OpenAI has proposed giving the US government a stake of roughly 5% in the company, the Financial Times reported on July 2, a move that would seek to ease growing political pressure in Washington.
At OpenAI’s March valuation of $852 billion, a 5% holding would be worth about $42.6 billion. According to the report, chief executive Sam Altman framed the idea as the best way to give the public a financial interest in AI’s upside, and pitched it as part of a broader arrangement in which each leading US AI developer would cede a similar stake to a government vehicle modeled on the Alaska Permanent Fund, which pays residents annual dividends from oil revenue. Reuters said it could not independently verify the report, and OpenAI and the White House did not comment.
A key detail is how the stake would be transferred. Rather than the government buying in with taxpayer money, OpenAI would donate the equity, avoiding any cash outlay. The shares would help seed what OpenAI calls a “Public Wealth Fund,” an idea from an April policy paper proposing a state-held fund that invests in AI companies and distributes returns directly to citizens, including those who own no stock.
No final terms have been agreed, and the figures remain subject to change. Altman has discussed the concept with President Trump, Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, and even with progressive Senator Bernie Sanders, over talks stretching back more than a year.
The report named OpenAI, Google and Meta as companies that might cede similar stakes, but that framing appears to overstate their involvement. Separate reporting on the same day indicated Anthropic is not part of the discussions, with a source saying the Claude maker is not negotiating government equity, in keeping with its tense relationship with the White House over AI safety policy.
Anthropic has instead floated a different idea, a “digital dividend” funded by taxes on the AI sector. Whether Google or Meta would entertain the proposal is also unclear, making OpenAI’s plan for now a one-company pitch dressed as an industry framework.
Why It Matters
The proposal reflects two pressures bearing down on OpenAI at once. One is political: Washington is increasingly wary of AI models’ security risks and of cheaper, capable Chinese rivals, and lawmakers are questioning whether ordinary Americans will share in AI’s profits.
The other is strategic: an equity tie to the government could buy goodwill for a company navigating export rules, a staggered model rollout demanded by the White House, and a possible public offering. The Trump administration has shown appetite for such deals, having taken a 10% stake in Intel and positions in IBM and other firms, and Trump has called public ownership of AI giants “a beautiful thing.” A donated stake is a low-cost way for OpenAI to align itself with that impulse.
The Objections
The idea draws criticism from both directions. Some analysts warn of a structural conflict: making the government a shareholder in the same companies it regulates could distort oversight and tempt officials to favor firms in which the state holds equity, a concern raised by the advocacy group Public Knowledge, while the libertarian Cato Institute cautioned against Washington picking winners.
From the other side, Sanders argues 5% is far too little; his bill would impose a one-time 50% equity levy on OpenAI, Anthropic and xAI, with government voting shares and board seats. Between those poles, OpenAI’s offer looks like an attempt to define the terms of an intervention that may be coming regardless, on terms it can live with. Whether rivals join, and whether the government accepts, remains unresolved.
