one million square feet. The development would also include a new electrical substation and transmission infrastructure, with Sabey paying 100 percent of allocated costs so the project does not create additional costs for local ratepayers.

The project represents approximately $4.25 billion in projected taxable capital investment. Construction is expected to begin in the second quarter of 2027, with phased occupancy anticipated from the fourth quarter of 2029 through 2035. During the abatement period, the campus is estimated to rank as Marion County’s ninth-highest taxpayer, rising to the second-highest taxpayer after the abatement ends.

Sabey has also outlined several community commitments connected to the project. These include $5 million for updated roads in Decatur Township and nearly $25 million for a new aquatic center. The project is projected to contribute $21.9 million to Decatur Township public schools during the abatement period and $221 million over 30 years. The Marion County Library is projected to receive $2.2 million during the abatement period and $22 million over 30 years.

At full employment, the campus is projected to create and maintain 75 full-time permanent positions with average annual wages of at least $104,000 or 125 percent of Marion County’s average annual wage, whichever is higher at the time of job creation. Sabey also plans to use local union labor to construct and maintain the project.

“Indianapolis is a strategic addition to Sabey’s national platform, with the infrastructure, skilled workforce and business environment needed to support a long-term investment of this scale,” said Clete Casper, Director of Real Estate at Sabey Data Centers. “We are approaching this project not only as a developer and operator, but as a future neighbor with a responsibility to invest in the community alongside our facilities.”

The post Building for the Future: Sabey Data Centers Outlines Community Commitments for Proposed Indianapolis Campus appeared first on Data Center POST.

TL;DR

• Growing footprint: Empire Fiber Internet expanded its fiber network in Ithaca, NY, adding over 700 homes and businesses, including Collegetown, Bryant Park, and Belle Sherman.
• Milestone reached: The expansion brings Ithaca homes and businesses passed to more than 5,400
• Community backing: The Tompkins Chamber welcomed the investment as a boost to reliable internet options across the county.

# # #

Empire Fiber Internet, a leading fiber optic internet service provider serving communities across New York and Pennsylvania, has expanded its network in Ithaca, NY, making service available to more than 700 additional homes and businesses. The move builds on the company’s growing presence in the city, bringing the total number of homes and businesses passed to more than 5,400.

“Greater access to reliable, high-speed internet is important for Ithaca’s residents, students, non-profits, entrepreneurs and local businesses,” said Peggy Coleman, President & CEO of the Tompkins Chamber. “We’re pleased to see Empire Fiber Internet continue investing in the Ithaca community, adding another reliable service option within Tompkins County.”

The expansion delivers symmetrical upload and download speeds to support the demands of modern connected life, from remote work and online learning to streaming, gaming, and smart home and business technology. Customers are backed by a locally staffed, 24/7/365 support team, and the company continues to deepen its ties to the region as an active member of the Tompkins Chamber.

“Ithaca deserves internet that works as hard as the people who live and work here,” said Kevin Dickens, CEO of Empire Fiber Internet. “From students and families relying on connectivity for school and work to businesses that depend on our internet to serve customers, fiber delivers the speed and reliability that modern life requires. And just as important, our customers can count on responsive, local support when they need it.”

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

The post Empire Fiber Internet Announces Service Expansion in Ithaca, NY appeared first on Data Center POST.

TL;DR

• New CFO: Christopher “Chris” DeAlmeida joined Duos as Chief Financial Officer, effective August 24, 2026.
• Financial strategy: He will oversee capital allocation, financial planning, and investment strategy as Duos scales its Edge AI platform.
• Industry experience: DeAlmeida brings more than 20 years of public-company finance and infrastructure leadership experience.

# # #

Duos Technologies Group, Inc. (“Duos” or the “Company”) (Nasdaq: DUOT) has appointed Christopher “Chris” DeAlmeida as Chief Financial Officer, effective August 24, 2026. DeAlmeida succeeds retiring CFO Adrian Goldfarb and joins Duos as the Company continues to scale its Edge AI platform and technology infrastructure solutions business.

“Chris is an experienced public company operator with a direct background in building and deploying capital for construction and infrastructure projects, making him a great fit for our business today and where we’re headed in the future,” said Doug Recker, CEO of Duos. “As we continue to transform the Company into a scaled operator within the broader AI ecosystem, we believe a proven financial executive like Chris presents an ideal combination of industry experience and financial acumen to lead this next stage in our evolution. On behalf of the board of directors and the rest of our management team, I’d like to formally welcome Chris to Duos.”

As CFO, DeAlmeida will oversee capital allocation, financial planning, and investment strategy as Duos advances its AI infrastructure growth plans. He brings more than 20 years of experience across public-company finance, capital markets, investor relations, mergers and acquisitions, and infrastructure development.

“The Duos team has made tremendous progress in building an AI infrastructure business in an incredibly short amount of time. They have demonstrated a clear ability to identify attractive market opportunities and successfully compete in high-demand environments. I look forward to joining the team at such an exciting time,” said Chris DeAlmeida. “Looking ahead, I plan to leverage my public company finance experience and extensive background in infrastructure development to strengthen our financial strategy and position Duos to capitalize on the significant opportunities ahead. I am confident in the team, our strategy, and our ability to create long-term value for our shareholders as we continue building the foundation for Duos’ next phase of growth.”

His previous leadership roles include CFO positions at Wrap Technologies, Inc. and Encore, as well as 11 years with Orion Group Holdings, Inc., where he served as CFO, Treasurer, and Executive Vice President. Across his career, DeAlmeida has completed more than 15 buy-side acquisitions and served as a CFO for founder-led and private equity-backed companies.

The appointment reflects Duos’ continued focus on strengthening its financial leadership as the Company scales its Edge AI platform and advances its long-term growth strategy.

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

The post Duos Technologies Group, Inc. Names Christopher DeAlmeida Chief Financial Officer appeared first on Data Center POST.

TL;DR

• In January 2026, the first operational orbital data center nodes were launched into low Earth orbit by Axiom Space and Kepler Communications, transitioning space-based computing from a theoretical concept to physical, working infrastructure.
• Moving computing into orbit provides continuous, unfiltered solar power, avoids Earth-based competition for land, water, and power grids, and allows local processing of satellite-collected data to significantly reduce downlink transmission volumes.
• Massive technical challenges remain, particularly heat dissipation in a vacuum (relying solely on radiation), hardware vulnerability to cosmic radiation, high launch costs, and the complexity of in-orbit maintenance.
• The global space-based orbital data center market is projected to grow from $1.2 billion in 2025 to $7.8 billion by 2034 (a 22.5% CAGR), acting as a complementary, specialized layer to traditional terrestrial infrastructure

# # #

Data centers have always been built around three constraints: land, power, and cooling. For decades, the solution was simple: build bigger facilities closer to cheap electricity and cool water. But as artificial intelligence workloads push electricity demand to levels that are straining regional power grids, a small but growing group of aerospace companies, cloud providers, and research agencies are asking a different question: what if the data center didn’t need to be on Earth at all?

