By Charles Kennedy of OilPrice.com

Pacific Gas and Electric has unveiled a $73-billion spending program through 2030 to overhaul California’s grid as electricity demand from artificial intelligence and cloud infrastructure explodes, Reuters reported on Monday. The utility said the plan will support as much as 10 gigawatts of new load from data centers slated for development in its service territory over the next decade.

Brent crude and copper traders aren’t the only ones recalibrating supply-demand curves. U.S. utilities are now staring down a structural load shock driven by AI. PG&E’s multibillion-dollar commitment indicates that AI has shifted from a niche workload to a dominant driver of electricity consumption. The scale is staggering, with a single hyperscale AI campus capable of drawing as much power as a small city, overwhelming systems already balancing renewable intermittency and wildfire risks.

California’s grid operator, CAISO, projects peak demand climbing from about 46,094 megawatts in 2025 to nearly 52,940 MW by 2030, which represents a 15% increase before counting speculative AI loads. Its 2025 assessment shows only 2,163 MW of new capacity coming online by mid-year, mostly batteries (1,654 MW) and solar (354 MW). That buildout won’t keep pace if AI data centers scale at current trajectories. 

California’s regulators estimate the state will need more than $30 billion in transmission and distribution upgrades over the next two decades just to stay level.

PG&E’s plan folds wildfire-hardening and undergrounding projects into a broader grid expansion blueprint, but the headline number is being read as a direct response to the “energy monster AI is creating.” Utilities from Virginia to Texas are reporting similar strains, and California’s tight reserve margins leave less room to maneuver.