ERCOT Forecasts Record 92,211 MW Demand in Texas, 8% Above Historical Peak

The Electric Reliability Council of Texas (ERCOT) projects a peak demand of 92,211 megawatts (MW) for summer 2026, 8% above the historical record of 85,464 MW set in August 2023, according to its seasonal report of May 24.

The growth is driven by the expansion of data centers and cryptocurrency mining. ERCOT reports 466 MW of new crypto mining load since September 2025 and projects 1,725 MW of additional large-load additions between May and September 2026. The U.S. Energy Information Administration (EIA) estimates that demand on the Texas grid will grow 14% in the first nine months of 2026, the fastest pace among all U.S. electric interconnections.

The phenomenon mirrors, on the U.S. side of the energy bridge, the same data-center load pressure facing Mexico's National Electric System (SEN). Texas exports the bulk of the natural gas Mexico consumes by pipeline, and rising Texas demand competes directly with cross-border supply.

To meet the growth, ERCOT added 11 GW of new capacity since last summer, comprising nearly 5 GW of batteries, 4 GW of solar generation, and 1.5 GW of natural gas. The EIA documents that utility-scale solar generation in ERCOT averaged 24 GW during midday hours in summer 2025, double the 2023 level, and that batteries dispatched an average of 4 GW during the evening peak hour. The probability of a grid emergency in July is estimated at 0.21%, according to ERCOT itself.

On the transmission side, Sempra announced in June an investment plan of more than $7 billion to expand the Texas grid between 2026 and 2034, aimed at serving 16 GW of new load, primarily data centers and artificial intelligence, in the Dallas-Fort Worth area.

The performance of the Texas grid this summer will test a system that is adding load faster than any other U.S. interconnection. For Mexico, ERCOT's demand trajectory is a leading indicator of pressure on the Permian gas that supplies both sides of the border.

This article was drafted with artificial intelligence assistance from verified sources and reviewed by a human editor before publication.