Here's what will replace the paycheck.
What UBI, Universal Basic Services, and citizen-equity warrants miss.
Where previous articles answered “when will AI cause changes to the workforce?”, this one answers “what are we going to do about it?” There are tons of ideas out there—UBI, universal basic services, citizen-equity warrants—but ideas require implementation. One approach has a higher likelihood than others to being politically feasible and addressing the real problem (it isn’t jobs). This article is about what’s most likely to work.
In 1914, Henry Ford did something his peers thought was insane. He doubled the wages of his factory workers to five dollars a day. The press treated it as a stunt. His competitors treated it as a betrayal. His investors sued him.
Ford’s defense was almost boring. He said his workers needed to be able to buy his cars. If the people making the Model T couldn’t afford the Model T, the whole thing came apart. The wage wasn’t charity. It wasn’t even loyalty. It was circulation. The money had to come back, or the system stopped.
For roughly a century, that quiet idea underpinned the entire industrial economy. The wage paid the worker enough to be a customer. The customer’s spending paid the company. The company paid the wage. The loop closed.
That loop is opening again.
Circulation is breaking
Nine months ago, in The Last Normal Year, I argued that the leading indicators of AI-driven displacement were screaming, and that we had roughly eighteen months before the lagging data confirmed it. We are now well past the point where that’s a hypothesis. The composite signal score on the dashboard sits at 4.25 out of 5. Hyperscaler AI capex went from $256 billion in 2024 to $443 billion in 2025 to a projected $602 billion in 2026. Indeed’s job postings index has fallen back to 2017 levels, erasing a decade of growth in eighteen months. Workers aged 22 to 25 in high-AI-exposure roles have seen a 13 percent employment decline since late 2022, while their colleagues over 30 in the same fields have grown by six to twelve percent. The labor market isn’t shrinking. It’s bifurcating, and the bottom is falling out of the bottom.
The natural response to this is to ask what to do about it. I have been avoiding that question on purpose for nine months, because I wanted to track the signal long enough to be honest about the answer. We’re at the point where I can be.
You’ve heard the answer everyone in the conversation wants it to be. It’s the one we have to walk past to get to one that works.
Universal Basic Income. A monthly check, no strings. Andrew Yang campaigned on it. Sam Altman keeps writing about it. Elon Musk thinks it’s the obvious answer. Pilot programs in Stockton, Finland, and Kenya keep producing reasonable-but-not-revolutionary results.
The slightly fancier version is Universal Basic Services. Free healthcare, housing, transit, childcare, education. The dignity floor.
The newer version, with more political momentum than either of the first two, is some form of citizen-equity. Yanis Varoufakis has been pushing a Universal Basic Dividend for a decade, funded by mandatory equity contributions from public companies. Matt Bruenig has been quietly building out what he now calls the American Solidarity Fund, which would issue a non-transferable share to every American and pay an annual dividend from the investment income. New York State Assembly member Alex Bores introduced an AI Dividend bill this April, funded by an AI consumption tax and equity warrants on major AI companies. Elon Musk has been talking about “universal high income” funded by AI productivity. The Trump administration signed an executive order in February 2025 directing the Treasury to begin establishing a federal sovereign wealth fund.
I want you to notice something about that list. Varoufakis is a self-described libertarian Marxist. Bruenig runs a left think tank. Bores is a Democratic state legislator. Musk is… Musk. The Trump administration is the Trump administration. These people don’t agree on much. They are arriving at the same general mechanism from completely different starting points.
That’s interesting. It’s also the most over-interpreted finding in the field.
The convergence trap
The temptation, when you see five ideologically incompatible actors propose roughly the same architecture, is to treat the convergence as a signal that the architecture is correct. That is a trap. Convergence is evidence of political viability, not structural correctness. It tells you what is likely to survive a contested process. It does not tell you whether the thing that survives actually solves the problem.
We need to look at the problem before we look at the proposals.
