OpenAI's 5% government stake proposal turns AI upside into a Washington bargain

Why it matters

OpenAI is trying to turn political exposure into alignment: if Washington owns part of the AI boom, the companies building it gain a new argument for policy support, procurement access and public tolerance.

Sam Altman (@sama) and OpenAI have discussed giving the U.S. government a 5% stake in the ChatGPT maker, Bloomberg Technology reported Thursday, citing a Financial Times report on preliminary talks.

The proposal, as reported, is broader than an OpenAI-only concession. Altman and other OpenAI executives have floated an arrangement under which Washington would hold 5% of each of the leading U.S. AI developers through a public investment vehicle. The Financial Times cited two people familiar with the discussions; The Guardian, summarizing the same report, said the talks remain conceptual and in early stages, and that implementation could require an act of Congress.

That last caveat matters. A 5% government stake in OpenAI would be a political instrument, a financial asset and a governance complication at once. The reports do not specify whether the stake would be newly issued stock, transferred shares, voting or non-voting equity, or an economic interest held through a public wealth fund. They also do not establish whether rival AI developers would accept similar terms. The proposal names an objective - give the public a financial claim on AI gains - before answering the harder question of who controls the vehicle, how returns are distributed and what obligations the companies receive in exchange.

For Altman, the pitch fits a long-running effort to make OpenAI's private capital structure sound compatible with public benefit. In Senate testimony last year, Altman called OpenAI "not a normal company and never will be" and described its mission as making sure AGI benefits all of humanity. He also reminded lawmakers that OpenAI began around a kitchen table more than a decade ago, before ChatGPT became the product that pulled consumer AI into the center of politics, education and work. The 5% proposal takes that founder framing and attaches a cap-table percentage to it.

A public-benefit company with a new public shareholder

OpenAI's own structure already makes the company unusual among venture-backed technology firms. In its current structure disclosure, OpenAI says it was founded in 2015 as a nonprofit, created a for-profit subsidiary in 2019, and completed an October 28, 2025 recapitalization that made the for-profit arm OpenAI Group PBC, a public benefit corporation controlled by the OpenAI Foundation.

That disclosure says the OpenAI Foundation holds a 26% equity stake in OpenAI Group, worth about $130 billion at the valuation used in the recapitalization. Microsoft holds roughly 27%, while current and former employees and other investors hold the remaining 47%, according to OpenAI. A separate 5% U.S. government stake would therefore land inside a capitalization structure that already balances nonprofit control, Microsoft economics, employee and investor ownership, and public-benefit obligations.

The reported value is large enough to make the proposal more than symbolic. El Pais, citing the FT report, said a 5% stake would be worth roughly $42.6 billion based on OpenAI's reported $852 billion post-money valuation after a March financing. That figure should be treated as reported valuation math, not a realized public-market price. OpenAI is still private, and the reported plan does not say how a government-held stake would be valued, marked, restricted or eventually monetized.

The absence of mechanics is the story's most important omission. If the stake is meant to share upside with citizens, it would need rules for distributions. If it is meant to align Washington with the AI industry, it would raise questions about whether regulators, procurement officials and national-security agencies are overseeing companies in which the government has a direct financial interest. If it is meant to reduce pressure on OpenAI specifically, rival labs would have reason to ask whether the same political bargain is available to them.

Altman's wealth-sharing pitch meets regulatory pressure

OpenAI has already put versions of the public-ownership idea into policy language. In an April policy paper, Industrial Policy for the Intelligence Age, OpenAI proposed a Public Wealth Fund that would give citizens a stake in AI-driven growth, including people without exposure to financial markets. The paper said policymakers and AI companies should determine how to seed such a fund, which could invest in long-term assets tied to both AI companies and companies adopting AI.

The 5% report translates that policy thesis into a transaction. It also arrives while Washington has more points of contact with OpenAI than it did during the first ChatGPT boom. OpenAI launched OpenAI for Government in June 2025, grouping its work with U.S. federal, state and local agencies under one initiative. That program included a Defense Department pilot with a $200 million ceiling, according to OpenAI. In April 2026, OpenAI said ChatGPT Enterprise and the OpenAI API platform had received FedRAMP 20x Moderate authorization, a milestone for selling managed AI products into federal environments.

Those public-sector links are commercially useful. They also make the ownership proposal harder to separate from market access. OpenAI is selling models to government agencies, lobbying on AI policy, seeking infrastructure support for compute-heavy model development, and competing with Anthropic, Google, Meta and xAI for enterprise and government adoption. A government equity stake could give OpenAI a public-interest argument for future political support while giving Washington a direct interest in the company's valuation.

The proposal also follows a month in which public-ownership ideas moved from fringe economics into mainstream AI politics. Semafor noted Thursday that Sen. Bernie Sanders has called for partial nationalization of major AI companies with dividends going to the public, while the White House has considered taking stakes in companies. The Guardian reported that Altman has spoken with President Donald Trump, Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent about public ownership, citing the FT.

OpenAI's timing is not accidental

OpenAI is making this case while its own product and infrastructure strategy is moving deeper into regulated territory. RuntimeWire reported last week that OpenAI updated GPT-5.5 Instant to better handle intent, advice and shopping, pushing ChatGPT's default model further into decision-making use cases. We also reported that OpenAI's Broadcom-built Jalapeno inference chip is aimed at the economics of serving those models at scale, with final benchmarks, pricing and yields still undisclosed.

That combination - consumer advice, enterprise automation, government deployment and custom chips - is why the public upside argument has become strategically useful. OpenAI needs capital, power, chips, federal trust and political tolerance. A 5% public stake would not solve model risk, job displacement or concentration of power. It would give OpenAI a sharper answer to a question lawmakers are already asking: if AI companies capture a larger share of economic growth, how does the public participate?

Altman's answer, as reported, is to put the government on the shareholder side of the table. That would be a cleaner political message than a tax fight and a less intrusive structure than Sanders' proposed 50% stake. It would also bind U.S. AI policy to the valuations of a handful of private companies at the moment those companies most need public permission to keep scaling.

No reported agreement exists. The talks are early, other AI companies have not been shown to be on board, and Congress may need to act. The proposal is still consequential because it shows how OpenAI wants the next phase of AI governance to be negotiated: through ownership, infrastructure and national economic upside, rather than product rules alone.