Sam Altman’s idea of handing Americans a slice of the wealth created by artificial intelligence has resurfaced. It’s a way to share the “common resource” on which models are trained. According to the Financial Times, the OpenAI CEO is discussing with President Trump a 5% stake for the U.S. government, which could eventually be distributed to citizens.

The math, borrowed from MIT Technology Review, is curious: at the $852 billion valuation OpenAI reached in its March funding round, 5% would be about $42.6 billion. Split among roughly 133 million American households, that’s a little over $320 per household. Not exactly a dividend that calms automation anxiety, but the figure is almost beside the point.

The plan is, above all, a story — a narrative Altman has nurtured for five years: first a 2.5% levy on large companies to feed a universal fund, then repeatedly scaled back to today’s narrower stake reserved for OpenAI. There’s little concrete detail so far, just informal meetings with an administration that loves industrial-style deals.

The real purpose is twofold. First, to soothe widespread distrust: a majority of Americans don’t trust companies to use AI responsibly and oppose local data centers. Second, to secure privileged access to the White House: avoiding having one’s models labeled a supply-chain risk, or getting support against Chinese rivals. Staying in the small circle of geopolitical players is becoming essential for AI firms.

For those evaluating on-premise deployment of LLMs, the initiative signals a regulatory environment in motion, where political proximity could translate into competitive advantages or sudden constraints. Should public ownership ever morph into mandates for national infrastructure use or model certification requirements, organizations already invested in self-hosting and data sovereignty could adapt faster, without getting locked into cloud platforms that risk becoming tools of industrial policy.

The blueprint has a clear ancestor: the Alaska Permanent Fund of the 1970s, created to distribute oil profits — a collective resource, and a finite one. Altman seems to accept the first part of the equation — AI feeds on human-produced work — but rejects the second: he argues AI will generate inexhaustible wealth for decades. So the $300-a-household stake is meant to persuade people that the AI boom will be large enough to share, rather than to fill their wallets. It’s a message, not a check.