A paper that came out earlier this year, by Hemenway Falk and Tsoukalas at Stanford, reframed AI displacement in a way that I think is going to define the policy conversation for the next decade. Their argument is that AI substitution isn’t primarily a redistribution problem. It’s a market failure. When a firm automates a task, the firm captures all of the productivity gain. The costs are borne by the rest of the economy. Fewer wages circulate. Aggregate demand weakens. Downstream firms feel it before the automating firm does. The private return to automation is higher than the social return. Companies, behaving rationally, will over-automate. Their estimate is that we’ll automate at roughly twice the rate that’s collectively efficient.
That’s a wedge. A wedge between what’s good for the firm and what’s good for the system. Economists have a name for that wedge and a tool for closing it: the Pigouvian tax. You tax the activity at the level of the externality. The price corrects. The behavior changes.
Now look back at the popular proposals through that lens.
UBI doesn’t change the wedge. It cushions the people the wedge displaces.
UBS doesn’t change the wedge. It builds a dignity floor underneath the people the wedge displaces.
Citizen-equity dividends, in any of the Varoufakis, Bruenig, or Bores variants, don’t change the wedge either. They give the displaced a slice of the upside the firms are capturing.
None of them touches the per-task incentive that’s driving the over-automation in the first place. They are all redistribution. The market failure goes on producing the imbalance, and the redistribution mechanism races to keep up. That’s a treadmill, not a fix.
The only architecture in the popular conversation that operates on the wedge itself is a Pigouvian automation tax. And almost no one is talking about it, because it’s politically unsexy and the entities that would have to design it are also the entities most exposed to it.
This is the part of the analysis that took me six months to get comfortable saying out loud. None of the popular proposals, on its own, addresses the structural problem. The combination of them might.
What actually has a chance of getting built
Designing the right answer is one task. Getting the right answer designed, legislated, funded, and operating before the displacement window closes is a different task, and the harder one. The clock on installation is shorter than the clock on payout. By the time AI companies hit displacement-scale valuations, they will be the most politically powerful entities in the economy. Architectures that depend on legislating against them after that point are structurally weaker than architectures installed before.
So we need a candidate that can actually be standing up by 2028.
UBI is well-designed but funding it at scale runs into the same fiscal-pressure spiral that has eaten every previous attempt. UBS is durable but slow to install. Building free healthcare and housing infrastructure is a multi-decade project. Citizen-equity warrants on AI companies are fast under crisis conditions but fragile if the AI companies themselves are caught in a financial reset. Pure public ownership of AI capital is theoretically the cleanest answer and legislatively the most distant.
The architecture that comes closest to surviving all four constraints, the ones that matter (designed, legislated, funded, operating), is the citizen-equity sovereign wealth fund.
It already has working precedent. Alaska’s Permanent Fund Dividend has been paying out to every Alaska resident since 1982. Forty-four years. It survived four oil crashes and seven governors from both parties. It pays roughly $1,500 to $3,000 per person per year. It is one of the most popular government programs in the reddest state in the country. The mechanism isn’t hypothetical. It runs.
The federal version has more political infrastructure than people realize. Bruenig’s American Solidarity Fund has the operational design. Bores’s AI Dividend bill has a legislative vehicle in New York. The Trump executive order has the Treasury already drafting structure. There are pieces in place across three ideological camps that, if connected, look more like a coalition than people are admitting.
It’s not a perfect architecture. The objection that bothers me most is that an SWF, like UBI, is still redistribution. It doesn’t touch the wedge. The companies still over-automate. The fund just gives the displaced a slice of what the companies capture.
That’s why the SWF alone isn’t the answer. It’s the floor. The Pigouvian piece has to be built on top of it. The fund pays out the dividend. The tax on automation rate corrects the per-task incentive and helps fund the dividend. UBI or warrants act as fast cushioning during installation. Public ownership operates as framing pressure that keeps the whole conversation honest.
That’s the combination. SWF as the politically viable installation site. Pigouvian elements layered on. UBI as the cushion. Public ownership as the pressure.
None of the popular answers, on its own, works. The combination does.
What this looks like for a real person
I want to make this concrete, because the architectures stay abstract until you can see a person inside them.
