The United States is using a 1935 wartime law three times in six months to manage a problem created by artificial intelligence infrastructure, and grid experts say this pattern cannot continue. On June 30, Energy Secretary Chris Wright signed emergency orders forcing AI data centers onto diesel backup generators to prevent rolling blackouts across 13 states and Washington, D.C. This marks the third time in 2026 that the Department of Energy has invoked Section 202(c) of the Federal Power Act, a statute originally designed to compel utilities to cooperate during wartime.

Why Is the Power Grid Struggling With AI Data Centers?

The underlying problem is structural and predictable: data centers are being built far faster than the electrical infrastructure needed to power them. A data center can be constructed in 18 to 24 months, but the transmission lines and substations required to supply electricity take three to seven years to build. Transformer lead times have stretched to two to five years, creating a widening gap between demand and supply.

PJM Interconnection, the regional grid operator serving 67 million people, projects that data centers will add 65 gigawatts of new demand over the next decade. For context, that is roughly equivalent to the total electricity consumption of a large state. The grid missed its reliability reserve target for the first time in its history in December 2025, when the capacity auction cleared 6,625 megawatts short of PJM's own reliability requirement.

On July 2, 2026, PJM demand hit approximately 163 gigawatts, nearly matching the all-time record of 165.5 gigawatts set in 2006. Day-ahead electricity prices spiked past $2,000 per megawatt-hour in parts of the system, and operating reserves fell to just 5,091 megawatts, leaving almost no cushion against an unexpected outage.

What Do Emergency Orders Actually Do?

Wright's orders authorized data centers and other large electricity consumers with at least 50 megawatts of peak load to be disconnected from utility-supplied power and shifted to on-site backup generation on demand. Hospitals, 911 centers, water treatment plants, and defense facilities are exempt from these forced disconnections.

The orders also granted temporary relief from normal environmental permit restrictions for power plants, allowing them to operate beyond their standard limits for sulfur dioxide, nitrogen oxide, and carbon monoxide to maximize available generation capacity. PJM officials acknowledged the orders could result in exceedances of normal pollution limits.

How Are Communities Bearing the Health Costs?

• Diesel Pollution Exposure: When PJM directs a data center off the grid, facilities typically switch to Tier 2 diesel backup generators, among the most polluting forms of power generation permitted under EPA standards. These units are normally subject to strict EPA limits on emissions, but emergency waivers suspend those restrictions.
• Heat Island Effect: University of Cambridge researchers documented that temperatures in the immediate vicinity of data centers run approximately 2 degrees Celsius (3.6 degrees Fahrenheit) higher than surrounding areas on average, and as much as 9.2 degrees Celsius (16.6 degrees Fahrenheit) directly adjacent to facilities.
• Vulnerable Populations: Roughly 38 percent of Americans, approximately 125 million people, live within five miles of one of the country's approximately 3,000 operational data centers, according to Pew Research. In Virginia's Henrico County, home to 37 data centers, school administrators were asked by their utility to limit power usage during peak periods.

In the context of a heat dome already pushing temperatures to historic levels, the localized thermal loading from data centers compounds the public health exposure of living near large data center campuses. Earlier this year, approximately 50,000 California customers near Lake Tahoe were notified they would need to find a new electricity provider, a move linked to surging data center demand in the region.

What Do Experts Say About Using Emergency Powers as a Permanent Solution?

The Department of Energy's own regulations, finalized in 1981, state explicitly that the emergency authority "does not intend to replace prudent utility planning and system expansion." Three orders in six months for a predictable, structural load problem contradict that intent, and grid planners are beginning to say so openly.

A February 2026 Congressional Research Service analysis flagged the pattern as "novel" and noted that DOE's own regulatory language characterizes Section 202(c) as a tool for "unexpected inadequate supply," not a substitute for capacity planning. The Balch and Bingham law firm, which advises energy sector clients on federal regulatory compliance, wrote in March 2026 that the authority has transformed from an emergency backstop into "an active instrument of reliability management," a characterization that carries significant implications for how the U.S. regulates its electricity system going forward.

For historical context, Section 202(c) was enacted in 1935 and used roughly 29 times in the eight decades before 2026, with 22 of those uses during World War II. From 2000 through February 2026, the DOE used the authority approximately 26 times across all U.S. grid operators, typically in response to brief, unpredictable events like hurricanes, winter storms, or sudden equipment failures. Those orders generally lasted hours to days. Three orders within a single regional grid in a single calendar year is, by any historical measure, unprecedented.

What Are the Financial Implications for Households?

The emergency orders and grid strain are already translating into higher electricity costs. Families face potential increases of $70 per month in their electricity bills by 2028, according to projections tied to the escalating pattern of grid emergencies and the need for infrastructure expansion.

The mismatch between data center build speed and grid expansion speed is structural, not seasonal, and emergency law cannot resolve it. Grid planners acknowledge that the only lasting solution requires massive investment in transmission infrastructure, transformer manufacturing capacity, and generation resources. Until that infrastructure is built, the U.S. power system will remain vulnerable to the kind of emergencies that have now become routine.