Underneath the reported proposal is a familiar corporate maneuver in regulated industries: when Washington frames you as an existential risk, the cheapest exit is to make the government a co owner rather than an adversary.
OpenAI's reported pitch to give the U.S. government a roughly 5% equity stake is not a capital raise. It is political insurance. The mechanism underneath the proposal is a familiar one in regulated industries: when the government has decided your category poses an existential risk, the cheapest exit is to make Washington a co-owner rather than an adversary.
The Financial Times reported on Thursday that OpenAI has discussed handing the federal government a stake of about 5%, citing two people familiar with the talks, and that CEO Sam Altman and other executives have held meetings with Trump administration officials about the structure. The paper also said OpenAI floated the idea that other major U.S. AI companies hand the government similar stakes, though it is unclear whether any peer would agree. Reuters carried the FT's reporting but said it could not immediately independently verify the scoop. Bloomberg, the Guardian, and CNBC all ran the same story. Neither OpenAI nor the White House had commented at the time of publication.
The framing matters more than the number. A 5% slice of a private company worth hundreds of billions of dollars would be a large windfall on paper. As governance, it is a much bigger concession. A government shareholder does not just collect dividends; it acquires standing to question board decisions, demand disclosure, and slow transactions it dislikes. For a frontier AI lab already navigating export controls, state-level model legislation, and a bipartisan worry that the technology is moving faster than the rules meant to govern it, a 5% block owned by the federal government is a lever the company cannot easily dislodge.
Why offer it? Because the alternative has gotten more expensive. President Trump said in June he was exploring ways to give the public a stake in leading AI firms, citing concerns that Americans would not share in sector profits, and OpenAI had previously pitched a "public wealth fund" to invest in AI companies and distribute proceeds to citizens. Anthropic separately said it is exploring a "digital dividend" funded by taxes on the sector. The two proposals read as different responses to the same political pressure: if Washington is going to treat frontier AI as a strategic resource, firms have an incentive to be on the right side of that treatment rather than the target of it.
The structural logic is not new. U.S. nuclear utilities spent decades accepting rate-base guaranteed returns on capital in exchange for shouldering strict liability and the accident-coverage pool created by the Price-Anderson Act. The trade was not generosity; it was a way to convert a politically fatal framing, that nuclear power was an uncontrollable hazard, into a regulated industry with a built-in federal partner. OpenAI's pitch, if the FT's sources are right, is the same maneuver applied to a different existential frame: convert the AI lab from "too dangerous to leave alone" into "too entangled to unwind." The White House becomes a co-owner with reasons to defend the company's franchise, not just to police it.
That logic also explains the "other AI companies should do it too" line in the FT report. The proposal is only useful as collective action. If OpenAI is the only firm that hands the government a stake, it has bought itself a regulator with skin in the game and given rivals a regulator that still has every reason to come down hard on the one firm that signed up. OpenAI's pitch to peers is effectively: let us all make the same concession, so the political cost of being singled out disappears.
Whether peers agree is the open question, and it is where the FT scoop runs out of road. The paper does not name which other companies OpenAI approached, and the major U.S. AI labs besides OpenAI and Anthropic have not been named in the reporting. Some, like Anthropic with its separate digital-dividend proposal, are already constructing their own political accommodations and may see a direct equity gift as more exposure than it is worth. Others, particularly firms with strong foreign investment or antitrust exposure, may calculate that a U.S. government shareholder complicates more than it solves.
What to watch next: on-record confirmation or denial from OpenAI, the White House, or peer labs; any legal or national-security pushback inside the administration over the precedent of sovereign equity in a frontier AI firm; and whether the "other companies should do it too" line of the FT report hardens into a formal inter-firm proposal or fades as a negotiating posture. The story's weight sits in the precedent. A confirmed 5% federal stake in one frontier AI lab is a precedent for the next one, and the next, until the category stops being exceptional.