Imagine someone we’ll call Jenna. She’s 34, a marketing coordinator at a mid-sized B2B software company. She lives in Denver. She has a partner who teaches middle school and a four-year-old. Her job, on paper, is the kind of role that the Stanford analysis flags: cognitive, AI-exposed, ages 22 to 35 most affected. Her income covers about 60 percent of household costs. The mortgage assumes both incomes.
Under the system we have now, if her job goes in 2027, the family is one severance check away from a hard problem. Unemployment runs out. The house is the question. She moves down-skill, takes a job that pays 30 percent less, and the family absorbs the loss for the next decade. That’s the path the data is currently pointing at for a meaningful slice of the workforce.
Under the system I’m describing, the picture is different. The federal sovereign wealth fund is paying her a citizen dividend that runs maybe $4,000 to $8,000 a year, scaled up over time as the fund’s assets grow. The Pigouvian tax on automation is generating a portion of that revenue, and is also slowing the rate at which her employer over-automates, because the per-task math now includes the social cost. UBI-style cushioning kicks in if her income drops below a threshold, buying her time to retrain. The dividend isn’t replacing her wage. It’s making sure that when her wage compresses, the floor she lands on is real.
That’s not utopia. She still has to work. The transition is still hard. But it’s the difference between a family that’s one bad quarter from foreclosure and a family that has six months of breathing room and a household income floor that doesn’t depend on a single job. That’s the version of the future I think is actually achievable. It’s also, I’d argue, the version that capitalism survives. The alternative is the long, ugly contraction.
The single signal to watch
Six months ago, my advice to readers was to track leading indicators of displacement. Today I want to give you one to track on the other side of the question. A leading indicator of whether a working answer gets installed in time.
It’s this. Whether the Trump administration’s February 2025 executive order on a federal sovereign wealth fund acquires a dedicated seeding mechanism through legislation by mid-2027.
If it does, the SWF lean strengthens substantially, and the rest of the architecture has something to attach to. If eighteen months pass without it, the political-viability story has to be revised down. Hard.
A few weeks ago I would have told you I was watching half a dozen signals. I’m down to one. It’s the one where the political infrastructure is closest to existing and the failure mode is most visible. The other proposals are interesting. This is the one where the loading bar is actually moving.
Why this is, secretly, an optimistic piece
I started by saying the loop is opening. I want to close by saying that’s not the same as the loop ending.
What Henry Ford figured out in 1914 is still true. An economy in which the people producing the goods can’t afford the goods is structurally fragile. He was talking about cars, but the principle scales. An economy in which the people whose labor used to circulate the money can no longer circulate the money is one wave of bad news away from a long, ugly contraction.
But that contraction’s not inevitable. The circulation isn’t broken because the underlying productive capacity is breaking. It’s the opposite. We are about to be richer, in real terms, than any society has ever been. The output is going up. What’s broken is the connection between the output and the people. That’s a design problem, not a scarcity problem. Design problems get solved. The question is whether we solve it before the disconnection becomes the new equilibrium and the political conditions for solving it disappear.
I think we have a window of roughly two to three years to put a working architecture in place. I think the SWF-plus-Pigouvian combination is the candidate most likely to get there, because it’s the only one with both a structural answer to the wedge and a politically viable installation path. I think we’ll know whether the answer is being installed by mid-2027.
If we get this right, the post-wage economy isn’t a downgrade. It’s a system in which a country’s productive capacity flows back to its citizens through ownership rather than through a job they happen to be lucky enough to hold. That’s not a worse story than the W-2 economy. In some ways it’s a better one. It’s certainly more honest about where the value is actually coming from.
The full brief on this analysis (comparing six architectures across seven structural criteria and three stress scenarios, with the math behind every claim above) is at josephlogan.com, free under CC BY 4.0; share it with anyone working on this question. The monthly Hari displacement updates continue. And if you’re working in or near a state legislature, a federal agency, or a policy shop where any of this is being discussed, I’d genuinely like to hear what you’re seeing.
Henry Ford figured this out a hundred and twelve years ago, and he wasn’t even trying. The instruments are still showing the loop opening. We have something close to two years to close it on purpose this time.
This article is based on research available here.