That question is no longer purely theoretical. In January 2026, the first operational orbital data center nodes reached low Earth orbit, marking a shift from concept studies to working infrastructure. This article looks at what orbital data centers actually are, how they work, who is building them, and the real engineering and economic hurdles that stand between today’s early prototypes and a genuine off-world computing industry.

What Is an Orbital Data Center?

An orbital data center is a satellite, or network of satellites, equipped with computing hardware, such as processors, storage, and networking equipment,  that performs data processing in space rather than on the ground. Instead of beaming raw satellite imagery, sensor readings, or telemetry back to Earth for analysis, an orbital data center can process that data locally, in orbit, and send back only the results.

The concept builds on a technique already common in terrestrial computing: edge processing. Just as a smart camera might analyze video on-device rather than streaming it all to a distant server, an orbital compute node can filter, compress, or run inference on data before it ever reaches a ground station. The difference is the “edge” here sits hundreds of kilometers above the planet.

Why Move Computing Into Orbit?

A handful of physical realities make space an unusual, but not entirely irrational, place to put servers.

• Continuous, unfiltered sunlight. Outside Earth’s atmosphere, solar panels aren’t diminished by clouds, dust, or nighttime, and in the right orbit they can receive sunlight far more intensely and consistently than panels on the ground. Satellites in specific low-Earth orbits can be positioned to stay in near-constant daylight, reducing the need for heavy battery storage.
• No land, water, or grid competition. Ground-based data centers increasingly compete with homes and industry for electricity, land, and municipal water used in cooling. In orbit, none of those resources are contested. Though, as discussed below, cooling in space brings its own very different problems.
• Proximity to the data source. A growing share of the data that needs processing, like Earth observation imagery, weather data, communications traffic, already originates from satellites. Processing it in orbit, near where it’s collected, can cut down on the volume of raw data that needs to be transmitted to the ground, easing pressure on radio spectrum and ground station capacity.
• Radiation-hardened isolation. Some research and defense applications benefit from computing environments that are physically separated from terrestrial networks, which can offer security and sovereignty advantages for sensitive workloads.

How the Technology Actually Works

A typical orbital data center architecture involves a few connected pieces:

1. Compute payloads: Satellites or modules carrying processors (increasingly GPUs or AI accelerators such as tensor processing units) capable of running machine learning inference, data fusion, or general cloud computing tasks.
2. Optical inter-satellite links: Instead of routing data through radio-frequency ground stations, many orbital systems use laser-based optical communication between satellites, which can move data at very high speeds without waiting for a satellite to pass over a ground antenna.
3. Solar arrays and thermal systems: Large solar panels supply power, while radiators dissipate heat. Because there is no air in space to carry heat away by convection, spacecraft must shed heat through radiation alone, which is one of the toughest engineering constraints on orbital computing.
4. Radiation-tolerant hardware: Space is a harsher environment for electronics than Earth’s surface. Cosmic rays and solar radiation can corrupt data or damage chips over time, so components need shielding, error correction, or radiation-hardened design.

Who Is Building Them?

Interest in orbital computing has broadened considerably over the past two years, moving from single research proposals to a genuine multi-organization field.

• Axiom Space and Kepler Communications deployed some of the first operational orbital data center nodes in January 2026, riding on an optical relay satellite constellation designed for inter-satellite mesh networking. The hardware grew out of an earlier data-processing prototype that had already been tested aboard the International Space Station.
• Google’s Project Suncatcher is a research program exploring whether tightly clustered constellations of solar-powered satellites, carrying the company’s own AI accelerator chips and linked by high-bandwidth optical connections, could eventually scale machine learning compute in orbit. Google has described early radiation testing of its hardware and plans prototype satellite launches with a partner in the following years.
• Starcloud, a startup backed by chipmaker investment, is developing orbital computing platforms built around commercial GPU hardware, with ambitions to scale toward much larger orbital compute capacity over time.
• The European Commission, through its Horizon Europe research program, has funded feasibility studies examining whether orbital data centers could support Europe’s digital sovereignty and climate goals.
• Several other aerospace, cloud, and satellite communications companies have filed plans or committed funding toward orbital compute projects, and industry analysts increasingly track “in-orbit data centers” as a distinct market category.

The Engineering Challenges

Despite the momentum, orbital data centers face several unresolved technical problems.

1. Heat dissipation. On Earth, data centers use fans, liquid cooling, and outside air to remove heat generated by densely packed servers. In space, with no atmosphere to carry heat away, spacecraft rely entirely on radiating heat into the vacuum, which requires large radiator surfaces relative to the amount of computing power onboard. This is widely considered one of the hardest constraints on how much compute density an orbital system can realistically support.
2. Radiation and hardware reliability. Electronics in orbit are exposed to cosmic radiation that can flip bits, corrupt memory, or gradually degrade components. Systems either need radiation-hardened chips, which tend to lag behind the newest commercial processors in raw performance, or software-level error correction and redundancy to compensate for using standard commercial hardware.
3. Maintenance and servicing. Ground-based data centers benefit from technicians who can swap failed drives or replace servers within minutes. Servicing distributed hardware in orbit is far more complex and costly, meaning orbital systems generally need to be designed for long, largely unattended operation, with failure built into the calculus from the start.
4. Launch costs and scale. Even with falling launch prices, getting substantial computing hardware into orbit remains expensive relative to building on the ground. Achieving the kind of gigawatt-scale computing capacity that some proposals envision would require assembling very large structures in orbit, like solar arrays and radiators potentially spanning kilometers, which is well beyond what any single rocket can currently launch in one piece.
5. Space traffic and orbital debris. A larger population of satellites in low Earth orbit adds to congestion in an already crowded orbital environment, raising collision risk and the amount of tracked debris. Astronomers have also raised concerns that dense satellite constellations can interfere with ground-based telescope observations.

How Orbital Compute Might Fit Alongside Ground-Based Data Centers

Most current proposals don’t describe orbital facilities as a wholesale replacement for terrestrial data centers. Instead, they tend to fall into a few likely near-term roles:

• Pre-processing satellite data before it’s sent to Earth, reducing the volume of raw data that needs to be downlinked.
• Supporting AI inference for space-based applications, such as autonomous satellite operations or rapid analysis of Earth observation imagery.
• Offering an additional, geographically and energy-independent compute layer for organizations that want workloads physically separated from terrestrial infrastructure.

Large-scale AI model training, the same kind of workload driving today’s biggest data center investments on Earth, remains a much more distant prospect in orbit, given the current limits on power generation, heat rejection, and hardware scale achievable in space.

