OpenAI Is About to Have Washington on Its Cap Table

The Trump administration is reportedly considering taking an equity stake in OpenAI. Not a regulatory relationship. Not a research contract. An equity stake — the same kind you get when you wire money to a VC fund.

According to Bloomberg reporting covered by TechCrunch, Trump has said he's discussing "concepts where pieces could be given to the American public, where the American public essentially becomes a partner." CNBC has confirmed the discussions are real. And Sam Altman, OpenAI's CEO, has reportedly been pushing the idea of a government stake in major AI companies since early 2025. The subject is helping the buyer. That's the first unusual thing.

The second is the Intel precedent. Last August, the Trump administration acquired a roughly 10% stake in Intel, funded through CHIPS Act money and Defense Department commitments. "I said, I think you should pay us 10% of your company," Trump said afterward. "They said yes." The Intel deal was framed around semiconductor manufacturing: strategic asset, national security rationale, legible narrative. AI is a different category. Intel makes chips. OpenAI makes systems that write, reason, and increasingly execute autonomous processes. The question of who controls that is not the same question.

Governments have always moved toward transformative infrastructure. Railroads got land grants and then regulation. Telecom got licensed spectrum. Nuclear power was built under government monopoly before private capital was allowed near it. Electricity became so intertwined with public interest that many countries nationalized it. The pattern is consistent: transformative technology, followed by recognition that it's too important to leave entirely to private actors, followed by some form of public involvement.

Equity is a different form of involvement than regulation. Shareholders want returns. Shareholders lobby for market access. They sit on earnings calls and ask about growth trajectory. A government that owns part of OpenAI has a financial interest in OpenAI succeeding as a business. That's not the same as a government that regulates OpenAI in the public interest. Those interests can align. They can also diverge. The question nobody is articulating clearly: when they conflict, which one wins?

Think through a specific scenario. OpenAI restricts its models in certain countries, citing safety concerns or export controls. A government shareholder now has a new pressure surface. The case for restriction is cleaner when there's no equity relationship to navigate. Add a shareholder to the mix and the calculation gets murkier.

We've already watched AI labs negotiate their government relationships with the Defense Department. Those negotiations mostly ended with usage policies and access agreements. An equity stake goes further. It doesn't just create a relationship between Washington and OpenAI. It creates a financial incentive structure.

OpenAI's own governance is already complicated. The company spent the last year converting from a nonprofit to a for-profit structure, a transition that generated genuine tension about its stated mission to develop AI "for the benefit of humanity." Adding a government shareholder doesn't simplify that. It adds another stakeholder whose interests aren't automatically aligned with anyone else's.

The bipartisan dimension is worth noting. Senator Sanders has introduced the American A.I. Sovereign Wealth Fund Act, which proposes a one-time 50% stock tax on major AI companies paid as shares, depositing the equity into a public fund with voting rights and board representation. The mechanism differs from what the Trump administration is considering. The underlying assumption is identical: private ownership of frontier AI isn't a settled question.

When Anthropic went public at a valuation approaching $965 billion, the question of who should benefit from AI's economic gains became concrete rather than theoretical. Sanders is answering it one way. The Trump administration is answering it with a more selective version of the same idea.

Maybe this is the right direction. The infrastructure comparison isn't empty. If large language models become as foundational as electricity or telecom, the question of who controls them and who benefits from them is legitimate. The Public Wealth Fund concept — government equity converted into citizen dividends — isn't obviously wrong as a mechanism for distributing gains that would otherwise accrue to a narrow group of shareholders.

But equity isn't regulation. Regulation creates rules. Equity creates alignment of interest. Washington buying into OpenAI isn't the same as Washington overseeing it. A regulator with no financial stake in the outcome can act on public interest grounds. A shareholder has to consider what the outcome does to the investment.

Washington figured out that regulating AI is hard. The technology moves too fast. The political economy of slowing it down is difficult. So they're buying in instead. Whether that's pragmatic governance or a quiet admission that effective oversight was never really on the table depends entirely on whether government shareholder interest and public interest end up pointing in the same direction.

That's not a guarantee I'd assume.

---

If you're thinking about AI's impact on the economy, society, and your daily life, the About.chat Weekly newsletter covers it every week — no hype, just analysis.