Why OpenAI Wants the US Government as a Shareholder

OpenAI has been quietly pitching an unusual idea to the Trump administration: give the US government a 5% ownership stake in the company, no purchase required. At OpenAI’s March valuation of $852 billion, that slice would be worth roughly $42.6 billion. Sam Altman reportedly took the pitch directly to President Trump, Commerce Secretary Howard Lutnick, and Treasury Secretary Scott Bessent, and he’s framed it as bigger than just OpenAI — he wants every major US AI lab to hand over a comparable slice of equity to a shared public fund.

The model he’s citing is the Alaska Permanent Fund, the sovereign wealth fund Alaska set up in 1976 to invest its oil revenue and pay yearly dividends to residents. It currently holds about $91 billion and has paid out every year since 1982. Altman’s version would do something similar with AI: instead of oil money flowing to a state, a slice of the country’s leading AI companies would flow into a vehicle that, in theory, lets the public share in whatever wealth these companies generate.

It’s worth being clear about what this is and isn’t. It isn’t a done deal, or even close to one. People familiar with the discussions describe them as early-stage and conceptual, and any real implementation would likely need an act of Congress — this isn’t something a handshake in the Oval Office can finalize. Google, Meta, and Anthropic, the other labs whose participation the plan depends on, haven’t signaled any interest in joining. So far this is a proposal from one company, aimed at building political momentum, not an agreement in motion.

Why now? The timing lines up with a rough few weeks for OpenAI in Washington. The company’s newest model, GPT-5.6, got held up by a national security review before its release, a reminder that the government now sees frontier AI systems as something it needs to vet before they ship. OpenAI is also facing a trade-secrets lawsuit from Apple over hundreds of former Apple employees who now work there — covered here in our look at that case — plus ongoing scrutiny over its nonprofit-to-for-profit restructuring and the eye-watering capital commitments behind its data center buildout. Offering the government a direct financial stake reads less like philanthropy and more like a company trying to convert a regulator into a stakeholder with skin in its success.

That’s the part worth sitting with. If the federal government owns 5% of OpenAI, it also has a financial interest in OpenAI doing well — in its valuation climbing, in fewer inconvenient regulations landing on it, in competitors facing tougher scrutiny instead. That’s a fundamentally different relationship than government-as-referee. We’ve already seen versions of this tension play out: government agencies buying AI services from companies they’re supposed to be independently evaluating, as in the Pentagon-Anthropic email exchange we wrote about last week. Turning the government into a shareholder doesn’t create that conflict of interest — it makes it official policy.

There’s a more sympathetic reading, too, and it’s not crazy. AI companies like OpenAI are, by their own account, built on decades of publicly funded research: government grants, university labs, open datasets, the internet itself. A public equity stake is one way to make the public’s claim on that value explicit rather than leaving it to trickle down through taxes on corporate profits that may or may not materialize, or through vague promises about “shared prosperity.” The Alaska model has actually worked for nearly fifty years — Alaskans get a real, predictable check every year, funded by a resource extracted from public land. Applying that logic to AI, a technology many argue is extracted from public knowledge, has a certain internal consistency.

But the differences matter more than the analogy. Oil is a physical commodity with a market price; a company’s equity value is a bet on future performance, heavily influenced by the very regulatory environment its new part-owner controls. Alaska’s fund doesn’t have veto power over how much oil Exxon can pump, or influence over safety rules that affect Exxon’s stock price. A government holding equity in OpenAI would have exactly that kind of leverage, in both directions — it could tighten rules on AI safety and inadvertently shrink its own portfolio, or loosen them and pad it. Either way, the incentive is no longer purely public interest.

None of this is likely to resolve quickly. A deal spanning multiple AI companies, run through a new sovereign-style vehicle, with congressional approval required, is the kind of thing that gets discussed for years before anything concrete happens, if it happens at all. But the proposal itself tells you something about where the industry currently sits: flush with valuation on paper, increasingly entangled with the government it needs both as a customer and as a regulator, and looking for ways to make that entanglement work in its favor rather than against it. Worth watching less for whether the 5% figure survives, and more for whether “the government as AI shareholder” becomes a normal way to talk about resolving AI policy disputes.