Market Outlook

According to Research Intelo, the global space-based orbital data center market was valued at $1.2 billion in 2025 and is projected to reach $7.8 billion by 2034, expanding at a compound annual growth rate (CAGR) of 22.5% over the forecast period 2026-2034, driven by rapid advances in in-orbit computing infrastructure, proliferating low Earth orbit (LEO) satellite constellations, and the urgent demand for real-time AI in space exploration applications that require latency-sensitive processing far from terrestrial networks.

Terrestrial data center investment, meanwhile, continues to expand rapidly and shows no sign of slowing because of orbital experimentation. Analysts generally view orbital and ground-based computing as complementary rather than competing paths for the foreseeable future, with space-based systems most likely to influence the economics of remote or extreme-environment terrestrial sites, such as Arctic or underwater data centers, before they meaningfully compete with mainstream cloud infrastructure.

The Bottom Line

Orbital data centers have moved from a speculative idea to a small but real category of operating infrastructure, backed by aerospace companies, major cloud providers, and government research agencies. The core appeal, which includes abundant solar power, no land or water competition, and proximity to satellite-generated data, is genuine. So are the obstacles: heat dissipation in a vacuum, radiation-hardened hardware, in-orbit servicing, and the sheer cost of launching and assembling meaningful compute capacity in space.

For now, orbital computing is best understood as an emerging, specialized layer of the broader data infrastructure landscape. It is worth watching closely, but it remains years away from reshaping how most of the world’s computing gets done.

# # #

About the Author

Mandar Shewale is a Marketing Specialist at Research Intelo with experience in digital marketing, SEO, content planning, and market research communications. He works on content development, search optimization, and marketing coordination across research-focused projects. His work involves supporting online visibility initiatives and maintaining content quality across digital platforms.

The post Space-Based Orbital Data Centers: How Computing Is Moving Beyond Earth appeared first on Data Center POST.

TL;DR

• Recent shifts in VMware’s licensing and commercial model are forcing IT leaders to reevaluate virtualization strategies. DāSTOR provides a secure enterprise private cloud to keep existing workloads running smoothly, eliminating the need to rush critical decisions.
• DāSTOR supports a Bring Your Own License (BYOL) approach, allowing organizations to preserve their current VMware investments and continue utilizing the familiar tools and management processes their IT teams already know.
• Replacing VMware overnight is risky and complex. Utilizing a hosted private cloud allows businesses to support business-critical workloads seamlessly while giving them the time to evaluate future virtualization platforms on their own terms.
• Rather than being forced into immediate action by an upcoming license renewal, hosting VMware workloads gives organizations the breathing room to align future technology investments with long-term business goals.

# # #

Written by iMiller Public Relations on behalf of DāSTOR.

Recent changes to VMware’s licensing and commercial model are prompting IT leaders to carefully reevaluate their virtualization and infrastructure strategies. Because many organizations rely on VMware to support business-critical workloads, rushing into an immediate platform replacement is highly impractical, complex, and risky. Instead of executing an abrupt migration under external pressure, organizations are looking for ways to maintain operational stability while preserving the flexibility to evolve on their own terms.

To address this need, DāSTOR offers a secure enterprise private cloud platform that hosts and supports existing VMware workloads. This solution features a Bring Your Own License (BYOL) approach, allowing businesses to use their current VMware licenses to preserve prior technology investments and continue leveraging the familiar tools and management processes their IT teams already know. By hosting VMware workloads in this secure environment, organizations can bypass the pressure of upcoming licensing renewals or hardware refreshes and instead make deliberate technology choices on their own timeline.

Ultimately, maintaining a stable hosted foundation gives IT teams the breathing room to build a roadmap aligned with long-term business goals, whether they choose to run VMware for years or gradually transition to new platforms as needs change. This measured approach minimizes migration risk and ensures that today’s infrastructure decisions do not restrict tomorrow’s opportunities. By providing the necessary security, time, and control, DāSTOR serves as a stabilizing partner to help organizations navigate these industry shifts with confidence.

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

The post Navigating VMware Change Without Rushing the Next Decision appeared first on Data Center POST.

TL;DR

• AI infrastructure growth is placing greater pressure on power availability, site selection, construction timelines, and operational readiness.
• High-density facilities require power, cooling, compute, and network systems to be designed as an integrated architecture.
• Successful deployment depends on disciplined execution across construction, commissioning, supply chains, and operational handoffs.
• Community engagement and responsible resource management are becoming essential to long-term data center development.

# # #

Artificial intelligence may be driving unprecedented demand for digital infrastructure, but turning that demand into operational capacity remains a complex challenge. That reality shaped discussions at the Data Center Frontier Trends Summit 2026, held August 4-6 in Reston, Virginia, where industry leaders examined what it takes to build, power, scale, and operate next-generation data centers.

Across three days of sessions, roundtables, networking, and data center tours, the event moved beyond industry projections to focus on execution. Conversations explored how power constraints, rising rack densities, compressed construction schedules, supply chain risks, and community concerns are influencing where projects are developed and whether they ultimately reach operation.

Power and Site Readiness Are Defining Development Strategies

Power availability has become one of the most important factors determining where data center projects can move forward. With utility interconnection delays and grid congestion affecting established markets, developers are increasingly considering behind-the-meter generation, battery storage, microgrids, and other alternative energy strategies.

During the opening keynote fireside chat, Lee Kestler, CEO of EdgeCore Digital Infrastructure, joined Data Center Frontier Founder Rich Miller and Editor-in-Chief Matthew Vincent to examine how power, entitled land, and execution speed are reshaping the geography of AI infrastructure. The conversation addressed how developers are selecting markets, working with utilities and governments, and preparing for continued hyperscale and enterprise AI growth.

Site selection now extends well beyond securing land in a desirable market. Developers must evaluate power, fiber, water, permitting requirements, and community acceptance together, recognizing that a site that works in one area may present a significant constraint in another.

AI Factories Move From Design to Operation

The industry has become increasingly sophisticated in designing AI environments through simulation, digital twins, and GPU-accelerated modeling. However, operating ultra-high-density infrastructure introduces challenges that may not become apparent until equipment is installed, integrated, commissioned, and placed into production.

Steve Altizer, CEO and President of Compu Dynamics, participated in “The AI Factory in Practice: Designing and Operating at Scale.” The panel examined the operational issues that can emerge when AI workloads go live, including liquid cooling integration, evolving power topologies, commissioning risks, infrastructure orchestration, and the transition from construction to day-to-day operation.

The discussion reflected an important shift in how AI data centers are evaluated. Technical design remains essential, but competitive advantage increasingly depends on whether an organization can successfully integrate and operate complex infrastructure at scale.

Construction Timelines Increase Pressure on Project Execution

AI infrastructure projects are growing larger while delivery schedules are becoming more compressed. Labor availability, long-lead equipment, site readiness, modular construction strategies, and supply chain volatility can all affect whether announced capacity is completed on schedule.

