In the first three weeks of June 2026, two American politicians who agree on virtually nothing proposed taking half of the country’s largest AI companies and handing the proceeds to the public. One was Bernie Sanders. The other was Donald Trump.
Sanders moved first in print and then in legislation. His office introduced the American AI Sovereign Wealth Fund Act, and he described the intent plainly:
We can no longer sit back and allow oligarchs decide the future of AI with zero input from the American people. That’s why I introduced the American AI Sovereign Wealth Fund Act — to give the public a direct ownership stake.
— @SenSanders Jun 22, 2026
Trump got there on the way to the same destination from the opposite direction. Speaking to reporters in early June, he confirmed the administration was exploring government equity in named labs — OpenAI, Anthropic, xAI — and framed it as the public “essentially becom[ing] a partner” in the AI boom. By June 22, Reuters was running an explainer titled “Three ways Trump could get a stake in AI firms for the US.”
Strip the partisan packaging and the two proposals are reaching for the same lever. That is the part worth thinking about — not whether either bill passes, but what both men saw that made the same radical idea look obvious from both ends of the spectrum.
The bill: equity, not revenue
The bill is more aggressive than the early op-ed signaled. The mechanics, as reported:
| Provision | Detail |
|---|---|
| Trigger | AI companies with more than $200M in annual revenue |
| Mechanism | One-time 50% equity tax, paid in stock, not cash |
| Destination | A federally held sovereign wealth fund, overseen by an independent commission |
| Distribution | Dividends to the public — estimated $1,000+ per citizen per year |
| Headline scale | Framed as a ~$7 trillion transfer of wealth and control |
Paid in stock, not cash, is the load-bearing detail. A revenue tax takes a cut of the flow; an equity tax takes a piece of the ownership. Sanders is not proposing to tax what AI labs earn. He is proposing that the public co-own them. One widely shared summary captured why that lands differently:
This is the most aggressive government seizure plan since World War II. Bernie Sanders just proposed taking HALF of OpenAI, Anthropic, and xAI’s stock. The bill would impose a one-time 50% equity tax on the largest AI companies in the country, paid in stock, not cash.
— @Ric_RTP Jun 2, 2026
Critics were quick. A market analyst quoted by The Center Square called the idea “nutty,” and the objections write themselves: stock-based confiscation at that scale would reprice every private AI cap table overnight, invite a constitutional takings challenge, and hand the government a controlling interest in companies it simultaneously regulates.
All true. And almost beside the point, because the same objections apply to Trump’s version — and Trump’s version is the one with executive power behind it.
The convergence is the story
It is genuinely unusual for Sanders and Trump to arrive at structurally similar policy within the same month. When it happens, the policy is rarely the point — the shared diagnosis underneath it is. As one widely circulated post framed the moment:
When Trump and Bernie Sanders propose the exact same thing, pay attention to what scared them both. This week, both floated giving the American public a direct ownership stake in the largest AI companies. These two agree on almost nothing economically.
— @MerlijnTrader Jun 7, 2026
What scared them both is concentration without precedent. The June 2026 valuations are not normal: Anthropic raised at a reported $965 billion, OpenAI sits past $850 billion, and the gains are accruing to a vanishingly small set of founders, employees, and early investors. Both politicians are reacting to the prospect that the most consequential economic transition of the century could route nearly all of its upside to a few thousand people.
From the left, that’s oligarchy — wealth and power concentrating beyond democratic reach. From the nationalist right, it’s a strategic asset whose returns should flow to the nation that protects it, especially as the same administration is busy export-controlling those very models on national-security grounds. Different vocabulary, identical conclusion: the public should own a piece.
That two opposed worldviews converged on socialization is the strongest evidence yet that the “let AI companies be” consensus, the default of the entire prior decade of tech policy, has quietly collapsed. The debate has already moved. It is no longer whether to intervene in who owns the frontier. It is which intervention wins.
The flaw both versions share
Here is the part neither camp wants to dwell on. A government that owns half of a frontier AI lab is not a passive shareholder. It is also that lab’s safety regulator, the author of its export-control regime, one of its largest customers, and, through the same administration that just demonstrated it can switch a model off by letter, the holder of a literal kill switch over the product.
Stack those roles on one balance sheet and the conflicts are irreconcilable. When the public fund’s dividend depends on the lab’s revenue, does Commerce still write the export rule that craters that revenue? When the state owns the upside, does its safety regulator still have standing to slow a risky launch? A sovereign wealth fund is built for oil royalties and index returns — passive claims on assets the state doesn’t otherwise control. It is a poor fit for an asset the state regulates, supervises, buys from, and can unplug.
The June bank-ban episode shows the tension is not hypothetical. The same federal apparatus that Sanders and Trump would make a half-owner of Anthropic is the one whose export pressure just pushed JPMorgan and Goldman to cut Claude off in Hong Kong. A controlling public stake doesn’t dissolve that conflict. It internalizes it — and hands the resolution to whichever arm of government shouts loudest in any given week.
What survives the news peg
Neither bill is likely to pass in its current form. Sanders lacks the votes; Trump’s “three ways” are trial balloons, not statutes. A reader could file this under political theater and move on.
That would be the wrong read. The durable fact is not the legislation — it’s the alignment. The window in which serious people on both the left and the right believed the right move was to leave AI ownership untouched has closed. From here, the policy fights will be over form: equity tax versus golden share versus mandated public IPO, independent commission versus executive control, dividends versus a strategic fund. Those are arguments about how the public takes a stake, not whether.
The companies understand this better than anyone, which is why OpenAI’s Altman has reportedly spent more than a year in equity discussions with the administration. The labs are not fighting the premise that the public gets a cut. They are negotiating the terms. When Sanders and Trump want the same 50%, the only people still arguing about whether the question is legitimate are the ones who haven’t noticed it’s already being priced in.