November 22, 2025 - A new federal reliability report warns that energy-hungry artificial intelligence data centers are driving winter electricity demand to record highs, putting large parts of North America at risk of blackouts.

The North American Electric Reliability Corporation (NERC) has warned that growing electricity demand from artificial intelligence (AI) data centers could push the nation’s power grids to the breaking point this winter.

In its 2025–2026 Winter Reliability Assessment, NERC said “much of North America” faces a heightened risk of failing to meet demand during “extreme operating conditions.” The problem is especially acute in regions dense with AI data centers, where round-the-clock power consumption is surging.

“Winter electricity demand is rising at the fastest rate in recent years, particularly in areas where data center development is occurring,” the agency said Tuesday. “Although resources are adequate for normal winter peak demand, any prolonged, wide-area cold snaps will be challenging. This is largely due to rising electricity demand, which has grown by 20 gigawatts since last winter, significantly outpacing winter on-peak capacity.”

Winter storm warning echoes 2021 crisis

NERC officials compared current risks to those seen during Winter Storm Uri, which crippled Texas and the South-Central U.S. four years ago.

“Electricity demand continues to grow faster than the resources being added to the grid, especially during the most extreme winter conditions where actual demand can topple forecasts by as much as 25% — as we saw in 2021 in ERCOT and SPP,” said John Moura, NERC’s director of reliability assessments.

He noted that, while utilities have improved cold-weather readiness, major vulnerabilities remain in fuel delivery and energy storage under severe cold conditions.

Strain from intermittent power and 24-hour operations

NERC’s report warned that reliance on intermittent renewable sources such as wind and solar could worsen reliability problems during winter peaks.

“In winter, peak demands typically occur before sunrise and after sunset, coinciding with the unavailability of solar generation,” the agency said. “Data centers are altering the daily load shape due to their round-the-clock operating pattern, lengthening peak demand periods.”

The organization said the rapid growth of AI infrastructure, electric vehicle charging networks, and semiconductor plants is changing grid dynamics faster than utilities can adapt.

DOE issues emergency order to reinforce grid

Responding to the warnings, the Department of Energy (DOE) announced a new emergency order on Wednesday to bolster grid reliability. The order extends operations at Michigan’s J.H. Campbell coal plant through February 2026. The facility sits on the Midcontinent Independent System Operator (MISO) grid, which suffered a major outage over Memorial Day weekend — two days after DOE first intervened to prevent the plant’s early retirement.

Energy Secretary Chris Wright said the administration is moving quickly to reverse “energy subtraction policies” that weakened grid stability.

“Because of the last administration’s dangerous energy subtraction policies targeting reliable and affordable energy sources, the United States continues to face an energy emergency,” Wright said Wednesday. “The Trump administration will keep taking action to reverse these policies, lowering energy costs and minimizing the risks of blackouts.”

AI expansion outpacing grid capacity

The Biden administration had pushed electrification and green energy initiatives, while President Donald Trump has accelerated federal permitting to support AI and data center expansion. But grid experts warn that infrastructure upgrades have not kept pace with power-hungry data centers and semiconductor facilities.

Reports from WSLS and Business Insider have echoed those concerns, warning that regions with dense clusters of AI computing operations may face rolling blackouts this winter.

As NERC’s report concludes, the grid’s ability to handle extreme cold will depend on how effectively operators can manage both surging demand and intermittent supply — a challenge made more urgent by America’s growing dependence on data-driven technology.