Trump and Sanders Now Want the Same Thing: Government Equity in the AI Labs. The Timing Is the Story.

On Friday, June 6, President Trump told reporters the US government may take direct equity stakes in AI companies like OpenAI, Anthropic, and xAI. "You make them a partnership in this revolution," he said. "It would be a beautiful thing." That came days after Bernie Sanders published a New York Times op-ed calling for a 50 percent public ownership stake in the same companies. The most prominent democratic socialist in the Senate and a Republican president who ran on deregulation are now describing the same policy outcome in different vocabularies.

The convergence is strange enough on its own. What makes it a story I have to write this weekend is the calendar underneath it. SpaceX prices the largest IPO in history on Wednesday. Anthropic confidentially filed its S-1 on June 1 at a $965 billion private valuation. OpenAI is finalizing its own confidential filing with Goldman Sachs and Morgan Stanley for a September window. Washington decided to start a public conversation about owning pieces of these companies in the exact five months they are all trying to sell shares to everyone else.

What Is Actually on the Table

There are three distinct proposals circulating, and they get conflated constantly. They should not be, because they differ by an order of magnitude.

The Sanders version is the most aggressive. His American AI Sovereign Wealth Fund Act, which followed the op-ed and has not yet been formally introduced as of this writing, proposes a one-time 50 percent tax on frontier AI companies, paid in stock rather than cash. The collected shares would sit in a federal sovereign wealth fund with voting rights, board representation at each company, and eventual dividend distributions to the public. His stated justification: the models were trained on the creative output of millions of people who were never asked and never paid, so the public should hold the upside.

The Trump version is vaguer and, for that reason, harder to price. Senior administration officials have held preliminary discussions about government shares in AI firms, per reporting from NOTUS confirmed by CNBC and others. Friday's comments are the first time the president endorsed the concept on camera. No mechanism, no percentage, no legislation. Just a stated openness to the idea from the executive branch that would have to implement it.

The third version is the one the market should study, because it came from inside the building. Sam Altman has been pitching a government equity concept to the Trump administration since early 2025, and revisited it with senior officials this week. OpenAI's own April 2026 policy proposal sketches a "Public Wealth Fund" seeded by donated equity. Donated, not taxed. A sliver, not half. Altman's answer to Sanders, who confronted him directly: he cannot support the 50 percent number, but he wants to work together on the general idea of public participation.

One detail worth flagging: the Sanders bill targets OpenAI, Anthropic, and xAI by description, and conspicuously not Google or Meta. The three named companies are private and heading to market. The two omitted ones have millions of public shareholders who would litigate instantly. That tells you the proposal is aimed at the IPO moment, when ownership structures are still wet cement.

This Administration Already Owns Things

The reflex is to dismiss all of this as theater. I would not. Government equity in private companies stopped being hypothetical under this administration a while ago. The golden share in US Steel that conditioned the Nippon deal. The roughly 10 percent stake in Intel taken in August 2025. The Defense Department's preferred position in MP Materials. Each was unthinkable until it was a term sheet.

So when the president muses about partnership stakes in AI labs, the relevant precedent class is not campaign rhetoric. It is three executed transactions in eighteen months. The mechanics exist, the lawyers have done it before, and the political coalition behind it now spans from Bannon (who called voluntary donations "tip money" and wants the full 50 percent) to the Vermont left.

Anthropic, for what it is worth, is reportedly not participating in any equity discussions with the administration. That is a cleaner posture for an S-1 review. It is also a riskier one if the political weather keeps moving, because the company that did not pre-negotiate is the company that gets terms written for it.

The IPO Window Just Got a New Risk Factor

Here is the calendar these proposals are landing on.

An IPO roadshow is a machine for converting uncertainty into a price. Institutional buyers will discount anything they cannot model, and "a sitting senator wants half your equity while the White House muses about a partnership stake" is close to unmodelable. It does not need to pass to cost money. It only needs to sit in the risk factors section while the bankers build the book.

SpaceX is the immediate test. Its roadshow finished week one on June 6, the same day Trump made his comments, and 30 percent of the float is earmarked for retail through Robinhood, Fidelity, and Schwab. If SPCX prices at target on Wednesday and holds, the market is saying it treats all of this as noise. If pricing comes in soft or the allocation gets restructured, the September and October filers will read that as a political risk premium with a number attached.

I wrote two weeks ago that Altman and Amodei were softening their AI jobs rhetoric because apocalypse framing is an asset when raising private capital and a liability under public-market disclosure rules. The equity-stake conversation is the same dynamic from the other direction. Eighteen months of telling Washington that AI will reorder the entire economy worked. Washington believed it. And institutions that believe a technology will reorder the economy historically do not stay passive shareholders of it.

Why Altman Is Pre-Negotiating

Read Altman's position as a hedge, not a concession. If some form of public participation in AI upside becomes politically inevitable, the company that designed the vehicle gets to set the terms: donated rather than taxed, a sliver rather than half, governance-free fund units rather than board seats. A voluntary 2 percent donation that defuses the pressure for a legislated 50 percent transfer is the cheapest insurance policy in corporate history.

It also buys something an S-1 cannot: political cover for the listing itself. OpenAI is asking public markets to fund a company that lost money at a negative 122 percent operating margin in Q1 while projecting losses through 2029. A populist backlash against AI wealth concentration, arriving mid-roadshow, is a real threat to that ask. A Public Wealth Fund announcement, arriving mid-roadshow, is a counter-narrative.

The cynical read writes itself: both labs spent the spring asking Congress for safety regulation they can live with, and now one of them is designing the equity participation scheme it can live with. Regulatory capture is when you write the rules. This is a step further. This is writing the expropriation.

Our Take

The near-term odds that either the Sanders bill or a Trump executive action puts real government equity in these companies are low. The bill is not introduced, the administration has no mechanism, and the IPO timelines are moving faster than the legislative calendar. Nothing here changes a filing date this month.

But the Overton window does not move back. As of this weekend, public ownership of frontier AI labs is a position held simultaneously by the president, the most visible senator on the left, the MAGA media flank, and the CEO of the largest lab. Nobody prominent is making the opposite case. That is a one-directional ratchet, and every future negotiation between these companies and this government (compute permits, export rules, federal contracts like the $0.42 GSA deal xAI just signed) now happens with equity as a live bargaining chip on the table.

Three signposts for the next ninety days. First, whether the Sanders bill gets formally introduced with co-sponsors, which converts an op-ed into a CBO-scoreable object. Second, whether OpenAI's eventual public S-1 mentions government equity discussions in its risk factors, which it almost certainly must if the talks are material. Third, whether Anthropic's October window slips, which would be the first hard evidence that the politics are pricing into the calendar. We will be tracking all three.