OpenAI proposed giving the US government a 5% stake worth about $42.6 billion

OpenAI has proposed handing the US federal government a stake equal to about 5% of the company, according to a Financial Times report on July 2 that CNBC, Bloomberg, and CNN matched the same day. The plan is not a sale. OpenAI would contribute the equity to a US sovereign wealth fund, a vehicle the company floated as a version of the Alaska Permanent Fund, which pays state residents an annual check out of oil revenue.

Chief executive Sam Altman has discussed the idea directly with President Trump and with Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, according to the reporting. Everyone involved describes the talks as early and conceptual. Actually doing it could require an act of Congress.

What the number is measured against

The figure attached to the proposal, roughly $42.6 billion, is one multiplication: 5% of the $852 billion valuation OpenAI reached in the funding round it closed in March. That valuation is a private mark. It is what late-stage investors agreed to pay for a slice of the company, not a price quoted on any exchange, and it comes with none of the liquidity a public share carries.

So a 5% stake dropped into a sovereign wealth fund is a claim on a private, capped-profit company that pays no dividend and cannot be sold into a market that does not exist. It is worth $42.6 billion the way a house is worth its last appraisal: real enough on paper, untested until someone sells. The equity is not nothing. But the headline number is doing a lot of work, and it rests on a valuation OpenAI set with its own investors.

Why it is surfacing now

The pitch lands in the middle of real political pressure on OpenAI: job-loss worry, local resistance to the data centers going up to power these models, and complaints that cheaper Chinese systems are closing the gap. Altman has argued that giving the public a direct financial stake is the cleanest way to share whatever upside AI produces. He first raised a version of this with the administration in early 2025, so the idea is not new. The timing is. It comes days after Washington slowed OpenAI's rollout of GPT-5.6, a model the company had already limited to a short list of vetted customers. Read one way, an equity offer buys down regulatory risk. Read another, it ties the company's windfall to the government refereeing it. Both can be true at once.

The parts nobody has agreed to

OpenAI's plan assumes it will not be alone. As reported, the same fund would eventually hold stakes contributed by Anthropic, Google, and Meta. None of those companies has said publicly that it will take part, and the framing that every leading lab pays in is OpenAI's, not theirs.

Then there is the governance question the proposal raises and does not answer. The federal government would own equity in a company its own agencies regulate, and would sit on the cap table of one AI lab while the others watch. Whether the stake would carry a vote, who would manage the fund, and what the government could actually do with the shares are all unspecified.

Strip the framing and the concrete event is modest: a sourced report that OpenAI has revived an idea it has floated before, in talks all sides call preliminary. No term sheet, no vote, no signature. It is worth tracking, because a government equity position in a frontier AI lab would be a new kind of arrangement, and because it shows how OpenAI intends to manage Washington. It is not a deal yet.