AI is becoming a power-grid problem
Building AI used to be a software problem; now it is increasingly an electricity problem. Training and running frontier models requires data centres at industrial scale, and the bottleneck has shifted from chips to the gigawatts of power those data centres consume. Local communities are pushing back, utilities are racing to build, and the question of where AI gets built is starting to be answered by where the electricity is.
Timeline
- January 25, 2024
Meta announces it will build a $10 billion data centre cluster in Louisiana — one of the first signals that the new round of AI infrastructure is moving where the power is, not where the talent is.
- September 20, 2024
Microsoft signs a 20-year deal to restart Three Mile Island's reactor 1 to power its AI data centres. Restarting a closed nuclear plant for AI sets a new bar for what AI infrastructure now demands.
- December 9, 2024
OpenAI, Oracle, and SoftBank announce "Stargate," a $500 billion AI infrastructure plan focused on US data-centre build-out. The plan reframes AI competition as an industrial-policy and power-supply contest.
- April 15, 2025
The first major utility — Dominion Energy in Virginia — publicly warns it cannot meet projected AI data-centre demand without major new generation capacity. Other utilities echo similar warnings within weeks.
- October 8, 2025
Several US states begin drafting moratoriums on new data-centre approvals, citing strain on power grids and water resources. The push-back is bipartisan.
- May 14, 2026
Denver becomes the first major US city to enact a formal moratorium on new AI data-centre construction within its limits. Other cities are watching and considering similar moves.
Where things stand right now
Power and land have become the binding constraints on AI build-out, not chips. The frontier labs and hyperscalers are racing to lock down generation capacity through nuclear restarts, behind-the-meter generation, and long-term utility contracts — while local communities are pushing back on the industrial footprint AI requires. Where AI gets built in 2027 will be decided as much by zoning boards and utility commissioners as by engineering teams.