Pacific Gas and Electric (PG&E) entered 2026 forecasting a year's worth of new electricity demand. Barely two months later, nearly all of that capacity was already spoken for, as interconnection requests piled up faster than planners anticipated. The surge is overwhelming a regulatory system designed for a time when electricity demand barely budged.
That era is over. Load growth, which historically hovered below 1% annually, hit 4% at some grid operators last year, according to a Lawrence Berkeley National Laboratory report. Bain and Company projects AI data centers alone could consume up to 9% of total U.S. electricity by 2030, adding more than 150 terawatt-hours of demand the grid was never built to handle. A third of this new demand is concentrated in Virginia, Texas, and California, per the Pew Research Center.
Yet AI is only part of the picture. Electric vehicles, new factories, and industries shifting away from diesel and natural gas are all pulling from the same strained system. At the Southwest Power Pool, which manages electricity across 17 states, officials compared last year's demand surge to two simultaneous pressures, underscoring how broad-based the growth has become.
The grid's aging infrastructure faces a test it was not designed for. While the rapid buildout of data centers signals economic momentum, it also exposes bottlenecks in transmission capacity and interconnection processes. The mismatch between new demand and old hardware raises questions about reliability and the pace of necessary upgrades.
Some analysts caution that the projections may overstate the scale and speed of the shift. Grid operators could accelerate permitting and expand capacity faster than historical trends suggest, potentially softening the near-term strain. Still, the fundamental collision between AI's energy appetite and a creaking power system is set to define the coming decade for utilities and regulators alike.