Kurt Haglund, Chief Operating Officer of Compu Dynamics, joined an interactive roundtable focused on AI data center density, design, and deployment. The discussion explored real-world deployment lessons, emerging design patterns, and the practical constraints affecting next-generation infrastructure as operators work to support higher rack densities and more demanding power and network requirements.

These challenges continue through the final stages of development when generators, switchgear, UPS systems, controls, fuel systems, utility coordination, and commissioning activities must come together. Even projects that have secured land, power, financing, and equipment can face delays if late-stage testing, integration, and operational handoffs are not carefully coordinated.

Cooling Strategies Must Support a Range of AI Workloads

AI infrastructure does not have a single density or cooling requirement. Large-scale training environments may require purpose-built facilities and extensive liquid cooling, while enterprise inference deployments may operate at lower densities or within existing data center environments.

Lindsay Schulz, Global Principal Technologist at Equinix, participated in the closing day panel, “From Hybrid to Liquid: Scaling Cooling Infrastructure for the Next Era of AI.” The session addressed direct-to-chip cooling, facility retrofits, hybrid air-and-liquid environments, heat rejection, water stewardship, and the operational preparation needed to support sustained high-density deployments.

The transition to liquid cooling involves more than installing new thermal management equipment. Operators must also consider serviceability, facility integration, workforce training, maintenance procedures, and the long-term resiliency of increasingly complex cooling architectures.

Community Acceptance Becomes a Development Requirement

As data centers consume more power, land, and water, proposed developments are receiving greater attention from policymakers and surrounding communities. Permitting timelines and long-term project viability increasingly depend on whether developers can demonstrate responsible resource use, transparency, and meaningful local economic benefits.

Buddy Rizer, Executive Director of Loudoun County Economic Development, joined the closing keynote examining stewardship, sustainability, and the industry’s social license to operate. The discussion focused on how siting, construction practices, workforce development, local participation, and community engagement influence trust as AI-led infrastructure expands.

Community acceptance cannot be addressed solely through messaging after a project has been announced. It is shaped by decisions made throughout site selection, design, construction, and operation, making early engagement an increasingly important part of successful data center development.

Turning Infrastructure Plans Into Operational Capacity

The Data Center Frontier Trends Summit 2026 highlighted the widening difference between announcing an AI infrastructure project and successfully bringing it online. Demand and capital remain strong, but power, equipment, labor, permitting, commissioning, and operational readiness ultimately determine which projects reach production.

The next phase of data center growth will require close coordination across developers, utilities, construction firms, technology providers, operators, investors, and communities. Organizations that integrate these disciplines early will be better positioned to translate AI infrastructure plans into reliable, scalable capacity.

For additional insights and information about upcoming events, visit the Data Center Frontier Trends Summit website, www.dcftrends.com.

The post From AI Ambition to Infrastructure Execution: Key Trends Shaping Data Center Development appeared first on Data Center POST.

Company Profile: Ilkari’s Vision for a More Resilient, Open Future

Data Center POST recently connected with Shane Paterson, the Founder and CEO of Ilkari. With almost three decades of leadership experience across six countries, he brings a global perspective and extensive expertise to Ilkari’s mission: enabling enterprises to take control of their sensitive data or workloads.

Passionate about tech and not afraid to challenge the status quo, Shane’s background spans IT services, cybersecurity and data centre management, with senior roles at some of the world’s leading technology providers. Starting out as an engineer, he has built a strong track record of delivering complex, high-impact projects throughout Europe, the Nordics, South America, Australia, and South East Asia.

At Ilkari, he combines his technical acumen and strategic leadership to help organizations meet data legislation requirements, reduce operational costs and scale more effectively.

What does Ilkari do?

Ilkari is a sovereign infrastructure provider delivering data centre services, sovereign cloud, and digital infrastructure solutions across Europe, and Latin America.

The company operates certified, high-performance infrastructure environments designed for operational resilience, regulatory alignment, and long-term digital sovereignty, supporting enterprise, regulated-industry, and data-intensive workloads.

Ilkari’s sovereign infrastructure platform spans data centre operations in Iceland, Croatia, and Colombia. Ilkari also operates a premium domain management and digital intelligence platform.

What problems does Ilkari solve in the market?

As a sovereign infrastructure provider delivering data centre services, sovereign cloud, and digital infrastructure solutions across Europe and Latin America, Ilkari helps companies solve the need for secure, compliant, regionally controlled digital infrastructure for enterprises and regulated industries.

What are Ilkari’s core products or services?

llkari operates across three closely connected areas:

1. Certified data centre and colocation services, including carrier-neutral connectivity and 24/7 onsite technical support.
2. Sovereign cloud infrastructure, delivered through shared and dedicated deployment models.
3. Corporate domain name services covering the registration, management and protection of digital identities globally.

Together, these services allow organizations to manage a broader part of their infrastructure through one accountable provider: from their domain portfolio and digital identity to the physical and cloud environments supporting their applications and data.

Our data centre footprint currently spans Colombia, Iceland, and Croatia. This gives customers access to infrastructure across Latin America and Europe, supported by local operational teams and international expertise.

What markets do you serve?

We primarily serve medium-sized and large organizations operating business-critical, regulated or data-intensive environments. These include financial services, government, and public-sector organizations, telcos, healthcare, and technology, as well as digital-first businesses and online platforms managing complex international domain portfolios.

Geographically, our infrastructure platform serves customers across Europe and Latin America, with data centre operations in Croatia, Iceland, and Colombia. Our domain name services have a broader international reach, supporting organizations in managing and protecting their or their customers’ digital identities across global markets.

We are particularly relevant to organizations that need to keep sensitive workloads within a defined jurisdiction, diversify infrastructure across regions or providers, establish business continuity outside a congested primary market, or reassess their dependence on hyperscale cloud and virtualisation vendors.

We also support European organizations seeking to grow or invest in Colombia, and Latin American businesses looking for infrastructure and expertise that can support their international expansion. Our combination of local teams with regional expertise and cross-regional operations allows us to serve as the link between those markets.

What challenges does the global digital infrastructure industry face today?

The industry is facing a collision between rapidly growing demand and increasingly constrained supply.

AI is accelerating infrastructure requirements, but it is arriving at a time when many established data-centre markets are already approaching capacity limits. Suitable land is scarce, grid connections can take years and power constraints are delaying or in some locations effectively freezing further development. According to the IEA, global data-centre electricity consumption increased by 17% in 2025 and is expected to double by 2030, with AI-focused facilities growing considerably faster. That makes access to reliable power one of the industry’s defining constraints.

This growth also has to be sustainable for the communities hosting it. Data centres compete with households and other industries for energy, land, and water. Operators therefore need to demonstrate tangible local value, invest in efficiency and renewable energy, and engage transparently with communities. The industry cannot assume that technical demand automatically creates permission to build.

