Trump administration weighs government equity stakes in US AI companies

The mechanics are where the idea wobbles. Rather than taxpayers buying into OpenAI at its $850B-plus valuation, the company would donate equity to seed a 'Public Wealth Fund,' whose proceeds could flow directly to citizens, including those who own no stocks at all ^[1]. That structure is clever because it sidesteps a cash outlay from a government already swimming in debt. But the cleverness ends at the implementation. As The Next Web's reporting bluntly frames it, the most fundamental obstacle is mechanical: there is no legal framework for moving private AI-company equity onto the federal balance sheet ^[3]. Compounding that, OpenAI, Anthropic, and xAI are wired differently at the corporate level, so a template that works for one could be useless for the next ^[3]. A 'partnership with the American public' is an appealing slogan; it is not yet a transaction anyone can describe.

This is not coming out of nowhere. Trump has taken partial government ownership of U.S. firms more aggressively than any recent president, with direct investments in at least 10 companies during his second term, including Intel, Lithium Americas, and Trilogy Metals ^[1][4]. The clearest precedent is MP Materials, where the Defense Department put up a $400M preferred-equity stake, committed to as much as $350M more, and added a $150M loan to stand up a domestic rare-earth magnet supply chain ^[5]. Those deals were sold on national-security and supply-chain logic, and Trump has called the Intel arrangement a 'direct windfall for American taxpayers' ^[1]. AI is the natural next target for that worldview: strategically vital, dominated by a few firms, and minting paper wealth at a scale no policymaker wants to watch from the sidelines. The equity-stake habit, in other words, has a runway.

What looks like one debate is really three competing proposals. Trump's version is the vaguest and softest: a donated stake, a 'partnership,' a promise to 'look into it' ^[2]. Bernie Sanders sits at the opposite pole, floating a 50% government equity stake in the largest AI firms via an AI Sovereign Wealth Fund Act, levied as a one-time 50% stock tax paid in shares ^[1][3]. And then there is the opposition from inside Trump's own coalition: former AI czar David Sacks, who argues that deeper government ownership would 'accelerate the corporate-government fusion we're already sliding toward' and warns against nationalization and bailouts ^[2]. The striking thing is the ideological scramble. A Republican president is entertaining what critics call nationalization, a democratic socialist senator wants to go further still, and a libertarian-leaning ally is the one yelling stop. The usual political map does not apply here.

The framing battle may matter more than the policy. The administration sells this as ordinary Americans sharing in AI's upside; the audience is not buying the pitch at face value. Public Knowledge's Nat Purser names the cleanest structural problem: the government would be 'a shareholder and a regulator at the same time, which creates substantial conflicts of interest' ^[1]. That critique landed even harder in the financial communities, where the dominant read was that this is a thinly disguised bailout or market pump for AI firms, profits accruing to insiders while taxpayers absorb the downside, with sharp skepticism that any 'public stake' would carry voting power or real economic interest. A smaller camp connected it to Sanders' sovereign-wealth logic, reasoning that if AI is going to take their jobs they may as well own a slice of it. Across business television the story was treated as a market-moving event, with traders tying the AI-stake chatter to broader tech volatility. When the same proposal reads as 'partnership' to its authors and 'self-dealing' to its audience, the gap itself becomes the political risk.