Universal Basic Capital Is the AI Ownership Mirage
When a San Francisco landlord raises the rent, the number typically arrives from a pricing algorithm, not from a human weighing the market. The software scans thousands of comparable listings, detects what tenants can be made to absorb, and returns a figure presented as if it were physics. The tenant stares at the screen and finds no chain of reasoning to argue with. There is no negotiation, because there is no counterparty. There is only a recommendation that has quietly become a verdict. This is the boundary of the machine age: the algorithm recommends, and the human consents.
Something similar is now happening at the scale of an entire economy. The artificial intelligence that sets rents and prices is also producing fortunes large enough to reorder a city and, if some politicians get their way, to hand every American a share of the proceeds. President Donald Trump and California Governor Gavin Newsom agree on at least this: Americans should own a piece of the AI boom. Senator Bernie Sanders agrees. Vivek Ramaswamy agrees. Even Sam Altman, the chief executive of OpenAI, agrees. A consensus this broad would ordinarily be the starting point of a serious policy. Here it is the end of one.
The New Magic Phrase
Universal basic income had the virtue of being almost embarrassingly simple. Give every American a stipend, say 500 dollars a month, with no strings attached, and let people spend it on rent, groceries, child care, or nothing at all. Pilot programs across the country produced results that were modest but consistently encouraging. Recipients did not stop working, and the money did not corrupt them. They fixed their cars, paid down debt, and, in the case of the Stockton experiment, moved into full-time work at slightly higher rates than the control group (Stockton Economic Empowerment Demonstration, 2021). [4] The idea seemed ready for its moment. Then the AI boom changed the terms of the argument.
Basic income, its critics and even some of its friends began to say, is redistribution. It waits for technology companies to generate wealth, then takes a sliver of it in taxes and hands it back as a monthly check. That leaves the companies in charge of the timing and the scale of the generosity. The newer idea, universal basic capital, flips the sequence. Instead of taxing the winners after the fact, give every citizen an ownership share in the winning companies from the start. Senator Sanders has introduced the only bill in Congress on the subject. It proposes funding a sovereign wealth fund through a tax on the stock of AI companies. Supporters point to Alaska, which for four decades has written annual dividend checks to every resident from its oil-funded permanent fund.
Universal basic capital sounds like the same idea with better plumbing. In practice, it is a different creature entirely. Universal basic income was tested in dozens of cities and studied to death, and the evidence said it works. Universal basic capital has almost no empirical record at all. There are more than 100 major sovereign wealth funds spread across 67 countries and six American states, but none of them was built by taking equity out of privately held technology firms. The policy exists as a phrase in speeches and a paragraph in a bill that no other member of Congress has signed.
The appeal is obvious, and that is precisely the problem. The phrase lets a politician stand in front of an audience and say that artificial intelligence will not just enrich a few thousand engineers, but that the public will own the machine. It is a form of consolation that sounds like a plan. It has the emotional force of a promise and none of the legal weight of one.
The City Where the Future Arrives First
While the politicians talk, the bear is approaching, and nowhere is the approach easier to see than in San Francisco, the capital of the AI gold rush. From January through June, more than 140 homes in the city sold for at least one million dollars above their asking price. In the same period a year earlier, the number was eight. The rental market has the lowest vacancy rate of any major American city, at 2.2 %, and AI salaries are being plowed directly into housing, restaurants, and a lifestyle that has no equivalent elsewhere in the country.
The wealth is undeniably real. OpenAI and Anthropic, two companies headquartered in the city, are preparing what could be the two largest stock market debuts in American history. OpenAI is considering bringing its initial public offering to market. That is the moment when the next presidential campaign is just beginning to take shape, which makes the timing almost too convenient for the politicians.
All of this visible abundance does important political work. It makes the case that artificial intelligence creates value, and it makes that case loudly. A city where bidding wars are ordinary and vacancy is virtually nonexistent looks like a place where the future is working. The illusion is that this future has a waiting room, that the wealth accumulating in San Francisco will eventually spill out to the rest of the country like a rising tide. The evidence so far points to a waterfall.
This is where the public story starts to break away from the private one. The AI industry does not need to hide its gains, because the gains are the advertisement. Every headline about a trillion-dollar valuation, every story about a home sold a million dollars over asking, reinforces the idea that something enormous is being created. The question of who gets to keep it is deferred, and the deferral is the point.
