OpenAI proposes handing the US government a 5% stake in itself

OpenAI has floated one of the most striking corporate-political proposals of the AI era: giving the US government a 5% equity stake in the company. The Financial Times reported on 2 July 2026 that the offer — worth roughly $42.6 billion at OpenAI's ~$852 billion post-money valuation — is intended to defuse mounting pressure in Washington and, in OpenAI's framing, to 'share the upside of AI' with the public. Two caveats belong up front: the reporting is second-hand via the FT, and the talks are described as conceptual, not agreed.

The mechanism matters more than the headline percentage. Under the reported plan, the 5% would not sit as a passive Treasury holding but would flow into a sovereign wealth vehicle modelled on the Alaska Permanent Fund — the state fund that pays Alaskans an annual dividend from oil revenues. Here the 'oil' is AI equity, and the dividend is a public payout funded by OpenAI's future value. Crucially, OpenAI reportedly envisions this as a template rather than a one-off: it wants other US labs — Anthropic, Google and Meta — to cede similar stakes into the same public fund. The idea extends OpenAI's own April proposal for a 'public wealth fund', giving it a concrete number and a concrete recipient.

The lobbying was senior and direct. Sam Altman engaged personally with President Trump, Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, and separately spoke with Senator Bernie Sanders — a signal that OpenAI is trying to build cover on both the executive and the populist-left flanks at once. That is a telling read on where the political risk is coming from: not one party, but a broad coalition uneasy about the concentration of AI wealth and power.

This is not as unprecedented as it sounds. In 2025 the US government took a ~10% stake in Intel (around $8.9 billion), alongside stakes in IBM and quantum-computing firms — establishing that Washington will now take direct equity positions in companies it deems strategic. Against that backdrop, OpenAI's 5% offer reads less like radical generosity and more like a pre-emptive, negotiated version of something the state has already shown it is willing to impose. And the ceiling is higher than 5%: Senator Sanders has pushed a far more aggressive design — a 50% government stake in major AI firms via a sovereign fund. OpenAI's 5%, in other words, is the low anchor in a negotiation whose upper bound is ten times larger.

The timing is not accidental. US frontier labs are under acute political heat: cybersecurity worries over what powerful models can do, and cheaper Chinese open-weight competition that undercuts the argument that American labs deserve a protected, lightly taxed run. Offering the government a share of the upside is a way to convert that hostility into partnership — to make Washington a stakeholder rather than an adversary. It also, not coincidentally, would make it far harder for regulators to break up or heavily constrain a company the Treasury co-owns.

But the obstacles are substantial, and WireRead would not price this as likely yet. The talks are conceptual; a stake of this kind would almost certainly require an act of Congress, not just executive agreement; and the multi-lab version depends on rivals playing along — yet Anthropic and the administration reportedly have not even discussed a stake, and the White House, OpenAI, Google and Meta largely declined to comment. Read plainly, this is a trial balloon with a real precedent behind it and a serious number attached — a signal of where the pressure is pushing the industry, not a done deal.