A second challenge is infrastructure concentration. organizations have gained enormous flexibility from cloud, but many now depend heavily on a small number of hyperscalers, software vendors, and geographic regions. Recent changes in virtualisation licensing and pricing have made this dependency tangible: technology choices made several years ago can suddenly become major cost and operational constraints. For an IT leader, migrating hundreds of virtual machines, applications and data repositories is not a quick procurement decision. Lock-in reduces negotiating power and can make the path to an alternative complex and expensive.

Regulation is adding further pressure. Frameworks such as the EU Digital Operational Resilience Act (DORA) require organizations to look beyond basic compliance and consider operational resilience, ICT third-party risk and concentration across critical providers. These are no longer questions that can be delegated entirely to the IT department; they increasingly affect enterprise risk and board-level decision-making.

This is also why sovereignty is moving into mainstream infrastructure procurement. PwC found that 44% of end users now rank sovereignty and data residency as a leading buying priority; almost level with the 46% prioritising total cost of compute. That tells us customers are no longer evaluating infrastructure on price and performance alone. They also want to know where their workloads reside, which jurisdiction governs them, who can access them and whether they have a credible exit path if their requirements change.

The central challenge is therefore not simply building more capacity. It is building infrastructure that is powerful enough for the next generation of workloads, sustainable enough to retain community support and sufficiently open, resilient and accountable to give customers meaningful control.

How is Ilkari adapting to these challenges?

We are building a multi-region infrastructure platform around four principles: regional capacity with a focus on emerging markets, operational control, customer choice, and sustainability and community impact.

First, we are investing in locations that can offer credible alternatives to the most constrained digital infrastructure hubs. The integration of DC North in Croatia, for example, gives customers access to a fully operational 4 MW carrier-neutral data centre within the European Union, with direct connectivity to internet exchange points in Austria, Hungary, and Slovenia. It adds capacity in a strategically useful location while supporting infrastructure diversification across Central and Eastern Europe.

Second, we combine certified data centres and colocation with a shared and dedicated sovereign cloud. This gives customers more flexibility in deciding where individual workloads should run and how much isolation, control and support they require. A regulated workload may require dedicated infrastructure within a defined jurisdiction, while another application may be better suited to a shared environment. Sovereignty should provide practical choices rather than impose a single architecture.

Third, we recognise that many organizations want to reduce platform dependency but cannot simply redesign their entire technology estate. Our role is to provide a credible alternative and a realistic migration path, supported by local engineers and clear accountability. That is particularly relevant for businesses reassessing virtualisation costs, looking to repatriate selected workloads from hyperscale cloud, or seeking a secondary environment to strengthen resilience.

AI reinforces this strategy. We are not positioning Ilkari as an AI company or applying an “AI-ready” label indiscriminately. Our focus is on the foundations that AI and other data-intensive workloads require: dependable power, resilient facilities, strong connectivity, predictable performance and clear governance over the underlying data.

Fourth, we embed sustainability and community impact into our infrastructure decisions. In Colombia, our facilities operate on a national grid powered predominantly by renewable energy, with hydropower accounting for the majority of generation, and are certified under the SS 564 Green Data Centre standard. In Croatia, DC North achieves a partial PUE (pPUE) of ≤ 1.14, holds an A+ energy-efficiency certificate and is ISO 14001 certified for environmental management. In Iceland, our infrastructure is powered by 100% renewable energy from a mix of hydroelectric power, fed by glacial rivers and melting ice, and geothermal energy from volcanic heat. Across our offices, we run frequent ESG initiatives, including local tree-planting activities, giving back to nature and supporting local communities.

Ultimately, customers should not have to choose between modern infrastructure and meaningful control. Our objective is to give them more choice over where their workloads operate, how their data is governed and which dependencies they are prepared to accept.

What are your company’s key differentiators?

Our principal differentiator is that we approach sovereignty across the infrastructure stack rather than presenting it as a feature attached to one product.

We combine corporate domain name services, certified data centres, colocation, and sovereign cloud. This gives customers a clearer line of accountability, from their digital identity through to the physical and cloud infrastructure supporting their applications and data.

Our geographic strategy is also deliberate. We are not attempting to replicate the scale or footprint of a hyperscaler. We invest selectively in markets where our infrastructure can offer meaningful capacity, connectivity and jurisdictional choice. Croatia, Iceland, and Colombia each provide distinct strategic advantages across Europe and Latin America.

We also combine international expertise with local operational teams. Sovereignty cannot be delivered through contractual language alone. It depends on who operates the infrastructure, who has access to it, who responds when support is required, and whether the customer can reach accountable technical experts.

Finally, we provide a credible alternative for organizations seeking to reduce dependency on large cloud and virtualisation vendors. That does not mean every workload must leave the hyperscalers. It means customers should have genuine options, clearer economics, and a viable path to move or diversify workloads if their requirements change.

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

Our immediate priority is to deepen the capabilities of the platform we have already established.

In Europe, that means integrating DC North into Ilkari’s wider operations and developing its role as a resilient infrastructure and interconnection hub for Croatia and the wider Central and Eastern European region. We see a clear opportunity to support organizations seeking greater infrastructure diversity and jurisdictional choice within the EU.

Across the business, we will continue developing our sovereign cloud capabilities and the services surrounding them, particularly in areas such as workload migration, security, connectivity, resilience and infrastructure management.

We also expect AI to increase demand for infrastructure that combines scalable capacity with clear jurisdiction, predictable performance and strong data governance. Ilkari will continue developing its data centre and cloud capabilities to support data-intensive workloads in markets where organizations require greater control over how and where their data is processed.

Over time, you can expect Ilkari to establish a presence in additional markets where regulation, data growth and infrastructure concentration are creating demand for credible alternatives. Our expansion will be selective. The objective is not to accumulate locations, but to build a connected platform in which each market serves a clear operational and customer purpose.

What upcoming industry events will you be attending?

Our next major ICT event will be ANDICOM in Cartagena at the beginning of September. Ilkari will exhibit at the event, contribute to its conference programme, and participate as part of the European Union delegation, with the EU joining this year as guest region.

We will showcase our sovereign cloud and certified data centre capabilities while sharing a practical European perspective on sovereign infrastructure with Colombian businesses and government organizations. As a company operating certified data centres across Europe and Colombia, we also have a valuable role to play as a local infrastructure partner for European organizations looking to grow or invest in the Colombian market.

Before then, we will attend Asobancaria’s Banking Convention in Cartagena, which brings together senior leaders from Colombia’s banking and financial-services sector.

Following the integration of DC North, we are expanding our engagement with data centre, cloud and wider ICT communities in Croatia and across Central and Eastern Europe. This will form an important part of our industry programme as we strengthen Ilkari’s regional presence and develop DC North’s role within the European digital infrastructure and ICT ecosystem.

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

Our most significant recent development is the integration of the data centre DC North in Varaždin, Croatia, into Ilkari’s European operations.

DC North is a fully operational, 4 MW carrier-neutral data centre located approximately 70 kilometres north of Zagreb. It is Croatia’s largest carrier-neutral data centre by operational power capacity and provides connectivity to internet exchange points in Austria, Hungary, and Slovenia.

The integration expands Ilkari’s operational capacity within the European Union and extends our data centre footprint to three markets: Croatia, Iceland, and Colombia. It strengthens our ability to support regulated, sovereignty-focused, and data-intensive organizations across Europe, while providing an additional location for infrastructure diversification and business continuity.

It also creates a foundation for bringing together colocation, connectivity, and sovereign cloud capabilities within an EU jurisdiction.

Earlier this year, we also expanded our infrastructure platform with Ilkari Cloud. Delivered through shared and dedicated deployment models, Ilkari Cloud is designed to provide predictable performance, operational control, and clearer jurisdiction over customer data and workloads. It offers organizations an alternative as they reassess hyperscaler concentration, virtualisation costs and long-term platform dependencies.

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

Sovereignty is sometimes presented as though organizations must choose between control and innovation. We do not believe that should be the case.

Businesses need infrastructure that performs, scales, and supports technologies such as AI. But they also need to understand the dependencies they are creating and retain meaningful control over their data, providers and operational choices.

Ilkari was established to help close that gap. We are building an alternative for organizations that want modern infrastructure without losing sight of jurisdiction, accountability, or resilience.

We are still a relatively young company, but the foundations are tangible: operational data centres across three markets, sovereign cloud capabilities, an international domain services business, and local teams with substantial infrastructure experience.

Our ambition is straightforward: to give organizations greater control over the digital infrastructure on which they increasingly depend.

Where can our readers learn more about Ilkari?

Readers can learn more about us on our website, ilkari.tech/about. You can also contact us through the contact form on our website.

# # #

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: Ilkari’s Vision for a More Resilient, Open Future appeared first on Data Center POST.

TL;DR

• Evaluating cooling components (such as cold plates) in isolation can be misleading; thermal and hydraulic performance must be engineered holistically to prevent degrading overall system efficiency.
• As liquid cooling scales across thousands of servers, small flow resistance inefficiencies accumulate, increasing pump power demands and potentially limiting infrastructure scaling.
• Much like Formula One teams constantly refine thermal packages for each race, data center cooling designs must evolve quickly to support the rapid arrival of new GPU generations with distinct thermal profiles.
• Computational fluid dynamics (CFD) and additive manufacturing (3D printing) allow engineers to design complex, monolithic cold plates that optimize both heat transfer and pressure drop while eliminating potential leak paths.

# # #

As cooling efficiency is emerging as a key constraint in scaling AI data center infrastructure, Michael Fuller, Executive Chairman of Conflux Technology and former F1 engineer, explores how Formula One’s system-level approach to thermal management could help data centers balance cooling performance with energy efficiency and operational demands.

As AI infrastructure scales, most conversations around data center performance still focus on compute capacity: faster GPUs, denser racks and increasing power availability. However, a critical constraint is quietly emerging alongside these advances – cooling efficiency.

While the industry has largely acknowledged that liquid cooling is necessary to support next-generation hardware, less attention has been paid to how efficiently those systems operate at scale. That distinction – between simply cooling hardware and engineering an efficient thermal system – is one Formula One has been refining for decades.

Engineering That Never Stands Still

Perhaps Formula One’s greatest lesson isn’t simply how to engineer an efficient cooling system, but how to engineer one that can evolve. Every race weekend brings new components, revised aerodynamic packages and evolving thermal demands, leaving engineers with little time to redesign, test and validate solutions. That relentless cycle has fostered an engineering mindset built around rapid iteration, where cooling systems are expected to evolve alongside the hardware they support. AI data centers are now entering a similar phase. Successive generations of GPUs are arriving faster than ever, each with different thermal profiles, power densities and packaging constraints, meaning cooling solutions can no longer remain static while the technology around them advances.

From Thermal Capacity to System Efficiency

That evolution also changes the engineering challenge. It is no longer enough for cooling systems to remove the required amount of heat. They must do so with maximum efficiency.

AI workloads are pushing rack power densities to unprecedented levels, and simply meeting thermal requirements does not guarantee an optimal system. Instead, operators must consider how much energy is required to deliver that cooling capacity.

This shifts the focus from peak performance to system efficiency, a change that has significant implications for both operational cost and infrastructure design.

The Hidden Cost of Pressure Drop

One of the most important, and often overlooked, aspects of cooling performance is pressure drop.

In liquid cooling systems, pressure drop has a significant influence on the energy required to move coolant through the system. While small, perhaps even negligible, at the component level, its impact becomes significant when multiplied across thousands of servers.

Inefficient flow paths can increase pump power requirements, raise energy consumption and reduce overall system flexibility. In extreme cases, they can become a limiting factor in scaling infrastructure.

Why Component-Level Thinking is No Longer Enough

Components such as cold plates are often assessed in isolation, focusing on metrics like heat transfer performance and flow rate. However, as systems become more complex, this approach can be misleading. Formula One engineers have long understood that a component delivering impressive standalone performance can still compromise the performance of the car as a whole.

A component that delivers strong thermal performance but induces high hydraulic resistance may degrade overall system efficiency. The net result is higher energy use, even if individual components appear optimised. To address this, cooling must be treated as a system-level engineering challenge, where thermal and hydraulic performance are considered together rather than separately.

Balancing Competing Requirements

Effective liquid cooling requires managing a fundamental trade-off between improving heat transfer and maintaining low resistance to flow. Design approaches that increase turbulence or surface area can enhance thermal performance, but they often do so at the cost of higher pressure drop. The challenge lies in achieving both low thermal resistance and low hydraulic resistance at the same time.

This balance demands precision engineering at both macro and micro levels. The most effective solutions focus on directing coolant exactly where it is needed, rather than applying uniform cooling across an entire component.

At Conflux, for example, our latest single-phase cold plate designs achieve thermal resistance as low as 0.0076°C/W at 6 LPM across a 2,500 mm² footprint, while maintaining exceptionally low pressure drop – demonstrating that high thermal performance and hydraulic efficiency can be achieved together, rather than traded off against one another. It’s the same engineering philosophy that underpins Formula One: the best solution isn’t the one that excels in a single metric, but the one that delivers the greatest overall system performance.

The Formula One Engineering Mindset

One of Formula One’s defining characteristics is that engineers do not evaluate components in isolation. Every design decision is based on its impact on overall system performance. A component that performs exceptionally well in isolation can still reduce the overall performance of the car if it compromises weight, reliability or aerodynamic efficiency.

This principle is directly relevant to data centers, where cooling systems must balance competing objectives under tight constraints. This is why cooling design is increasingly moving beyond component optimisation towards system-level engineering, where every design decision is judged by its impact on overall performance.