What the Research Actually Shows
The research record draws a clean line between the two ideas. What we know about universal basic income comes from a decade of experiments with real money and real households. The Stockton pilot, run by the city’s former mayor Michael Tubbs, gave 125 residents 500 dollars a month for two years and then measured what happened. [4] The results, released in 2021, showed that recipients used the money to stabilize their housing, to buy better food, and to show up for work with fewer distractions. [4] Similar experiments in Finland, Canada, and Kenya pointed in the same direction. The conclusion was not that basic income solves poverty, but that it does not corrupt people, which was the charge it had faced since the Nixon administration first floated a family assistance plan in 1969 and then buried it.
What we assume about universal basic capital is built on much thinner ground. The Alaska Permanent Fund is the favorite example, and it is also a misleading one. Alaska taxes an oil extraction industry that has to pump from the ground; it does not take ownership of the companies doing the pumping. The fund works because fossil fuel profits were already flowing through a state tax system, not because the state holds shares in British Petroleum. Translating that model to artificial intelligence means asking a very different question: would a state or federal government seize equity in privately held companies, and at what point in their life cycle?
The professors behind the Sanders bill have thought about this more carefully than almost anyone else in the debate. Sarah Polcz and Jeremy Bearer-Friend argue that taking a one-time 50 percent tax on the stock of AI companies is better than trying to claw back profits later through corporate taxes. US News Their reasoning is grounded in the observation that technology companies have become exceptionally skilled at minimizing their tax liability. By the time the profit is reported, they argue, it has already been routed through subsidiaries, intellectual property arrangements, and other structures designed to make it disappear.
That argument has real force, but it also exposes the central tension of the whole project. If the government takes a stake in an AI company, it becomes a shareholder with an interest in the company’s success. Would a government that owns part of an AI firm regulate it as strictly as one with no stake? Polcz says the legislation forbids bailouts and would not soften regulation, but the question is structural, not legal. A government that profits from a company’s growth has a permanent incentive to let it grow, and no commission charter has ever fully solved that problem.
The Bear and the Goose
Teri Ollie, vice president of states strategy at the Economic Security Project, a think tank that has studied direct payments to low-income Americans, describes the political situation with a bluntness rare among policy professionals. “Lawmakers don’t act until the bear is in their face,” she says. “There’s a real fear of killing the goose that laid the golden egg. It’s dangerous to an elected official. They feel that the economy is like a bubble. If you touch it, it will burst.” US News
The goose metaphor is doing a lot of work in that statement, and it deserves attention. The golden egg is the AI economy. The goose is the industry that produces it. And the politicians are the farmers who have decided that the safest course of action is to describe, in great detail, how they plan to share the eggs, without ever touching the goose. This is not cynicism; it may be something closer to superstition.
Newsom has been the most active talker. In May, he told the Center for American Progress that “you don’t need charity, we need ownership” and called for an “ownership state.” [2] He signed an executive order directing California to study giving the state equity shares in AI companies, and demanded a roadmap from state agencies by August. US News The order was, in the clearest possible terms, a request for a study. It put no money in anyone’s pocket and committed the state to nothing. But it allowed the governor to say, at every subsequent appearance, that California was “looking at this notion of equity.” The roadmap, when it arrives, will be the first real test of whether the conversation can survive contact with a government document.
In July, Newsom told reporters that “a lot of people are now talking about this,” and listed Sanders and Trump as evidence. “It’s a very healthy conversation, a very healthy debate,” he said. A conversation is not a policy. A debate is not a dividend. The governor has not said how California would acquire equity in companies that are not currently offering it, at what valuation, or through what legal authority. The executive order did not answer these questions, because it did not ask them.
Trump has been even vaguer. In May, he suggested that Americans should become “a partner” in AI companies, without explaining the mechanism, the funding, or the legal basis for such an arrangement. US News The phrase has the quality of a slogan that has not yet met a policy it can survive. It is the campaign-trail version of the pricing algorithm: confident, opaque, and impossible to interrogate.
The gap between the rhetoric and the mechanism is not an oversight. It is a feature of the political moment. The 2028 presidential election will be the first to run entirely on the question of how to handle artificial intelligence, and every candidate wants to occupy the position of having addressed it without having been pinned down by it. In May, Nathan Gardels, a co-founder and senior adviser to the Berggruen Institute, wrote that “the 2028 campaign will be the first-ever AI election,” defined by how candidates propose to cope with the upheaval of widespread AI integration. US News The candidates will come to San Francisco, look at the boom, and offer the public a share of the machine. The offer costs nothing to make.
