US issues emergency power order for PJM grid amid AI demand surge
- Emergency order allows extra generation across PJM grid serving 67 million people
- Directive runs from October 13 to October 16 under Section 202(c) of Federal Power Act
- AI data center electricity demand projected to triple to 29 GW by 2030
- New $581 million package aims to add 2.5 GW capacity via TVA projects

*this image is generated using AI for illustrative purposes only.
The US Department of Energy issued an emergency order allowing extra power generation across the PJM Interconnection as unusual October heat and rising demand raise blackout risks for more than 67 million people.
Energy Secretary Chris Wright signed the directive under Section 202(c) of the Federal Power Act. The order runs from October 13 through October 16, directing PJM to dispatch specified units and permitting backup generation as a last resort before or during an Energy Emergency Alert Level 3.
Weather and maintenance tighten supply
The intervention comes during the fall maintenance season, when weather conditions are constraining grid flexibility. The National Weather Service forecast record or near-record warmth across parts of the Ohio Valley and Mid-Atlantic. Highs are expected in the upper 80s Fahrenheit in central Indiana and well into the 80s around Washington and Baltimore.
PJM coordinates electricity across all or parts of 13 states plus Washington, D.C. The grid operator stated that recent capacity auctions reflected "tightening supply and increasing demand outpacing new supply," while it works to connect large data centers without sacrificing reliability or affordability.
AI infrastructure drives long-term pressure
The immediate emergency order highlights a structural challenge facing US utilities: the rapid expansion of artificial intelligence infrastructure. Federal regulators recently approved a PJM registry for data centers as electricity demand from those facilities is projected to triple to 29 gigawatts by 2030.
Power availability has emerged as a major bottleneck for the AI data-center buildout. Separate projections indicate that AI-related electricity demand could surge sharply through the next decade, exacerbating the supply-demand imbalance visible in current capacity auctions.
Broader reliability efforts expand
This is not the first use of emergency authority by the administration this year. In July, the Department of Energy issued similar PJM orders during a severe heat wave as electricity demand approached record levels and generator outages tightened supply. That earlier intervention also authorized additional generation to protect grid reliability.
Separately, the Department of Energy announced plans on Tuesday to support 84 Tennessee Valley Authority transmission and distribution projects across eight states. The roughly $581 million package, including $250 million in federal funding, is expected to add about 2.5 gigawatts of grid capacity and improve reliability for 2.58 million families and businesses.
What the numbers show
The juxtaposition of short-term emergency measures with long-term infrastructure spending reveals a widening gap between immediate capacity needs and future demand growth. While the current order addresses a four-day window of weather-driven stress, the projection that AI demand will triple to 29 GW by 2030 suggests that temporary dispatch orders may become insufficient without permanent capacity additions. The $581 million investment in TVA projects adds only 2.5 GW, a fraction of the projected AI-driven load increase, indicating that near-term reliability will likely continue to depend on regulatory interventions rather than solely on new build-out.
How might the repeated use of Section 202(c) emergency orders influence PJM's capacity auction prices and long-term investment signals for new generation?
What specific regulatory or infrastructure bottlenecks must be resolved to accommodate the projected tripling of data center demand to 29 GW by 2030?
Will the $581 million TVA transmission investment be sufficient to mitigate regional reliability risks, or will additional federal funding be required to close the gap with AI-driven load growth?

























