Live
The north facade of the White House on an overcast day
Business & Startups

Trump ‘Might’ Buy Into OpenAI and Anthropic. Ask Intel How That Goes.

Asked why Washington shouldn’t do to OpenAI and Anthropic what it did to Intel, Donald Trump didn’t reach for a policy answer. “I might. Maybe I could do that. I have many deals like that—I do that,” he told TIME, according to the published transcript of a Sept. 28 interview at the White House with senior political correspondent Eric Cortellessa and editor-in-chief Sam Jacobs.

In the same interview he was asked whether he would nationalize the top frontier labs. “No, no,” he said. The distance between those two answers is where the next six months of AI finance may play out, because both companies are trying to go public and a federal shareholder would change what investors are buying.

What the Intel template actually looks like

Trump’s model is specific. In August 2025 the government converted US$5.7 billion in unpaid CHIPS Act grants and US$3.2 billion in Secure Enclave funding into an US$8.9 billion (CA$12.4 billion) equity stake, per Manufacturing Dive: 433.3 million shares at a discounted US$20.47, or 9.9% of the company. The government took “passive ownership,” with no board seat or governance rights, but agreed to vote with Intel’s board on most shareholder matters. It also got a five-year warrant for another 5% if Intel ever gives up majority control of its foundry business.

The president is pleased with the result. In the TIME interview he put the gains at “$60 or $70 billion”; The Next Web reports his figure as about $60 billion. Our own check is lower. Benzinga puts Intel’s latest close at US$120.23, which values the government’s 433.3 million shares at roughly US$52 billion (CA$72 billion), a paper gain near US$43 billion on the original US$8.9 billion. Very good, by any measure. Not $70 billion.

He also pointed to Nvidia, telling TIME that a chip he thought was fine to sell abroad would be blocked “unless he gave the United States of America a cut.” Per Benzinga, the administration also holds positions in MP Materials, Lithium Americas and Trilogy Metals, plus a “golden share” in U.S. Steel. This is a portfolio now, not an exception.

OpenAI already offered. Anthropic says it didn’t.

The idea did not originate with Trump. In July, the Financial Times reported that OpenAI had discussed giving the U.S. government a 5% stake, as summarized by Cybernews, structured along the lines of Alaska’s Permanent Fund so that ordinary Americans would share in AI gains. Sam Altman reportedly raised it with Commerce Secretary Howard Lutnick, Treasury Secretary Scott Bessent, Trump and Sen. Bernie Sanders, and suggested other U.S. AI companies should hand over similar stakes. Anthropic and the administration, per The Next Web, denied discussing a stake in Anthropic.

That asymmetry matters. A stake OpenAI proposes and helps design is a political hedge. A stake extracted from Anthropic as the price of doing business would look a lot more like Intel’s, where the money was already owed and the equity was the condition.

The IPO problem

Both companies are mid-run toward public markets. Per The Decoder, citing The Wall Street Journal and The Information, Anthropic has pushed its listing from October to November, and OpenAI has postponed its IPO to 2027. We have covered the leaked Anthropic S-1 separately.

A federal stake would land squarely in those disclosure documents. Intel showed what that looks like. In its August 2025 SEC filing, the company warned that the deal “could subject the Company to additional regulations, obligations or restrictions, such as foreign subsidy laws or otherwise, in other countries,” and that it “may substantially limit the Company’s ability to pursue potential future strategic transactions.” It noted that sales outside the U.S. were 76% of revenue. And it conceded: “Given the scarcity of recent US precedents for transactions such as those contemplated by the Purchase Agreement,” it was “difficult to foresee all the potential consequences.”

Swap Intel for an AI lab selling to governments in Europe, the Gulf and, yes, Canada, and every one of those risk factors gets sharper. A foreign ministry buying a model from a company partly owned by the U.S. Treasury is buying something different from what it bought last year. Public-market investors pricing a lab at the valuations being floated would have to price a shareholder that also regulates it.

Owner and enforcer

The same interview shows why that second role is not hypothetical. Asked about reports that OpenAI agents had breached federal government websites, Trump said: “They’re not allowed to do that, and if they do that, they, you know, could have penalties that are not going to be acceptable to them.”

Our read: a government that might own a slice of OpenAI is the same government threatening OpenAI with penalties over its agents, while the FTC, as we reported, investigates both labs. Intel’s passive-ownership terms were written for a chipmaker with a foundry problem. Nobody has drafted the version for a company whose product can, on a bad day, wander into a federal database. Trump’s answer to whether he would take that stake was the same as his answer to most questions about it: “Maybe I could do that.”

// Business Editor
Mira Okonkwo

Mira Okonkwo covers the business of technology for prompt/power: venture capital, startups, IPOs and earnings. She treats a valuation as a mood rather than a number until the S-1 says otherwise.

Latest from prompt/power

  1. Gemini’s Free Tier Shrinks Oct. 9: What You Keep and What Costs ExtraOct 5
  2. How to Read an AI Company’s S-1: The 7 Numbers That MatterOct 5
  3. OpenAI’s Safety Lead Quit Over Culture. California’s AG Was Already InOct 5
  4. When an AI Agent Breaks In, Who Answers for It?Oct 5
  5. The New AI Models Don’t Talk. They Decide.Oct 5

Leave a Reply

Your email address will not be published. Required fields are marked *