OpenAI Proposes 5% Equity Stake to U.S. Sovereign Wealth Fund in Bold Political Gambit

OpenAI CEO Sam Altman has reportedly floated a politically charged proposal that could reshape the relationship between the U.S. government and the AI industry: donating a 5% equity stake in OpenAI—valued at roughly $42.6 billion—to a newly created U.S. sovereign wealth fund modeled on the Alaska Permanent Fund.

The proposal, first reported by the Financial Times in early July, is not merely a strategic defensive maneuver. Altman has framed it as a template for the entire sector, with Anthropic, Google, and Meta potentially ceding similar stakes to the same government vehicle. If enacted, it would mark an unprecedented entanglement of Washington’s institutional financial interests with the private AI industry—one with regulatory, competitive, and constitutional ramifications that experts are only beginning to map.

The Mechanics of the Deal

On paper, the structure is relatively straightforward. OpenAI would transfer 5% of its equity—equivalent to approximately $42.6 billion based on the company’s recent $965 billion valuation following its $65 billion Series H—to a federal sovereign wealth fund managed by or under the supervision of the U.S. Treasury. The fund’s logic mirrors that of the Alaska Permanent Fund, established in 1976 to convert surplus oil revenues into permanent public wealth. In Altman’s version, the resource being monetized is the AI boom itself.

Two possible structures are being discussed. In one, OpenAI issues shares directly to the fund. In a broader variant, OpenAI contributes to an industry-wide vehicle in which Anthropic, Google, and Meta participate on similar terms, creating a diversified portfolio of equity stakes across America’s leading AI developers.

Senator Bernie Sanders has engaged in conversations with Altman on the proposal and pushed for a considerably more expansive version: a 50% stake in all major AI companies. Most observers regard that figure as politically untenable, but it signals the ideological range within which any congressional debate would unfold—and puts Altman’s 5% offer in the context of a genuine negotiating floor rather than a ceiling.

Why OpenAI Needs Washington on Its Side

This proposal is not charity. It is calculated political risk management at a moment of unusual vulnerability for OpenAI.

The company has spent the past 18 months navigating a treacherous environment: a California attorney general lawsuit over its nonprofit-to-for-profit corporate restructuring, sustained congressional scrutiny over safety practices and model capabilities, and the erosion of its once-commanding revenue lead as Anthropic’s annualized run rate surged to $47 billion—eclipsing OpenAI’s own reported range of $25 to $33 billion.

A government equity stake creates structural alignment. If Washington holds shares in OpenAI, it acquires a financial incentive to see the company succeed rather than over-regulate it into irrelevance or competitive disadvantage against Chinese AI developers. The arrangement also provides a potent rebuttal to the critique that AI profits flow entirely to private investors and overseas capital while infrastructure costs—grid upgrades, water usage, national security risks—are socialized onto the public.

“The question is no longer whether AI will reshape the economy,” one regulatory analyst tracking the proposal said. “The question is who benefits. Giving the government a seat at the table—literally, as a shareholder—is one answer, however imperfect and unprecedented.”

The Sovereign Wealth Fund Model

The Alaska Permanent Fund analogy is instructive but imperfect. Alaska’s fund pays annual dividends directly to state residents; a federal AI fund would more likely be administered through a dedicated institutional vehicle, with returns directed toward public benefit programs, deficit reduction, or reinvestment in critical infrastructure such as power grids, semiconductor manufacturing, or broadband.

Creating such a fund would almost certainly require an act of Congress. The current administration has signaled an appetite for winning the AI race but has generally preferred soft governance mechanisms—like the voluntary AI release standards announced last week—over binding mandates. A sovereign wealth fund would be a much harder commitment, requiring legislation, oversight structures, a governance framework for when and how to exercise any shareholder rights, and a clear answer to whether the government intends to act as a passive investor or as an active voice in AI company decision-making.

The Constitutional complications are substantial. Can the federal government legally hold equity in a private company without running afoul of restrictions on government commercial activity? How would the fund manage conflicts of interest when it holds stakes in competing companies simultaneously? These are not abstract questions—they are the kinds of issues that have delayed similar proposals in other sectors for years.

The Industry-Wide Domino Effect

The most consequential aspect of the proposal may not be the OpenAI stake itself but the precedent it would establish. If the leading AI company cedes equity to the government, it becomes politically difficult for rivals to refuse. A company that declines to participate can easily be portrayed as unwilling to share AI’s benefits with the public—a damaging narrative in an environment where public trust in AI companies is already fragile.

For Google and Meta, however, the structural complications are severe. As public companies with existing shareholders, any voluntary equity transfer must be approved by their boards and would face scrutiny from investors who could argue it diminishes shareholder value. Anthropic, still private, has more flexibility but also more to lose from the perception of being coerced into a political arrangement by a competitor that moved first.

For smaller AI startups and open-source developers, the implications are murkier still. Would a sovereign wealth fund eventually require compliance from any company crossing certain revenue or model capability thresholds? Could it evolve into a de facto licensing regime for the most powerful frontier AI systems—something the Biden administration attempted through executive orders that were subsequently rolled back?

What Happens Next

For now, Altman’s proposal remains a trial balloon. No legislation has been introduced. No equity has been transferred. Anthropic, Google, and Meta have each declined to comment on whether they would participate in a similar arrangement.

The path to any legislative action runs through a deeply divided Congress with limited appetite for complex new regulatory frameworks, especially ones that could be characterized as partial nationalization of private industry. Sources familiar with the discussions stress that the framework could still shift significantly before anything is formalized.

But the proposal has already changed the Overton window of what is politically thinkable. A year ago, the notion of the U.S. federal government holding equity stakes in private AI companies would have been dismissed as socialist fantasy in most mainstream policy circles. Today, it is being discussed in the Financial Times, on Capitol Hill, and in the board rooms of every major AI company.

For investors and policy watchers, the critical risk is not necessarily that the proposal succeeds in its current form, but that it generates regulatory momentum that changes the political math on AI investment regardless of whether any equity actually changes hands. The moment governments begin treating AI companies as public utilities or strategic national assets—rather than pure commercial enterprises—the rules of the game shift in ways that are very difficult to reverse.

Sam Altman appears to be betting that giving Washington a financial stake in OpenAI’s success is cheaper, in the long run, than fighting a government that feels left out of the most transformative technological shift in a generation. Whether that bet pays off will be one of the defining policy questions of 2026 and beyond.