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

In the age of AIHuman resources are now facing a new, unexpected problem: Phantom data centers. On the surface, it may seem like a no-brainer: Why (and how) would someone build something as complex as a data center? But as AI seeks to skyrocket and the demand for more power increases, the ideas surrounding data center development are causing confusion, especially in areas like Northern Virginia, global data center. In the changing environment, the tools used are being shaped by the strong demand from the developers of the environment that may or may not exist. actually building the foundation they say.

Virtual data centers represent an urgent problem in expanding data infrastructure to meet computing needs. This upcoming event is preventing capital from moving where it is needed. Any business that can help solve this problem – maybe using AI solving a problem created by AI – will have limitations.

More gigawatts are needed

Dominion Energy, Northern Virginia’s largest utility, has received multiple requests 50 gigawatts of power from the data center project. That’s more energy than Iceland consumes in a year.

But most of these claims are fictitious or false. Developers are looking for potential locations and staking their claim to power long before they have capital or any ideas about how to land it. In fact, estimates show that about 90 percent of these requests are false.

In the early days of the data center boom, utilities didn’t worry about false demand. Companies like Amazon, Google and Microsoft – known as a “hyperscaler” because it uses a data center with thousands of servers – sends requests for direct power, and resources are automatically provided. But now, the frenzy to find power has led to a flood of requests from reputable manufacturers or speculators with dubious reputations. Utilities, which usually work with only a few power-hungry customers, are suddenly flooded with orders for power that will slow down their entire grid.

Resources struggle to transform reality from fiction

The real estate problem isn’t just about technology – it exists. They are responsible for knowing what is real and what is not. And they are not well equipped to do this. In the past, the tools used have been slow, risk-averse organizations. Now they are being asked to watch out for speculators, many of whom are playing the real estate game, hoping to flip their shares when the market is hot.

Resources have groups that are responsible for economic development, but these groups are not used to handle many speculative requests at the same time. It’s like a land rush, where only a few of those complaining plan to create something tangible. What’s next? Paralysis. Agents hesitate to allocate power if they don’t know which tasks will be done, and limit it development cycle.

A large wall

There is no shortage of money flowing into the data center space, but the amount is part of the problem. When capital is easy to find, it makes sense. In a way, this is similar to the mousetrap problem: Too many players chasing too much market. This proliferation of predictors leads to misjudgment not only in business, but also in communities that have to decide whether to grant land use and development permits.

Adding to the challenge is that the data center is not just for AI. Of course, AI is driving increased demand, but there is also an ongoing demand for cloud computing. Developers are building data centers to accommodate both, but distinguishing between the two can be difficult, especially when projects are combined. AI hype and traditional cloud infrastructure.

What is the truth?

Official players – the aforementioned Apples, Googles and Microsofts – are building real data centers, and many are taking steps like “behind the meter” dealing with renewable energy providers or creating microgrids to avoid grid connection constraints. But as real jobs proliferate, so do fake ones. Developers with limited experience in the field are trying to raise money, which leads to a frenzy of demand.

The problem is not a financial problem – although the capital required to build a single gigawatt campus can exceed several billion dollars – it is very difficult to build infrastructure on this scale. The 6-gigawatt campus looks exciting, but economic and engineering realities make it impossible to build in time. However, speculators toss these big numbers around, hoping they’ll gain momentum by thinking the process will change later.

Why this group can’t meet the demands of the data center

When utility struggles to distinguish fact from fiction, the grid becomes a hindrance. McKinsey recently said that the demand for data centers around the world could reach 152 gigawatts by 2030and add 250 terawatt-hours of new electricity demand. In the US, only data centers can count 8% of total energy demand by 2030a very impressive number considering the amount of demand that has grown in the last two decades.

However, the group was not prepared for this increase. Communications and broadcasting have grown, and it is estimated that the US could run out of energy by 2027 to 2029 if alternatives are not found. Developers are turning to on-site generation like gas turbines or microgrids to avoid gridlock, but these shutdowns only highlight the grid’s limitations.

Conclusion: Useful as a guard

The real obstacle is not a lack of money (trust me, there is a lot of money here) or technology – it is the lack of resources to work as security guards, to know who is real and who is just playing an imaginary game. Without a strong vetting process, the team runs the risk of being overwhelmed by projects that never get done. The age of the fake data center is upon us, and until the tools change completely, all companies may struggle to keep up with the real demands.

In this chaotic environment, it is not just about the distribution of power; it’s about tools to help learn how to navigate new, creative frontiers for business (and AI) success.

Sophie Bakalar is a partner at Collaborative Fund.