A proposal allowing data centers to reduce consumption during peak demand periods rather than building new power plants could impose billions of dollars in costs on utility customers, according to a report released Thursday.
Monitoring Analytics LLC, the independent market monitor for PJM Interconnection LLC, found that massive data center campuses promising to throttle back during grid emergencies without adding generation capacity could increase annual capacity costs by as much as $5.48 billion, with billions more in additional energy expenses.
“The assertion that large new data center loads can be demand-side resources and do not require new capacity is a regulatory fiction,” the report said, noting that data centers typically require 99.999% electricity reliability, and making customer interruptions an impractical solution.
Surging power demand has strained aging infrastructure, raised consumer costs and sparked political controversy, asserts PJM, which manages the 13-state eastern grid spanning from the mid-Atlantic to the Midwest and hosts the world’s largest concentration of data centers.
The analysis represents the first attempt to quantify the costs of making data centers flexible to maximize existing grid capacity without expansion — essentially requiring facilities to shut down during extreme stress periods.
Concerns over data center energy consumption is mounting as quickly as Big Tech is throwing hundreds of billions of dollars into construction of vast facilities across the U.S. In recent months, Meta Platforms Inc., Anthropic, OpenAI, Oracle Corp., Amazon.com Inc. have either announced additional funding or new data centers, raising questions around energy use and environmental impact in local communities.
A Duke University study published earlier this year suggested the current U.S. grid contains approximately 100 gigawatts of available capacity for new demand if data centers reduced consumption at critical times.
Monitoring Analytics modeled capacity cost increases if 20 gigawatts of new data center demand were added to the grid, with 90% of that demand agreeing to shut down completely during extreme events. These facilities would receive payments to potentially serve as emergency resources for a few hours annually.
Even with 100% participation from new demand sources, capacity costs would still rise by approximately $396 million, the analysis determined. The report concluded that the only way to prevent other consumers from bearing the costs of data center growth is for these large users to build power plants capable of generating their own electricity.
According to a Berkeley Lab report, data centers’ annual energy consumption could reach between 74 and 132 gigawatts by 2028—representing 6.7% to 12% of total U.S. electricity use.