Delivering that level of optimisation demands equally sophisticated design tools and manufacturing techniques.

The Role of Advanced Design and Manufacturing

Achieving high levels of efficiency in liquid cooling begins with how fluid flow is controlled within components.

Modern engineering approaches use simulation tools such as computational fluid dynamics to analyse and refine internal geometries. These tools make it possible to optimise coolant distribution, minimise losses and improve heat transfer where it matters most.

It’s an approach familiar to Formula One, where advanced simulation and rapid manufacturing allow engineers to continually refine components in pursuit of marginal gains.

In liquid cooling, those same principles are enabling a new generation of highly optimised cold plate designs.

Additive manufacturing makes this possible by enabling complex internal channels and geometries that would be difficult or impossible to produce using conventional manufacturing methods. This gives engineers far greater control over coolant flow, allowing thermal performance and pressure drop to be optimised together rather than traded against one another.

It also enables cold plates to be produced as single, monolithic components rather than assemblies of multiple parts. By eliminating bonded interfaces, potential leak paths are eliminated while improving structural integrity – a significant advantage in mission-critical data center environments, where reliability is just as important as performance.

Rethinking the Role of Cooling

Formula One has always been a sport of marginal gains, where countless small engineering improvements combine to create a meaningful competitive advantage. AI data centers are entering a similar era.

As processor power continues to increase, the question is no longer simply whether cooling systems can remove enough heat. It is whether they can do so with maximum efficiency, balancing thermal performance, pressure drop, energy consumption and reliability across the entire system.

As a result, the next phase of data center innovation may be defined not only by faster processors, but by smarter and more efficient ways of keeping them cool.

# # #

About the Author

Michael Fuller’s career was launched off the back of his boyhood dream of designing racing cars. In 1999 he migrated to Europe as a Design Engineer, and over 15 years he held various design engineering and management roles in motorsport organisations competing in series as varied as the World Endurance Championship, World Rally Championship, and the Formula 1 World Championship. As an early adapter of Additive Manufacturing (AM) in the motorsport industry he was acutely aware of the opportunities available to enhance performance.

After returning to Australia, he pursued a design idea on an AM heat exchanger that proved to have ground-breaking performance potential and launched Conflux soon thereafter. Conflux has now grown into a world leading organisation that combines specialisations in heat transfer and additive manufacturing.

He is actively involved in all aspects of the business with significant effort in design engineering, relationship building, partnership research projects and industry speaking engagements. A considerable amount of his time is invested in strategy development and ensuring the right resources are in place to fulfil our vision.

The post Why AI Data Centers Should Think More Like Formula One Teams appeared first on Data Center POST.

TL;DR

• Data centers are so vital to modern life that the UK has granted them critical national infrastructure status, elevating data and energy security to matters of national security.
• While AI-driven energy demand is rising rapidly, the real hurdle is managing highly unpredictable, variable daily load patterns rather than just generating more total power.
• Resilience requires a flexible electricity system capable of balancing supply and demand in real time using solutions like energy storage, dispatchable generation, smart energy management, and workload shifting.
• Energy flexibility has become a business differentiator with companies like Google and Meta diversifying their power sources, while grid trials and rapid battery storage growth show real momentum.

# # #

Data Centers are so essential to modern life that, in 2024, the UK granted them the same critical national infrastructure status as water, energy and emergency services.

Protecting data infrastructure relies in part on protecting power and maintaining energy security. This has exposed a new reality: as data and energy become strategic dependencies, both are matters of national security.

As further proof of this, data centers and critical energy sites are known to be targeted in times of international conflict; such actions can exacerbate ongoing supply disruption and price volatility.

Against this backdrop, it’s clear that systems built on concentrated energy sources, such as an overreliance on one fuel type, can leave countries and operators exposed.

At the same time, it’s impossible to ignore the speed at which AI-driven energy demand is climbing. Data Centers are already consuming 6% of electricity in the UK and US, and that growth is not anticipated to slow.

However, because AI workloads are unevenly distributed and highly variable, data centers can have unpredictable fluctuations in power needs, surging during processing, dipping when they pause to share data or sync, and then repeating that cycle when they start the next batch. While daily averages might look consistent, demand throughout the day can fluctuate by hundreds of megawatts every few minutes.

In an environment of rising demand, unpredictable and dynamic load patterns, and geopolitical risk, simply adding more generation will not be enough. The real challenge is flexibility.

Flexibility will define the next era of energy

There is no single technology that can deliver the secure supply that AI will require. Resilience comes from combining dispatchable generation, renewable energy sources, storage, and smart energy management into a system flexible enough to respond to growing demand while guarding against volatility and disruption.

A data center’s ability to consistently access energy depends on the electricity system’s ability to adapt to changing conditions in real time, balancing supply and demand while maintaining a steady flow of power. Generation provides the power. Flexibility determines how effectively that power can be delivered and used, acting as a shock absorber against swings in supply and demand.

Historically, electricity systems have been designed around relatively predictable demand patterns.

Today, AI is introducing a new kind of demand. This dynamic is the reason flexibility will become one of the defining aims for the energy industry in years to come.

Flexibility as a business differentiator

Technology companies like Google and Meta are already diversifying to ensure energy security for their facilities. Google, for example, is purchasing nuclear power from Kairos Power’s small modular reactors to serve its AI facilities, while Meta has introduced a space-solar partnership with Overview Energy.

These investments demonstrate how energy resilience and flexibility are emerging as commercial differentiators for operators. Reliable access to power can be the difference between a data center’s success or failure. It allows organizations to manage risk, control costs, and scale more confidently. Flexibility can come from both the supply side (storage, dispatchable generation, and intermittent renewable generation) and the demand side (smart energy management, shifting workloads).

• Energy Storage: Captures electricity when supply is abundant to be used later on when demand is high. Solutions include battery energy storage systems and pumped storage hydro.
• Dispatchable Generation: Can ramp up quickly to generate electricity when needed. Examples include biomass and natural gas power plants.
• Smart Energy Management: Using advanced software and AI, virtual power plants and other solutions can optimise cooling systems, server utilisation, and power consumption in real-time based on grid conditions.
• Shifting Workloads: Moves non-urgent workloads to off-peak periods or physical locations to reduce stress on energy systems when demand is higher.

Together, these approaches can bolster energy systems and enable them to better respond to AI’s increasingly dynamic demand, while insulating against volatility and price swings.

Where We Stand Today

The good news is that there are signs of meaningful progress. In the UK, battery storage capacity surpassed nuclear generation for the first time, leaping from just under 10 MW to almost 7 GW in less than a decade. Elsewhere, earlier this year Emerald AI and the National Grid joined with other partners to complete a trial which proved that, through smart software, AI facilities can adjust electricity consumption in response to real-time demand signals without impacting critical operations.

