US Utilities Struggle to Meet Big Tech’s Data Center Power Demands
Subscribe to our free newsletter today to keep up-to-date with the latest energy, oil and gas news.
As AI technologies scale, US utilities are facing record-high energy requests from tech companies. Data centers operated by Amazon, Google, and Microsoft are consuming vast amounts of electricity, straining regional grids.
Some requests now exceed utility capacity. Dominion Energy in Virginia received a proposal for 2 gigawatts of power, equal to two nuclear reactors. Georgia approved $4 billion in utility investment, largely to support data centers. Duke Energy in North Carolina is expanding transmission infrastructure to manage projected load increases.
These surges complicate energy planning. Utilities are unsure whether demand forecasts are accurate or inflated for negotiation purposes. Executives worry about overbuilding and wasting capital on projects that may never be used.
Forecasting is challenged by speculative demand and tech firm bidding strategies
Tech companies often seek bids from multiple regions for future data centers, leaving utilities in a bind. Without confirmed projects, utilities struggle to forecast load accurately.
A source from Reuters mentioned, “A survey of 13 major US electric utility earnings transcripts found nearly half have received inquiries from data center companies for volumes of power that would exceed their peak demand or existing generation capacity”.
If utilities overbuild, they risk financial losses. If they underbuild, they risk outages and penalties. To manage this, some utilities are delaying upgrades until contracts are secured. Others are redesigning systems to be more modular and adaptive.
Microsoft recently reduced its expansion plans by up to 2 gigawatts. As AI becomes more efficient, power needs may decline. These shifts create even more uncertainty for infrastructure planning.
Utilities face financial risk as capital investments increase
Utilities like Georgia Power and Dominion are committing billions to support rising demand. These investments, which cover substations, grid upgrades, and new power plants, typically require rate increases approved by regulators.
Public utility commissions are raising concerns. If tech companies cancel projects or overestimate usage, utilities could be left with stranded assets and unrecoverable costs. Regulators are asking for more evidence before approving large-scale infrastructure spending.
AI efficiency gains and regulatory scrutiny are reshaping energy strategy
New AI models now require fewer chips, lowering electricity demand per computation. This trend has already influenced corporate decisions. Microsoft’s reduced construction plans are partly tied to better chip and cooling efficiency.
Meanwhile, regulators are considering stricter policies. These may include clearer energy disclosures, tiered pricing, and contracts that lock in minimum usage. The Department of Energy has also begun reviewing AI-related power trends for broader oversight.
Utilities are increasingly tying infrastructure to renewable sources like solar and wind. However, renewables introduce variability, which complicates grid balancing during peak periods.
To reduce planning risks, utilities are pushing for fixed energy contracts with penalties for missed targets. These deals help ensure infrastructure is built to serve real, not hypothetical, needs.
In several states, utilities, regulators, and tech companies have formed working groups. These forums allow for shared planning, improved transparency, and faster regulatory reviews.
Sources: