Seizing the AI opportunity

OpenAI has filed a policy submission to the White House laying out what it thinks the U.S. needs to do to stay ahead in artificial intelligence, and the pitch boils down to infrastructure and people. Not model weights, not benchmarks — power grids, data centers, and a workforce that can actually build and run them.

The framing here is the "Intelligence Age," OpenAI's preferred term for whatever comes after the current AI boom settles into something more permanent. According to the submission, sustaining American leadership means expanding energy capacity at a pace that matches AI's appetite for compute, which has been growing far faster than the grid additions utilities are used to planning for. That's not a small ask. Data centers running frontier models already strain regional power supplies in places like Virginia and Texas, and the projected growth curve only gets steeper from here.

Workforce readiness is the second pillar, and it's the less flashy but arguably harder problem. Building gigawatt-scale data centers requires electricians, HVAC technicians, and grid engineers, not just machine learning researchers. OpenAI's submission ties AI leadership directly to whether the U.S. can train enough of these workers quickly enough, which puts the company in the odd position of lobbying for trade school funding as much as chip policy.

The subtext, of course, is competitive positioning against China, which has been aggressively building out its own AI-adjacent energy and manufacturing base. OpenAI isn't shy about framing this as a race with a winner and a loser. Whether the White House treats this submission as a serious policy roadmap or as one company's self-interested wish list is the actual question worth watching, since easier permitting and expanded energy capacity would obviously also just make OpenAI's own expansion plans cheaper and faster.