The United States is in the middle of a rapid buildout of data centers meant to power artificial intelligence workloads. Utilities and grid operators have been fielding requests for enormous amounts of new electricity capacity. Much of that demand is tied to AI training and inference infrastructure planned by technology companies and their partners.
According to a report from CryptoBriefing, some of that projected demand may not be as solid as it appears. The report points to what it calls a phantom project problem. Some announced data center developments may be speculative, duplicated across multiple sites, or unlikely to reach construction at all.
This matters because utilities use these announcements to plan grid upgrades years in advance. Building new transmission lines, substations, and power plants takes significant time and capital. If planners rely on inflated demand figures, they risk overbuilding infrastructure that never gets used.
The opposite risk also exists. If utilities dismiss legitimate projects as phantom demand, they could underbuild capacity needed for real AI expansion. Either mistake carries costs. Ratepayers, investors, and grid reliability could all be affected depending on which way planners err.
The AI boom has already pushed some regions to reconsider long-term power planning. Data center operators sometimes file requests for grid connections at multiple sites while deciding where to actually build. That practice can make it hard for utilities to distinguish real commitments from exploratory ones.
CryptoBriefing's report does not specify exact figures for how much announced capacity may be phantom. It also does not name specific companies or regions tied to the concern. The broader point is that the electricity demand tied to AI growth may be harder to measure accurately than headline numbers suggest.
This uncertainty arrives as AI infrastructure spending has become a major driver of capital investment across the tech sector. Data center construction has been cited by several analysts as a key contributor to recent electricity demand growth forecasts. Grid operators in multiple states have flagged AI-related demand as a planning priority.
The crypto and energy sectors have some overlap here. Cryptocurrency mining operations have previously raised similar concerns about speculative power requests inflating grid planning assumptions. Some data center operators have shifted between AI and crypto mining uses depending on market conditions, which can further complicate demand tracking.
Regulators and utility commissions in several states have begun asking data center developers for more documentation of committed capital and signed contracts. That shift reflects growing awareness that announcement volume alone is not a reliable indicator of future electricity use.
Market Impact
If the phantom project concern proves widespread, utilities could revise long-term capacity plans downward in some regions. That could slow planned investment in transmission and generation tied to AI-specific demand forecasts. Investors in power infrastructure and data center developers may face closer scrutiny of which projects are actually funded and under construction.
For energy markets, clearer distinction between real and speculative demand could reduce forecasting volatility over time. It may also affect how quickly new generation capacity, including natural gas, nuclear, and renewable projects tied to AI demand, moves forward. Crypto mining operators competing for the same grid capacity could see shifting dynamics depending on how utilities recalibrate their planning assumptions.
The report underscores a broader challenge facing grid planners as AI-driven electricity demand accelerates. Separating real commitments from speculative announcements will likely remain a central issue for utilities, regulators, and investors in the months ahead.
Frequently Asked Questions
What is a 'phantom project' in the context of AI electricity demand?
It refers to an announced data center or AI infrastructure project that may never be built or that duplicates capacity requests across multiple sites, according to the CryptoBriefing report.
Why does this matter for US electricity grids?
Utilities plan new transmission lines and power generation years in advance based on projected demand, so inflated or unreliable figures could lead to overbuilding or underbuilding of infrastructure.
Does the report name specific companies involved in phantom projects?
No, the CryptoBriefing report does not identify specific companies, sites, or regions tied to the concern.
How does this relate to cryptocurrency mining?
Crypto mining operations have previously raised similar concerns about speculative power requests, and some facilities have shifted between AI and mining uses, complicating demand forecasts for grid operators.