Data centers need more of these solutions quickly because the cost of getting system flexibility wrong can be high, including rising costs, renewable curtailment, and reduced energy security. Meanwhile, as the geopolitical shocks that rocked energy markets have shown, reverting to and relying on the cheapest or fastest deployed single source also carries great risk.

AI’s energy demand is often framed as a race for more megawatts, but that’s an oversimplification. AI’s growth will depend just as much on the electricity system’s ability to adapt and continue delivering as it will on the megawatts.

To build a secure energy system capable of supporting the next generation of AI infrastructure, the UK must invest in flexible technologies that can meet the dynamic needs of future energy demand.

# # #

About the Author

Bio: Paul Sheffield is Chief Commercial Officer at Drax, where he leads the group’s biomass sales, trading and optimisation, and customer businesses. He joined Drax as Chief Operating Officer of Haven Power in 2017, bringing more than 20 years of experience in energy trading and retail, and was appointed to Drax’s Executive Committee in 2019.

The post The Next Big Challenge for Data Centers Isn’t More Power appeared first on Data Center POST.

TL;DR

• Selecting a data center UPS system requires evaluating load capacity, runtime, battery technology, serviceability, scalability, and long-term lifecycle costs rather than simply comparing initial purchase prices.
• Organizations should maintain a 20-30% spare capacity to support future growth, with a runtime of 5-15 minutes being sufficient to bridge the gap between utility failure and generator startup.
• Utilizing modular UPS configurations simplifies maintenance and reduces downtime risks, while focusing on the total cost of ownership (TCO) ensures optimal long-term reliability and efficiency.

# # #

Over the years, with the rapid advent of digital transformation across different industries, businesses’ data centers have been under pressure to deliver uninterrupted power, support growing workloads, and maintain operational efficiency. Reliable power protection remains a critical component of organizational continuity, whether supporting cloud services, enterprise applications, colocation environments, or AI-driven workloads.

The most important protection strategy is an Uninterruptible Power Supply (UPS) system, and choosing the right one is more than just comparing purchase prices. It is essential to evaluate UPS options for a data center based on load capacity, runtime, battery technology, serviceability, and lifecycle cost. A UPS must be evaluated based on its ability to support current and future load requirements, provide adequate runtime during outages, reduce maintenance needs, and deliver long-term value throughout its operational life.

When assessing UPS solutions, industry experts recommend a holistic approach and suggest organizations include the Uptime Institute, IEEE, and leading data center infrastructure providers.

Why Is UPS Selection Critical for Modern Data Centers?

For businesses, costly downtime, data loss and business interruptions can be highly affected by power disruptions. Against utility failures and power quality issues, UPS systems serve as the first line of defence by buffering and delivering clean, consistent power to protect the system from fluctuations. Apart from this, modern UPS solutions contribute towards:

• Availability and improved uptime
• Enhancement of power quality
• Better energy efficiency
• Increased operational stability
• Support for future growth of infrastructure

How Should Data Centers Evaluate Load Capacity?

The UPS system you choose should support the current load and future growth, and understanding load capacity is one of the most important factors to consider. Maintaining an approximately 20-30% spare capacity is often recommended to accommodate infrastructure expansion and avoid operating the UPS continuously at maximum capacity.

Here are some of the key considerations:

• Requirements of current IT power
• Growth of future workload
• Additions of planned equipment
• Target capacity utilisation
• Strategies of business expansion

How Much Runtime Does a Data Center Really Need?

During a power outage, the runtime refers to the amount of time a UPS can continue supplying power. Based on the business continuity strategy, the right runtime largely depends on it. While a UPS was never designed to provide power for hours, it bridges the gap between utility power loss and generator setup and a runtime of 5-15 minutes is considered sufficient.

While understanding the runtime requirements, it is important to consider the following:

• Availability of generator
• Plans of disaster recovery
• Requirements of critical applications
• Continuity of business
• Acceptable downtime

Which Battery Technology Is Best?

Reliability, maintenance requirements, space utilisation, and cost of ownership are directly affected by the battery technology decisions:

Valve-Regulated Lead Acid (VRLA) Batteries

Being a standard choice for data center UPS systems, VRLA batteries offer advantages like:

• Low acquisition costs
• Available within a broad market
• Maintenance-free operation
• Proven reliability
• Instantaneous response

Some limitations that should be considered:

• Shorter lifespan, typically 3-5 years
• High temperature sensitivity
• Larger physical footprint
• High operational costs

Quick Facts on Lithium-Ion Batteries

In modern data centers, lithium-ion batteries are becoming increasingly popular for good reason; here’s why:

• Extended lifespan, generally 10-15 years
• Space-saving footprint
• Quick recharge capabilities
• Less hands-on upkeep
• Operate safely in higher temperatures

Why Do Serviceability and Scalability Matter?

The ease of maintaining and expanding a UPS system directly influences its uptime and operational flexibility. Traditional systems can pose challenges, as maintenance activities require planned downtime or complex arrangements.

Modular systems provide a scalable approach to aligning infrastructure investments with business needs for organizations experiencing rapid growth or evolving workload demands.

Having modular UPS architectures offers multiple benefits, such as:

• Maintenance procedures are simple
• Fault isolation is improved
• Downtime risks are reduced
• Capacity expansion is easier

How Should Organizations Evaluate Lifecycle Cost?

The lifecycle cost analysis of organizations must include:

• Investment in initial equipment
• Expenses on battery replacement
• Performance of energy efficiency
• Maintenance costs preventions
• Requirements for cooling
• Agreements of service
• Risks involved in downtime
• Costs of equipment disposal

Final Thoughts

It is concluded that choosing a UPS system for a data center involves more than just upfront costs; it requires evaluating load capacity, runtime requirements, battery technology serviceability, scalability, and lifecycle costs. This ensures that the chosen solution aligns with business objectives and operational requirements.

Organizations that focus on total cost of ownership rather than acquisition cost alone are often better positioned to achieve long-term reliability, efficiency, and scalability. Industry leaders such as Schneider Electric continue to advance UPS innovation through high-efficiency power protection systems, modular architectures, and lithium-ion battery integration, helping data centers improve resilience while supporting future growth. Ultimately, the right UPS balances current needs with long-term operational and financial goals.

# # #

About the Author

Anuj Kudesia, MD-SEPSL and Director of Business Development at Schneider Electric, leads strategy, product launches, and revenue operations for the Secure Power Division across the Greater India Zone. With over 20 years at Schneider Electric spanning enterprise data center sales, channel management, and strategic business development including P&L ownership for BFSI, IT/ITES, and Government verticals, he brings deep expertise in UPS systems, critical power infrastructure, edge computing, and data center modernisation for Indian enterprises.

The post Choosing the Right UPS for Your Data Center: Load, Runtime, Battery Type, and Cost Compared appeared first on Data Center POST.

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.