Data Center Demand Strains US Grid: Emergency Orders Become Routine

The rapidly escalating electricity demand from data centers, especially those driven by artificial intelligence (AI), is placing unprecedented strain on the United States' energy grid. This situation has led to federal emergency orders, issued under Section 202(c) of the Federal Power Act and typically reserved for rare crises, becoming a routine measure to prevent blackouts. Grid operators are signaling that with demand outstripping supply, the nation's energy infrastructure faces a critical need for transformation.

PJM Interconnection, a major grid operator serving 65 million people across 13 states, hit an all-time record peak load of 168,158 megawatts on July 2. To keep the lights on, PJM requested federal emergency orders authorizing it to pull data centers off the grid onto their own backup generators and allowing power plants to exceed pollution limits. These orders have been invoked 43 times in the past 14 months, a stark increase compared to just 20 times over the preceding nearly 50 years. On January 22, 2026, U.S. Energy Secretary Chris Wright expanded the scope of these emergency orders to include the ability to direct large customers to switch to their own backup generators during emergencies. Many of these backup generators are diesel-powered, intended as a last resort rather than a routine grid-balancing tool.

Northern Virginia, a key hub for data center growth, hosts more data center capacity than the next five largest U.S. markets combined. Dominion Energy, the utility serving most of Virginia, serves approximately 450 data centers and is navigating a 70-gigawatt interconnection queue, a number that dwarfs the utility's all-time system peak of 24.7 gigawatts. U.S. data centers consumed roughly 4.4% of total U.S. electricity in 2023, with projections indicating this could rise to 12% by 2028. A single modern AI data center can consume as much power as 100,000 homes.

This surging demand is creating significant challenges for grid stability and raising concerns that the cost of expanding energy infrastructure to support data centers may be passed on to taxpayers and ratepayers. Bloomberg reported that half of all data center construction plans were canceled or delayed at the beginning of 2026, illustrating the impact of grid bottlenecks on projects. The North American Electric Reliability Corporation (NERC) issued a rare Level 3 Essential Actions notice in May, directing data centers to report usage rates to avoid reliability issues. Concurrently, the Federal Energy Regulatory Commission (FERC) has directed grid operators to propose reforms or justify their rules governing how data centers and other large customers connect to the electric grid within 60 days.

In a broader economic and political context, this situation highlights the tension between rapid technological advancement, particularly the AI boom, and existing infrastructure limitations. Former President Donald J. Trump declared a national energy emergency on January 20, 2025, citing a “precariously inadequate and intermittent energy supply, and an increasingly unreliable grid.” The Department of Energy (DOE) warns that blackouts could increase by 100 times by 2030 if reliable power sources are retired without timely replacement. The rapid growth of data centers is fueled by billions of dollars in private investment, state tax incentives, and federal directives aimed at fast-tracking permitting.

Analysts and market experts emphasize the need for flexible solutions to manage this challenge. Grid operators are focusing on strategies such as shifting data center workloads or utilizing backup power during peak periods. Tech giants like Google and Nvidia are exploring ways to manage their consumption more effectively. Research from Duke University suggests that flexible management approaches could save between $40 billion and $150 billion in capital investments over the next decade. This scenario also underscores the critical need for more accurate tools to assist grid operators in preventing instability.