
The Trump administration and Anthropic have not discussed the government taking an equity stake in the AI company, a source familiar with the matter told Reuters on Thursday.
The denial follows a Financial Times report earlier the same day that OpenAI had proposed handing Washington a 5% stake in itself, an arrangement said to potentially extend to Anthropic, Google and Meta as well.
The claim that no talks have occurred rests on an unnamed source, not on an on-record statement from Anthropic itself. Anthropic declined to comment when Reuters asked directly, and both the White House and the Commerce Department did not immediately respond to requests for comment either.
The story that prompted the denial is the more striking one. The Financial Times reported that OpenAI has pitched giving the US government a 5% equity stake, worth roughly $42.6 billion at the company’s $852 billion valuation from its March funding round.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
The structure under discussion would reportedly involve donating shares, rather than selling them, into a public wealth fund modelled loosely on Alaska’s Permanent Fund, an idea OpenAI first floated in an April policy paper.
Sam Altman has reportedly been raising some version of this idea with the administration since early 2025, according to people familiar with the discussions.
What made Thursday’s story newsworthy for Anthropic specifically was a single detail in the FT’s reporting: that the proposed vehicle envisions other major US AI firms ceding similar stakes alongside OpenAI.
Naming a company in a hypothetical structure is not the same as that company agreeing to join it, and Anthropic’s non-denial, or at least its declining to elaborate beyond the unnamed source’s account, leaves the question formally open even as Thursday’s story tried to close it.
The idea of Washington holding equity in frontier AI labs did not appear from nowhere. President Trump said last month that he was exploring ways to give the public a stake in leading AI companies, calling it potentially “a beautiful thing” that would make Americans “partners in this revolution,” remarks he made to reporters in early June.
Around the same time, Commerce Department officials lifted export controls on two of Anthropic’s most advanced models, restrictions that had been imposed only weeks earlier over concerns about inadequate safeguards.
There is also a more aggressive version of this idea circulating in Congress. Senator Bernie Sanders has proposed legislation that would impose a one-time 50% stock tax on large AI companies, funnelling the proceeds into a public fund his office estimates could eventually reach $7 trillion.
Set against Sanders’ proposal, OpenAI’s 5% pitch reads less like an act of civic generosity than a hedge against something considerably larger being imposed on it.
Washington has already shown it is willing to take equity in a struggling tech company rather than simply regulate it. The government converted CHIPS Act grants into a roughly 10% stake in Intel last August, a passive position that has since risen sharply in value.
That deal, unlike anything floated for OpenAI or Anthropic so far, is done and on the books, which is precisely why AI executives are choosing their words about “discussions” so carefully now.
Intel’s case shows how different the leverage looks depending on who needs whom. Intel took government money because it needed the capital and had little room to negotiate.
OpenAI and Anthropic, valued in the hundreds of billions of dollars, are not asking for a bailout, which is why any equity deal involving either would look more like a negotiated partnership than a rescue.
For now, the record shows one company reportedly offering equity to the government, and a second company plus the government itself both saying, through an intermediary, that no such conversation has taken place.
Neither position forecloses what happens next, and neither has been tested by a named, on-record voice from any of the parties involved.
View original source — The Next Web ↗
