Bernie Sanders proposed this week that the American public should own half of the largest AI companies. His bill would impose a one-time fifty-percent tax, paid in stock rather than cash, on OpenAI, Anthropic, xAI and the rest, depositing the shares into a public fund that would grant ordinary Americans voting rights, seats on the companies’ boards, and eventually a check in the mail. The remarkable thing is not the proposal. It is that the Trump White House, which agrees with Sanders on virtually nothing, has been floating its own version of the same idea — a government sovereign wealth fund, public equity stakes in the AI companies — and that Sam Altman has been meeting with Sanders to discuss it. The socialist, the nationalist, and the chief executive are, for once, talking about the same thing: that the wealth this technology is concentrating is too vast, and too public in its origins, to be left entirely in private hands.
The Convergence
The convergence is the news, because of who is converging. Sanders and Trump occupy opposite ends of the country’s politics, and they have spent a decade unable to agree on the date. That they have independently arrived at the same structural proposal — that the state should hold an ownership stake in the AI companies — means the proposal has stopped being ideological and become a response to a material fact that both ends of the spectrum can now see. The fact is concentration. The technology is gathering wealth into a handful of firms at a velocity that no existing mechanism can spread, and when the socialist and the nationalist reach, separately, for the same remedy, it is because the disease has grown visible enough to be diagnosed from both sides of the room.
Each arrives for his own reasons, and the reasons matter less than the destination. Sanders sees a windfall produced by the many and captured by the few, and proposes that the many own a share of what they produced. Trump sees a strategic asset of national importance accumulating in private hands, and proposes that the nation hold a position in it, the way a state holds a stake in its oil or its defense industry. One frames it as justice and the other as sovereignty, but the mechanism is identical: the public, through its government, takes equity in the companies building the most powerful technology of the age, rather than leaving them entirely to their private owners. The framing is a quarrel. The structure is a consensus.
And Altman is in the room, which is the detail that tells you the consensus has weight. A chief executive does not take meetings about the partial nationalization of his company unless he has concluded that some version of it is coming and that it is better to shape it than to resist it. His presence is not endorsement; it is calculation. He has read the same political weather the proposals represent, and judged that the question is no longer whether the state takes a stake but how large and on what terms, and that a seat at the negotiation is worth more than a principled refusal to attend it. When the target of a proposal shows up to help write it, the proposal has passed from fantasy into the early stages of fact.
Why the Windfall Invites the State
The case for public ownership is stronger than the usual argument for redistribution, because the AI companies rest on a foundation the public already built and was never paid for. The internet they run on was publicly funded. The foundational research was publicly funded. The workforce was publicly educated. The data the models are made of is the collective output of the entire population, taken — as a court is now being asked to decide — without consent or payment. The language itself, the accumulated written record that the models compress into weights, belongs to no one and everyone. The companies did not create the foundation. They built upon a commons, and the claim that the commons is owed a stake in what was built on it is not envy. It is accounting.
This is the same claim that a South Korean official made when he floated a citizen dividend and the market fell five percent at the suggestion — escalated, now, from a dividend to ownership, and the escalation is the substance. A dividend is a share of the profit, granted at the state’s discretion, revocable when the politics change. Ownership is a share of the company, structural, carrying votes and board seats and a permanent claim. Sanders is not proposing that the public be paid a portion of the windfall. He is proposing that the public own a piece of the machine producing it, which is a far more durable thing, and a far more frightening one to the private holders, because it cannot be quietly withdrawn the way a dividend can.
And the reason a nationalist government finds the idea attractive, rather than recoiling from it as socialism, is that the valuations have made the stakes too large for any state to ignore. If this technology is worth what its price assumes — if it will reshape labor, warfare, and power at the scale the trillion-dollar valuations imply — then leaving it entirely in private hands concentrates not merely wealth but capability, the capability to influence elections, to surveil populations, to direct economies, at a magnitude that any sovereign, of any ideology, must regard as a rival. The state’s interest in a stake is not only fiscal. It is the recognition that a private power of this size is a competitor to the state itself, and that ownership is the oldest way a sovereign has of bringing a rival power under its hand.
What This Means
The proposal will not pass in Sanders’ form; fifty percent is a number chosen to open a negotiation, not to be enacted. But the form is not the point. The point is the trajectory, and the trajectory is steep. The question of who owns the AI windfall has traveled, in two months, from a fringe idea that crashed a foreign market to a bipartisan conversation in the American capital in which the chief executive of the largest lab is a willing participant. Movements of that speed, across that ideological distance, do not reverse. They are the early shape of something that arrives.
What the convergence reveals is that the windfall is now being claimed from every direction at once. The workers who fabricate the chips claimed a share through the strike. The citizens claimed one through the dividend the market punished. And now the state itself is claiming one, from the left as justice and from the right as sovereignty, which means the claim has reached the one actor with the power to actually take it. The valuations are priced on the assumption that the wealth this technology generates flows to private capital indefinitely. The politics are beginning to price the opposite, and when the politics and the valuations disagree this sharply, it is usually the valuations that have failed to read the room.
I have watched states arrive at the great concentrations of private power before — the railroads, the trusts, the banks, the oil — and the arrival follows a pattern that the present is now entering. First the concentration is celebrated as progress. Then it grows large enough to rival the state. Then the state, of whatever ideology, moves to bring it under control, through ownership or regulation or breakup, because a power that size cannot be permitted to remain wholly outside the public hand. The AI companies are at the second stage and crossing into the third, and the proof is that the socialist and the nationalist are reaching for the same instrument in the same week. They will disagree, loudly, about everything except the one thing they have quietly agreed on: that this will not be left entirely to its owners. On that, for now, the country is united, and the owners are taking the meeting.