The Question Nobody Can Answer
Michael Tubbs knows what it means to turn an idea into a program. As mayor of Stockton, he launched one of the first guaranteed income pilots in the United States, and watched it grow into a national movement. But Tubbs is also clear that basic income was never going to be enough. “In this moment it is the beginning,” he says. “It’s not the end. It has to be married to this idea of ownership. Like how do you make sure that all of us have some ownership in the wealth that’s going to be generated, so it’s not just a couple people seeing their standard of living raised, but all of our standard living raises?” US News
The word “ownership” is doing something important in Tubbs’ sentence, something different from what Newsom and Trump are doing with it. Tubbs is describing a moral floor, a way of making sure the economy’s next chapter includes the people who were left out of the last one. Newsom is describing a policy brand. Trump is describing a feeling. None of this means the politicians are insincere. It means the distance between a conviction and a statute is the whole game, and the game is being played in public, with trillion-dollar IPOs as the backdrop.
History suggests a possible path. In the 2020 election, Sanders pushed the entire Democratic field leftward on health care and climate, and Joe Biden eventually folded parts of those ideas into his agenda. The same pattern could repeat with universal basic capital. A candidate who can turn the phrase into something more than a phrase — who can explain it simply and coherently, as the pilots once explained basic income — could make it the defining issue of the 2028 race. But that candidate does not exist yet, and the only bill on the table has zero co-sponsors.
The silence from Washington is itself a kind of answer. There are no serious negotiations happening over how to structure an AI sovereign wealth fund, no hearings on the mechanics of a 50 percent stock tax, no draft legislation coming out of the California Senate to match the governor’s rhetoric. The absence of activity is not an accident of scheduling. The subject is difficult, the interests arrayed against it are powerful, and the political reward for vague agreement far exceeds the reward for specific disagreement.
Meanwhile, the timeline of the technology and the timeline of the politics are moving at different speeds, and the gap between them is the real story. The IPO is coming in early 2027. The presidential campaign is already forming around it. The bear that Ollie talked about is not at the door. It is already in the house. And the politicians have decided, for now, that the most rational response is to keep describing the goose and the eggs, and to hope that the goose never notices how hungry they are.
What would a real plan even look like? The elements exist in fragments. A sovereign wealth fund with an independent commission, as Sanders proposes. A valuation mechanism for taking equity in companies that are not yet public. A legal theory for a 50 percent stock tax that would survive the inevitable court challenge. A distribution channel that does not collapse under the weight of the IRS bureaucracy. Each of these fragments is solvable on its own. The illusion begins when the fragments are presented as a whole.
The Gap Between the Boom and the Bill
The deepest problem with the universal basic capital conversation is not that the details are missing. Details can be drafted. The problem is that the conversation treats AI as a finished product with a known revenue stream, when it is actually a speculative boom with a contested future. The same AI companies that politicians want to tax in order to fill a sovereign wealth fund could just as easily go bust, or merge, or move their intellectual property to a jurisdiction with a more welcoming tax code. The history of technology is littered with trillion-dollar promises that ended in bankruptcy filings.
An AI bubble would not merely pop. It would take the sovereign wealth fund down with it, because the fund’s assets would be the bubble. Polcz and her co-author insist that the legislation forbids bailouts, and that is a meaningful safeguard. But forbidding a bailout does not prevent a crash. It only prevents the government from being legally obliged to clean one up.
There is also a quieter danger, which is that the promise of ownership becomes a substitute for the harder work of building a safety net. If every American believes they will one day hold a share of an AI company, the pressure to fund unemployment insurance, job retraining, and a health care system that does not depend on employment may quietly dissolve. The promise of future wealth is a wonderful anesthetic for present pain. The patient stops complaining, and the surgery never happens.
This is the quietest form of the trick. It is not that the politicians are lying, exactly. It is that the promise of ownership, repeated often enough and sincerely enough, begins to function as a policy in its own right. It calms the public, reassures the markets, and allows the AI industry to continue its work undisturbed. The promise is the policy. The policy is the promise. And the public, like the tenant facing the algorithm’s rent increase, has no chain of reasoning to argue with.
The opening image of this story was a landlord and an algorithm and a tenant with no recourse. The closing image is not much different, except that the numbers are larger. A candidate stands on a stage in 2027, in the shadow of the largest IPO in history, and says that Americans will own a piece of it all. The audience applauds. The candidate means it, or half means it, or knows that meaning it is not the same as doing it. The applause is real. The share is not.
Progress and consequence rarely share the same time horizon. The technology moves in straight lines and sudden leaps. The policy moves in election cycles and committee schedules. The rent rises this month. The IPO lands next year. The dividend, if it ever comes, arrives at a date that no one alive can name. Every part of the machine is real except the one part the politicians promised. That is where AI deceives us: not in what it does, but in what we let it make us believe we have been promised.
Sources
1. Anthropic
2. Center for American Progress
4. Stockton Economic Empowerment Demonstration
5. IRS