AI Data Center Boom Faces a Power Grid That Can't Keep Up

A report published Tuesday frames the AI buildout as running into what it calls a three-year energy wall: an estimated $1 trillion in committed spending on AI infrastructure is arriving faster than new power generation and transmission capacity can be built in the United States. The gap between capital committed and electrons available is emerging as a constraint on how quickly AI data centers can actually come online. The timing matters because it lands alongside separate signs of caution elsewhere in energy-adjacent markets. Oil and gas private equity deals have tumbled as new platform buyouts dry up, according to a separate report, suggesting capital in traditional energy is pulling back even as AI-driven electricity demand grows. Widening credit spreads globally, flagged in a separate MarketWatch analysis, are also cited as a sign that how risk assets are valued more broadly may be shifting. What's unresolved is the pace of resolution: whether utilities, permitting timelines, or new generation technology can compress that three-year estimate, or whether it lengthens as AI capacity commitments keep growing faster than the $1 trillion figure already on the books.