Sam Altman pitched the structure to Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent as an Alaska style oil dividend fund. The White House has separately asked OpenAI to restrict GPT 5.6 release.
OpenAI's proposed 5% government stake is not, in any meaningful sense, an Alaska Permanent Fund. It is a gatekeeping deal wrapped in oil-revenue rhetoric, and the moment that makes the wrapper credible is a White House request to keep the next frontier model away from the open market.
Sam Altman raised the 5% figure in early discussions with Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, according to the Financial Times, as reported by HNGN. At OpenAI's $852 billion post-money valuation following its March funding round, 5% is roughly $42.6 billion. The proposal frames the stake as a contribution to a public investment vehicle modeled on the Alaska Permanent Fund, the state-owned corporation that distributes oil-revenue dividends to Alaska residents, with Anthropic, Google, and Meta asked to contribute 5% each as well.
The packaging is deliberate. Alaska's fund is a popular symbol: a state harvesting a finite natural resource and sending checks to its citizens. The pitch borrows that image to make a politically toxic arrangement sound like a dividend. What the proposal would actually establish is a government equity position in a frontier AI lab, with a seat at the table proportionate to a $42.6 billion check.
That seat is the mechanism. OpenAI is negotiating with an administration that has already shown it can use export controls, release approvals, and procurement rules to shape what frontier AI companies can sell, to whom, and when. A government equity position turns a relationship that currently runs through regulatory discretion into a structural governance right. The Alaska Permanent Fund language is the cover. The substance is that Washington gets a permanent claim on OpenAI's upside, and a corresponding claim on its decisions.
The trigger sits in the same news cycle. The White House asked OpenAI last month to limit the release of its GPT-5.6 model to government-approved customers, a step beyond the export-control regime that has governed advanced chips for years. Anthropic faces parallel scrutiny. Bloomberg and CNN both reported the talks as "early conversations" or "preliminary," and there is no signed agreement. The request to throttle GPT-5.6 is what gives a Permanent Fund pitch its operational meaning. If Washington can already condition release on political alignment, a 5% equity position makes that conditionality permanent.
The proposal has not been matched by other labs. A person familiar with the matter told the FT the Trump administration and Anthropic have not discussed a government stake in Anthropic, as CNBC noted. Google and Meta's participation is described in coverage as part of the structure Altman has proposed, not as something those companies have agreed to. Framing this as an industry coalition overstates OpenAI's reach.
Altman has also brought the idea to Senator Bernie Sanders, who has separately proposed a 50% government stake in major AI companies, funded through a one-time tax on their stock. The two proposals are distinct and should not be conflated. Sanders frames his version as compensation for AI built on collective human knowledge. Altman's version is corporate negotiation, dressed in the same Permanent Fund language but pointed at a smaller slice and a different buyer. President Trump told reporters last month he was "hearing concepts where pieces could be given to the American public," calling the idea "interesting" and saying his administration would "look into" it, per The Guardian.
The administration's track record makes the proposal less hypothetical than the language of "floated" suggests. The Trump second term has taken a 10% stake in Intel worth $8.9 billion, plus positions in IBM and several quantum-computing and critical-minerals firms. The model is not new; applying it to a private frontier AI lab at this scale is. OpenAI's April industrial policy document already floated a "Public Wealth Fund" alongside proposals for four-day work weeks and higher corporate taxes to offset AI-driven job loss, as covered by Euronews. The 5% figure is the negotiating conversion of that vision into a deal Washington can sign.
The next tell is whether the GPT-5.6 release restriction becomes policy and whether Anthropic, Google, or Meta are asked to match OpenAI's offer.