Phantom data centers: What they are (or aren’t) and why they’re hindering the true promise of AI

In the age of AI, utilities now face a new, unexpected problem: phantom data centers. On the surface, this may seem absurd. Why (and how) would anyone build something as complex as a data center? But as AI demand surges along with the need for more computing power, speculation about data center development is causing confusion, especially in regions like Northern Virginia, the data center capital of the world. In this evolving environment, utilities are being bombarded with power requests from real estate developers. actually Build the infrastructure they advocate for.

Fake data centers represent an urgent bottleneck in scaling data infrastructure to keep up with computing demand. This new phenomenon is preventing capital from flowing to where it is really needed. Companies that can help solve this problem, including leveraging AI to solve problems created by AI, will have a significant advantage.

The mirage of gigawatt demand

Dominion Energy, Northern Virginia’s largest utility, has received requests for a total of 50 gigawatts of power from data center projects. This is more electricity than Iceland consumes in a year.

However, many of these requests are speculative or outright false. Developers are eyeing potential sites and staking claims on power capacity long before they have capital or a strategy for how to break ground. In fact, estimates show that up to 90% of these requests are completely fake.

In the early days of the data center boom, utilities didn’t have to worry about spurious demand. Companies like Amazon, Google, and Microsoft (called “hyperscalers” because they run data centers with hundreds of thousands of servers) submitted simple power requests, and utilities were simply provided. But now the frenzy to secure power capacity is bringing in a flood of requests from unknown developers or speculators with questionable track records. Utilities that traditionally serve only a small number of power-hungry customers are suddenly inundated with orders for power capacity that overwhelms the entire grid.

Utilities have a hard time sorting fact from fiction.

The challenges for utilities are not only technical but also existential. They are tasked with determining what is real and what is not. And they are ill-equipped to handle this. Historically, utilities have been slow-moving and risk-averse institutions. Now they are being asked to investigate speculators, many of whom are simply playing the real estate game, hoping to reverse their allocation of power when the market heats up.

Utilities have groups responsible for economic development, but these teams are not accustomed to handling dozens of speculative requests at once. This is similar to a land rush, where only some of those claiming a stake actually plan to build something tangible. What are the results? paralysis. Utilities are hesitant to allocate power because they don’t know which projects will come to fruition, slowing down the overall development cycle.

wall of capital

There is no shortage of capital flowing into the data center space, but its abundance is part of the problem. Easy access to capital leads to speculation. In some ways, this is similar to the better mousetrap problem. There are too many players chasing an oversupplied market. This influx of speculators creates indecision not only within utilities, but also in communities that must decide whether to grant permits for land use and infrastructure development.

To further complicate things, data centers aren’t just for AI. Of course, AI is driving the surge in demand, but there is also a continued need for cloud computing. Developers are building data centers to accommodate both, but distinguishing between the two is becoming increasingly difficult, especially when projects mix AI hype with traditional cloud infrastructure.

What is real?

Legitimate companies like the aforementioned Apple, Google, and Microsoft are building true data centers, and many are adopting strategies such as “secret deals” with renewable energy suppliers or avoiding bottlenecks in grid interconnection. We are building a microgrid for this purpose. However, as the number of real projects increases, so do the number of fake projects. The utility landscape is becoming increasingly chaotic as developers with little experience in the field are trying to make money.

The problem isn’t just financial risk. The capital required to build a single gigawatt campus can easily exceed billions of dollars, but the complexities of developing infrastructure of this scale are daunting. A 6-gigawatt campus sounds impressive, but financial and engineering realities make it nearly impossible to build in a reasonable time frame. But speculators are throwing these huge numbers around, hoping to free up power capacity to flip the project later.

Why Grid Can’t Keep Up with Data Center Demand

The grid itself becomes a bottleneck as utilities struggle to sort fact from fiction. McKinsey recently estimated that global data center demand could reach up to 152 gigawatts by 2030, adding 250 terawatt hours of new power demand. In the United States, data centers alone could account for 8% of total electricity demand by 2030, a surprising figure considering how little demand has grown over the past two decades.

But the grid is not ready to handle this influx. Interconnection and transmission problems are widespread, and it is estimated that America’s electricity capacity could be depleted by 2027-2029 if alternative solutions are not found. Developers are increasingly turning to on-site power generation, such as gas turbines or microgrids, to avoid interconnection bottlenecks, but these stopgap measures only serve to highlight the grid’s limitations.

Conclusion: Utility as Gatekeeper

The real bottleneck is not lack of capital (trust me, there is a lot of it here) or technology. This is the utility’s ability to act as a gatekeeper, determining who is real and who is playing a speculative game. Without a robust process for vetting developers, there is a risk of the grid becoming overwhelmed with projects that will never come to fruition. The era of fake data centers has arrived. Until utilities adapt, the entire industry may struggle to keep up with actual demand.

In this chaotic environment, power allocation isn’t the only thing that matters. It’s about utilities learning how to navigate new speculative territories so that businesses (and AI) can thrive.

Sophie Bakalar is a partner at the Collaborative Fund.