Data Centers Are No Longer a Real Estate Story—They’re a Power Story

For years, commercial real estate professionals looked at data centers through a familiar lens: land, zoning, fiber connectivity, tax incentives, tenant credit, construction cost, and exit cap rates. Those still matter. But in 2026, they are no longer the first question.

The first question is simpler: Can you get enough power, fast enough, at a price that makes the project work?

That shift is changing how data centers are sited, financed, leased, and valued. In many markets, data center demand is not being limited by lack of capital, lack of tenants, or lack of land. It is being limited by electricity.

The AI boom changed the site-selection equation:

The traditional data center site-selection model prioritized proximity to users, fiber networks, tax incentives, land availability, and disaster-risk mitigation. Power was always important, but it was one factor among many.

AI has moved power from a checklist item to the central underwriting assumption.

JLL projects that nearly 100 gigawatts of new data center capacity will be added globally between 2026 and 2030, effectively doubling global capacity. The firm also estimates that this expansion could require up to $3 trillion in total spending when real estate, debt, and tenant fit-out costs are included.

That scale is enormous, but the more important point is not simply that more data centers are being built. It is that the next generation of data centers is different. AI workloads require far more power density than traditional enterprise cloud computing. Training large models is energy-intensive, and as AI adoption shifts from training to everyday inference — the act of running AI models for users — demand will become more distributed and persistent.

The International Energy Agency projects global data center electricity consumption will roughly double from 485 terawatt-hours in 2025 to 950 terawatt-hours in 2030, with AI-focused data center consumption growing even faster. ([IEA][2]) In the U.S., data center electricity consumption is projected to increase by about 240 terawatt-hours by 2030, a 130% increase from 2024 levels.

Those numbers help explain why the industry’s center of gravity has shifted. Data centers are still real estate assets, but the scarce resource is no longer just a well-located site. It is deliverable power.

“Speed to power” is the new location:

In most CRE sectors, location still defines the opportunity. For office, it may be transit access and talent. For retail, traffic counts and income density. For industrial, highway access and labor.

For data centers, the equivalent metric is now speed to power.

CBRE says power cost and delivery speed now outweigh connectivity in data center site selection. The firm notes that the ability to secure 300-megawatt-plus power deliveries in under 36 months can matter more than traditional connectivity advantages. It also warns that large AI campuses requiring 500 megawatts or more can push construction and interconnection timelines into multi-year territory, especially where new transmission or generation is needed.

That is a fundamental change for investors and developers. A parcel with strong zoning, cheap land, and nearby fiber may still be unusable if the local utility cannot deliver the load. Meanwhile, a secondary or frontier market with abundant energy resources may suddenly become more attractive than a historically dominant data center hub.

This is already redrawing the map. JLL reports that 64% of data center capacity under construction in North America is now located in frontier markets, with West Texas, Tennessee, Wisconsin, and Ohio among the beneficiaries. The firm also notes that Texas could overtake Northern Virginia as the world’s largest data center market by 2030 if current trends continue. ([JLL][5])

In other words, the next hot data center market may not be the one with the most established tech ecosystem. It may be the one with the most credible path to megawatts.

Demand is not the problem. Delivery is:

In a normal real estate cycle, a surge in development raises concerns about oversupply. Data centers are not immune to that risk, but the current bottleneck is different.

CBRE reported that the overall vacancy rate in primary North American data center markets fell to a record-low 1.4% at year-end 2025, while primary-market supply increased 36% year over year to 9,432 megawatts. Net absorption also reached a record 2,497.6 megawatts in 2025.

JLL’s year-end 2025 North America report tells a similar story: vacancy remained at 1% for a second consecutive year, and 92% of capacity under construction was precommitted through either binding leases or owner-occupied development.

Those figures suggest that, at least for now, demand is not the main weakness in the story. The harder question is whether developers can deliver enough powered capacity at the right time and cost.

That is where the power constraint becomes an investment constraint. A data center lease may be valuable, but if utility upgrades, substations, transformers, switchgear, transmission lines, or generation capacity are delayed, the economics can deteriorate quickly. Time is not neutral in data center development. Every delay can mean higher construction costs, missed tenant deadlines, and a changing capital-market environment.

The rise of “bring your own power”:

As utility interconnection queues lengthen, developers and operators are looking beyond the traditional model of buying land and waiting for the grid.

JLL says the average wait time for a grid connection in primary data center markets now exceeds four years, pushing operators toward behind-the-meter power arrangements and colocated battery storage. The firm also notes that some markets have adopted “bring your own power” approaches, forcing developers to solve part of the energy equation themselves rather than relying entirely on existing grid capacity.

This is creating a new convergence between real estate, energy infrastructure, and capital markets. Data center developers increasingly need expertise in power procurement, utility negotiation, power purchase agreements, on-site generation, battery storage, gas availability, renewable integration, and grid interconnection.

That does not mean every data center will become its own power plant. But it does mean that energy strategy is now inseparable from real estate strategy.

For CRE investors, this changes due diligence. A site’s value cannot be understood only by looking at acreage, entitlement status, fiber access, and tenant demand. Investors need to understand the local power market, the utility’s actual ability to serve the site, interconnection timelines, substation requirements, transmission constraints, and potential community opposition.

A “good site” without power is no longer a good site. It is a land-banking bet.

Communities are starting to push back:

The power story is also becoming a political story.

Data centers can generate tax revenue, construction jobs, and long-term investment, but they also raise local concerns about electricity demand, water use, noise, diesel backup generators, land use, and utility bills. As projects get larger, those concerns are becoming harder for local governments to ignore.

In Coachella, California, hundreds of residents recently protested a proposed data center campus, and city council members signaled support for a temporary halt while the city studies the issue further. Planning documents showed the project could include six data centers across up to 450 acres, with the initial phase including three buildings of roughly 1 million square feet each and 90 megawatts of capacity per building., per Business Insider.

In Albany, New York, a proposed mixed-use development with a data center component that could theoretically require up to 180 megawatts of electricity faced zoning hurdles, environmental review questions, and public concern. Local officials said the approval process would be lengthy and would require multiple legislative and planning steps. ([Times Union][8])

These are not isolated debates. They are early signs of a larger negotiation between digital infrastructure demand and local resource capacity.

For developers, community support is becoming part of power strategy. A project may have a utility pathway on paper, but if residents believe it will strain the grid, raise bills, consume water, or provide limited local benefit, political risk can slow or stop the project.

Even chipmakers are responding to the power wall:

The power constraint is not limited to real estate. It is influencing the technology stack itself.

TSMC, the world’s largest contract chipmaker, recently said energy efficiency is becoming a top priority for customers as AI power demands rise. A senior TSMC executive said customers are increasingly looking for performance improvements that do not drive up electricity use, especially in high-performance AI data centers.

That matters for CRE because the future data center footprint depends not only on how much AI demand grows, but also on how efficient chips, cooling systems, networking equipment, and software become. Better efficiency could ease some pressure on the grid. But it could also make AI cheaper and more widely used, increasing total demand. The net effect is uncertain.

What is clear is that power is now shaping decisions from the semiconductor lab to the zoning board.

What this means for CRE investors:

For commercial real estate investors, the data center boom remains one of the most compelling growth stories in the market. But it is not a simple “buy land near fiber and wait” thesis.

The winning projects will likely share several traits: secured power, credible utility timelines, experienced operators, strong tenant commitments, realistic construction budgets, and a community-relations strategy that begins before entitlement hearings. The losing projects may be those that underwrite tenant demand correctly but underestimate the complexity of electricity delivery.

This also changes how investors should think about markets. Historically dominant data center hubs will remain important, but the next phase of growth may favor places with available power, supportive utilities, energy infrastructure, land, and political willingness. That opens opportunities in secondary and frontier markets, but it also increases execution risk.

Data centers still sit on land. They still require leases, financing, construction, and asset management. But the asset class is no longer defined primarily by the real estate beneath it.

It is defined by the power behind it.

The bottom line:

The data center boom is not just a CRE cycle. It is an infrastructure cycle.

AI has transformed the data center from a specialized property type into a frontline asset in the race for electricity, compute capacity, and digital dominance. Developers who understand only real estate will be at a disadvantage. So will energy players who underestimate entitlement, community, and capital-market risk.

The next generation of winners will be those who can integrate both worlds.

In 2026, the most valuable data center sites are not simply the biggest, cheapest, or best-located. They are the ones that can answer the question every tenant, lender, investor, and local official is now asking:

Where is the power coming from?