Sam Altman walked into the Oval Office in early 2025 and pitched President Donald Trump on the idea of giving the American government an equity stake in OpenAI. That conversation, which took place before most of Washington had fully processed what the AI boom was going to mean for wealth distribution, has been developing quietly ever since. On June 5, Trump confirmed the talks aboard Air Force One, telling reporters there are “concepts where pieces could be given to the American public, where the American public essentially becomes a partner” with the companies building artificial intelligence.
The confirmation landed in the middle of a sharp Big Tech selloff and reignited a debate that has been building for months: whether Washington is drifting toward something that looks considerably more like state capitalism than the free-market model American technology policy has operated on for decades. The government currently holds ownership stakes in roughly 20 private companies across minerals, semiconductors, and quantum computing. The Intel deal — a 9.9% stake worth roughly $8.9 billion, converted from CHIPS Act grants — is the template everyone is pointing to. Whether the same model applies to artificial intelligence, and which of several competing proposals actually gets implemented, are the questions the industry and Washington are now actively working through.
The Three Models on the Table
Three main mechanisms are being discussed, and they are worth understanding separately because they produce very different outcomes for who controls what.
The first is a stock tax: most aggressively proposed by Senator Bernie Sanders, whose American AI Sovereign Wealth Fund Act would impose a one-time 50% levy on OpenAI, Anthropic, and xAI, payable in shares rather than cash. The proceeds would go into a sovereign wealth fund for the public. Two law professors at George Washington University have proposed a similar but less dramatic version, a stock-based tax that transfers equity without requiring public investment and without giving the government a controlling stake. The Sanders version would effectively bring the companies under substantial federal influence. The law professor version is considerably more modest.
The second is equity in exchange for federal funding: the Intel model applied to AI infrastructure. The tech sector needs enormous, regular injections of capital for data centers, chip plants, and computing infrastructure. OpenAI has discussed federal loan guarantees for chip plants with the administration. Alphabet announced it would increase its equity offerings to $84.75 billion this month. If the government becomes a funding source for this infrastructure buildout, equity stakes become the natural compensation, the same logic that produced the Intel arrangement, the MP Materials deal where the Pentagon took a 15% stake in exchange for $400 million, and the broader pattern of direct government investment across strategically sensitive industries that has been building since 2025.
The third is voluntary donation — OpenAI’s own preferred model. The company’s April policy paper proposed that AI firms donate equity to seed a nationally managed Public Wealth Fund, which would invest in diversified long-term assets across AI companies and distribute proceeds to citizens. The parallel OpenAI draws is to the Alaska Permanent Fund, which distributes annual dividends to state residents from oil revenues. The company frames the public data that trained its models as a form of public infrastructure that justifies public ownership of a share of the returns. Under the framework being discussed, OpenAI would donate equity voluntarily rather than having it extracted through taxation or conditioned on government funding.
Why This Is Different From the Intel Deal
The government taking stakes in American companies is not new. TARP in 2008 put the Treasury into banks and automakers during the financial crisis, but those were emergency rescues explicitly designed to be unwound, and they were. The current wave is different in kind. These are healthy, often dominant companies at the top of one of the most valuable industries in history, and the government is not rescuing them from failure. It is positioning itself as a permanent co-owner of their future upside.
That distinction matters enormously for what comes next. A government stake in a company it is rescuing creates pressure to exit as quickly as possible, restoring normal market conditions. A government stake in a dominant AI company creates a permanent relationship between a regulator and the entity it is supposed to regulate, with financial interests now entangled in the regulated company’s success. Critics from the right, including Jennifer Huddleston of the Cato Institute, have called this picking preferred companies and warned it intrudes on traditional principles of private enterprise. David Sacks, who recently stepped down as Trump’s AI and crypto czar, said he can see why the idea resonates across the political spectrum but warned it would accelerate the corporate-government fusion already underway.
The conflict of interest concern is not theoretical. A government that holds equity in OpenAI has a financial incentive for OpenAI to succeed, which shapes how it regulates competitors, how it allocates spectrum and data access, how it responds to antitrust complaints, and how it handles the national security questions around AI that NSPM-11, signed on June 5, has put squarely on the policy agenda. The government became OpenAI’s largest shareholder in the Anthropic dispute, where the Pentagon designated Anthropic a supply chain risk, and the equity entanglement would make navigating similar disputes considerably more complicated.
The Model US Mocked China For, Is Now American Policy
Every major competitor to the United States in AI, China most notably, operates with deep state involvement in its AI sector. The Chinese government shapes which companies receive capital, which research priorities get funded, which international markets are accessible, and which capabilities get integrated into national security infrastructure. American AI policy has historically operated on the opposite premise: the government sets the rules and the market decides who wins. The equity stake idea represents a meaningful departure from that premise, moving Washington toward a model where it holds financial interests in the companies it is simultaneously regulating, funding, and depending on for national security capabilities.
That convergence toward state involvement in strategic technology is happening across allied governments too. France has been investing directly in AI champions through BPIFrance. The UAE’s sovereign wealth funds hold positions in AI infrastructure globally. Saudi Arabia’s Public Investment Fund is building AI capacity through direct state ownership. If the United States moves toward government equity stakes in its frontier AI companies, it is joining a global trend rather than initiating one, but it is doing so in a way that would reshape the American technology model more fundamentally than any policy change since the postwar science funding framework was established.
The bipartisan nature of the interest is itself significant. Trump and Sanders arriving at the same general conclusion from completely different starting points; Trump from a populist instinct that taxpayers should benefit from government-supported technology, Sanders from a socialist critique of concentrated AI wealth, suggests this is not going away regardless of which party controls Washington after November. The specific mechanism will be contested fiercely. The underlying direction appears to have genuine political momentum on both sides.
What Happens to the IPOs
OpenAI is targeting a valuation of up to $1 trillion for its IPO, currently expected as early as September 2026. Anthropic filed confidentially this month. SpaceX went public this week at $1.8 trillion. The equity stake discussions are happening simultaneously with the most significant wave of AI company public listings in history, and the two processes interact in ways that complicate both.
If the government secures an equity stake in OpenAI before its IPO, that stake becomes publicly traded, creating a situation where the US Treasury holds a position in a company that other government agencies regulate, fund through contracts, and depend on for classified AI capabilities under NSPM-11. The valuation implications cut both ways: government involvement could boost investor confidence by signaling strategic importance, or it could depress valuations by introducing regulatory entanglement and political risk that private investors find difficult to price. Either way, the government becomes a party whose interests have to be managed alongside those of ordinary shareholders, institutional investors, and the company’s own board.
The legal mechanism for any of this remains unclear. The White House declined to comment on the specifics. The people familiar with the talks cautioned the deal may ultimately not come together. What is clear is that the conversation has moved from think-tank whiteboards to the Oval Office, and the precedent set by Intel, MP Materials, and the pattern of direct government investment across strategically sensitive industries suggests the administration is operating with a systematic rather than ad hoc logic. Whether AI is next in that sequence, and in what form, is the question that will define the next phase of American technology policy.
