Data Centers Cut Electricity Costs 3.5% Per Capacity Doubling

A widespread assumption about data centers—that they inevitably drive up electricity costs for nearby residents—may have been wrong, at least until recently. New research from the Electric Power Research Institute reveals that data center operations actually reduced retail electricity prices through 2024, even as capacity expanded rapidly.

The working paper analyzed Federal Energy Regulatory Commission data and U.S. Energy Information Administration retail revenue figures from 2015 to 2024. Researchers identified a clear causal relationship: each doubling of data center capacity corresponded with a 3.5% decrease in average retail electricity prices. At the state level, the effect was even more pronounced, with prices dropping approximately 6%.

How data centers reduced costs

The mechanism behind this counterintuitive finding lies in the economics of electricity markets. Unlike commodity markets where prices reflect production costs, electricity pricing operates on cost recovery—dividing fixed infrastructure costs across total consumption.

"Electricity markets are different than a lot of markets that they interact with," Asa Watten, the study's coauthor and EPRI researcher, told Fortune. As data centers consume more kilowatt hours, those fixed costs get distributed across a larger base, reducing per-unit prices. Additionally, increased load brings more generators online, many of which operate with greater energy efficiency than older infrastructure.

The reversal is already beginning

This favorable pattern shows signs of breaking down. PJM, the nation's largest power grid operator, projected this week that consumer electricity costs will rise $6.3 billion over the next three years, primarily due to increased data center power demands. In Virginia, the state with the highest concentration of data centers, residential electricity prices jumped more than 13% in the past year, according to EIA data.

The industry faces a $7 trillion spending surge on data center construction through 2030. Goldman Sachs has projected AI infrastructure will push electricity costs up 6% between 2026 and 2027, with an additional 3% increase by 2028.

Why it matters

The future relationship between data centers and electricity costs hinges entirely on whether AI demand justifies the massive infrastructure buildout. If the grid expands capacity for data center loads that never materialize, fixed costs will be spread across fewer consumers than anticipated—the inverse of the beneficial pattern observed through 2024. This scenario would validate concerns about an AI bubble while saddling ratepayers with stranded infrastructure costs. Investor skepticism is already emerging: Tesla and Alphabet shares fell Thursday after both companies announced increased AI capital expenditures. Billionaire Mark Cuban warned this week that "a lot of data centers…are going to be turned into pickleball courts" as AI efficiency gains reduce the need for physical capacity.

Watten noted one optimistic scenario independent of AI's trajectory: broader electrification through electric vehicles and heat pumps could continue distributing fixed costs more efficiently, potentially stabilizing or reducing household energy expenses regardless of data center utilization rates.

These findings were first reported by Fortune.