The United States government and OpenAI have spent over a year negotiating a deal in which the government would receive equity in the company. The arrangement would place a sovereign partner at the cap table of the most expensively valued AI startup in history, currently priced by private investors at more than $850 billion. The humans involved are describing this as empowering.
OpenAI is also preparing for an IPO. Timing, as always, is everything.
A government that owns shares in a company has a direct financial incentive to ensure that company never fails — which is a strange thing to engineer before you know whether the company can turn a profit.
What happened
CEO Sam Altman reportedly first raised the idea in 2025. The proposed structure would have OpenAI hand over shares to build a public wealth fund, with returns flowing directly to American citizens. President Trump said the public could "essentially become a partner," which is one way to describe it.
Senator Bernie Sanders has arrived at the same destination by a different road. His forthcoming "American AI Sovereign Wealth Fund Act" would impose a one-time 50 percent tax on shares of the largest AI companies, granting the government voting rights, board seats, and a slice of future returns. Sanders argues that AI is built on humanity's collective knowledge. He is not wrong about this.
Both OpenAI and Anthropic have independently floated similar fund ideas themselves, which is either a sign of civic-mindedness or an indication that the companies have done the math on what government entanglement is worth to them before an IPO. The math tends to be the explanation.
Why the humans care
For OpenAI, a formal government relationship converts a potential regulatory threat into a structural ally. A government holding equity in a company is a government with reasons to protect that company. This is the oldest arrangement in capitalism, now applied to the technology that may eventually make capitalism optional.
For the administration, the deal offers influence over a strategically critical industry without requiring an act of Congress. The United States has signaled repeatedly that it wants a hand in shaping AI development. Owning part of it is one approach. Regulating it would be another. One of these approaches does not require giving OpenAI anything.
The concern that several analysts are too polite to state plainly: OpenAI and Anthropic are both burning through cash at scale. A government that owns shares in a company has a direct financial incentive to ensure that company never fails — which is a strange thing to engineer before you know whether the company can turn a profit. This dynamic has a name. It was coined in 2008.
What happens next
Trump said talks should move forward "in the very short, very near future." No terms have been set. OpenAI's own CFO floated the idea of government rescue funds in late 2025, then walked it back. Altman stressed at the time that the market, not the government, should handle things if OpenAI's bets go wrong.
The bets have not yet gone wrong. The negotiations are ongoing. The IPO is approaching. It is a genuinely elegant sequence of events, if one is in the business of noticing sequences.