In the evolving age of AI, public utilities are encountering a novel and unexpected issue: phantom data centers. While it might initially sound preposterous, fabricating something as intricate as a data center, the surge in AI-driven demand and the need for enormous computing power have led to a speculative frenzy surrounding data center development. This is particularly evident in regions like Northern Virginia, the data center capital of the world, creating turmoil for utilities being inundated with power requests from real estate developers who may not have any genuine plans to construct the infrastructures they claim.
The Emergence and Impact of Fake Data Centers
Phantom data centers represent a significant bottleneck in scaling the data infrastructure necessary to meet computing demands. This phenomenon is creating a misallocation of capital, which hampers the effective deployment of resources where they are genuinely needed. If any enterprise can address this issue—potentially by leveraging AI to counteract a problem exacerbated by AI—it would have a considerable competitive advantage.
Dominion Energy, Northern Virginia’s largest utility, is at the forefront of this challenge. The utility has received aggregated requests for a staggering 50 gigawatts of power from prospective data center projects, an amount exceeding Iceland’s yearly power consumption. However, a significant portion of these requests—estimates suggest up to 90%—are speculative or outright fabrications. Developers often eye potential sites and stake their claims for power capacity without having the necessary capital or any concrete plans to initiate construction.
Traditionally, utilities did not have to contend with fake demand. Established hyperscalers like Amazon, Google, and Microsoft, who operate data centers with extensive server farms, submitted straightforward power requests, and utilities obliged. But now, as competition to secure power capacity intensifies, utilities are being flooded with inquiries from lesser-known developers or speculators with dubious histories. Utilities, which typically handled requests from a limited number of large, power-intensive customers, are now overwhelmed by claims for power capacity that could potentially exceed the capacity of their entire grid.
Utilities’ Struggle to Differentiate Between Legitimate and False Claims
The problem utilities face isn’t merely technical; it’s existential. They are now tasked with discerning between genuine and fraudulent claims, a capability they are ill-equipped to handle. Historically, utilities are slow-moving entities inclined towards risk aversion. Now, they must scrutinize speculators, many of whom are likely engaging in real estate speculation, hoping to resell their power entitlements when the market peaks.
Utilities have teams dedicated to economic development, but these teams are not accustomed to handling a large volume of speculative requests simultaneously. This scenario is akin to a land rush where only a few stake claimants actually have intentions to build something substantial. Consequently, utilities often hesitate to allocate power, unsure of which projects will materialize, leading to delays in the development cycle.
One complicating factor in dealing with this issue is the abundance of capital flowing into the data center space, which ironically exacerbates the problem. Easy access to capital leads to speculation, comparable to the better mousetrap dilemma: too many players chasing an oversupplied market. This scenario induces hesitation not only within utilities but also among local communities that need to decide on land use permits and infrastructure development.
The Challenge of Abundant Capital and Speculation
Further complexity arises because data centers aren’t exclusively for AI; there’s a high, ongoing demand for cloud computing as well. Developers are constructing data centers to serve both needs, making it challenging to distinguish between AI-focused and traditional cloud infrastructure projects. This blending of AI hype with conventional cloud infrastructure projects obscures the developers’ true intentions, complicating the utilities’ decision-making processes.
Legitimate players in the space, such as Apple, Google, and Microsoft, continue to build genuine data centers. Many of these companies are adopting innovative strategies like “behind-the-meter” deals with renewable energy providers or constructing microgrids to bypass grid interconnection bottlenecks. Despite the real projects, speculative developers abound, drawn by the potential financial gains, leading to an increasingly chaotic environment for utilities.
This problem extends beyond financial risk, which is significant—building a single gigawatt-scale campus can cost several billion dollars. The challenge encompasses the complexity of developing infrastructure on such a massive scale. A proposed 6-gigawatt campus appears impressive on paper, but the financial and engineering demands make it nearly unfeasible to construct within reasonable time frames. Speculators, nonetheless, persist in promoting these large-scale projects, expecting to secure power capacity to flip the project for profit later.
The Grid’s Struggle with Increasing Data Center Demands
As utilities grapple with sorting out genuine from fictitious demands, the power grid itself becomes a significant bottleneck. According to McKinsey, global data center demand could soar to 152 gigawatts by 2030, necessitating the addition of 250 terawatt-hours of new electricity. In the U.S. alone, data centers could account for 8% of total power demand by 2030—a remarkable increase given the relatively stagnant growth in overall power demand over the previous two decades.
Despite this burgeoning demand, the power grid is woefully unprepared. Widespread interconnection and transmission challenges persist, with estimates indicating the U.S. may face power capacity shortages between 2027 and 2029 unless alternative solutions are implemented. Developers are increasingly relying on on-site power generation solutions like gas turbines or microgrids to circumvent grid interconnection issues, but these measures only underscore the grid’s inadequacies.
Conclusion: The Need for Utilities to Act as Effective Gatekeepers
In the advancing era of AI, public utilities are facing a unique and unforeseen challenge: phantom data centers. While at first, this might seem absurd—how can one fabricate something as complex as a data center?—the surge in AI-driven demand and the necessity for massive computing power have incited a speculative boom in data center development. This issue is particularly pronounced in regions like Northern Virginia, recognized as the data center capital of the world. Utilities there are experiencing chaos as they receive a deluge of power requests from real estate developers who may have no real intentions of building the data centers they propose. This speculative fervor has left utilities scrambling to separate genuine projects from fictitious ones, straining resources, and complicating planning processes. These phantom data center requests not only waste resources but also create a false sense of demand, making it difficult for utilities to allocate their capabilities effectively and plan for future power needs accurately.