This is chapter 2 of a set of ideas I’m putting together that may (or may not) end up as a short book, pamphlet etc. I am continually trying to understand how to make the world a better place. This is one set of proposals. I’m happy to hear feedback! Please share. If you are able, start to engage in sharing your wealth with others. You can start with your family, but you can then branch out to others!
What the seed actually becomes
The last chapter ended on a question: what would it take to carry every newborn from a single seed all the way to a real retirement? Here’s the answer, and it’s smaller than you’d guess.
Eighteen thousand dollars, invested once at birth and then left completely alone, grows to about $1 million in today’s dollars over sixty years. One deposit. Nothing added after it. No discipline required, no dependence on whether the parents could spare a cent. Seven percent above inflation — roughly what U.S. stocks have delivered over the long run — compounded across six decades turns eighteen thousand dollars into a million in real purchasing power.
I use $18,000 as the do-the-whole-job number — the top of the range, the deposit that fully funds a millionaire outcome from a single seed.
About 3.6 million babies are born in this country each year. At $18,000 apiece, seeding every one of them runs about $65 billion a year. Set that against a federal budget of roughly $7 trillion and it’s under 1% of what the government already spends. The maximum version of this idea — every newborn on track to a real million — costs less than a penny on the federal dollar.
And we already run a miniature version. The Trump Accounts seed every citizen newborn with $1,000 in a low-cost index fund. That $1,000, left alone at 7% real, becomes about $58,000 in today’s dollars by retirement — which, remarkably, is more than the typical American saves in an entire working life. Not because $1,000 is a lot, but because it gets sixty uninterrupted years and most people never give themselves that. The last chapter’s tables showed the same thing from the other direction: the young are absurdly cheap to fund and the old break the bank, and the whole difference is compounding. Time does the heavy lifting, and the young have all of it.
So the mechanism is right; the size is wrong and fixable. The harder question isn’t how much — it’s why this at all. Why ownership, not just a better wage or a monthly check?
The floor most people stand on is a wall
To see why a share beats a wage, you have to be honest about what a wage actually is.
Take ordinary work — driving a bus, picking a warehouse order, running a register, answering a support line. This work has real value; a city without bus drivers is worse off, and so are the people who couldn’t get to their jobs. But the value isn’t scarce. Lots of people can do it. And a price is set by scarcity, not by usefulness. So why does the Swedish bus driver earn a Swedish wage instead of an Indian one, for the identical task? Because a border and a license keep the person in India from showing up and doing the job for half the pay. If that person could simply come and drive the bus, the wage would fall to meet them.
That border is a wall. It holds a price up by keeping the cheaper competitor out — and it’s been running the whole time, long before anyone mentioned AI. Half the wages in the developed world are propped, to some degree, by a wall that keeps a willing, lower-priced worker on the other side of a line.
There’s a cruel version of this point and a true one, and only the true one is mine. The cruel version is “most people add no value.” That’s false. The true version is narrower: the value is not scarce. The wage measures scarcity, not worth. The entire humane core of this project is the claim that a person’s worth is not their market price.
So the wage floor was always a propped one. Here’s why AI is different from every earlier wave of cheaper labor, when “someone will do it cheaper” is the oldest story in economics: for the first time, the cheaper competitor is on the inside of the wall. You can keep a foreign worker out with a visa. You cannot keep a model out with a visa. It needs no work permit, crosses no border, and lives in a data center down the road. The wall that held the wage up does nothing against it. And a propped floor doesn’t sag gently when the prop goes — it holds, and holds, and then fails all at once.
The only floor that doesn’t need a wall
Here’s the whole case for the shareholder society in a single contrast.
Every floor built on a wage needs a wall to hold it up — a border, a license, a union rule, a tariff, an occupational cartel. Walls are politically fragile; they’re one election or one trade deal from coming down. And against AI they’re not even fragile, they’re useless, because the competitor is already inside.
A floor built on a share needs no wall at all. A share’s value doesn’t depend on keeping any competitor out. It’s a claim on production however cheaply that production gets done — and it gets stronger as production gets cheaper. The same AI that destroys the wage increases the profit, and the shareholder owns the profit. Ownership is the one floor that rises as labor gets cheaper instead of collapsing with it. The thing that kills the wage feeds the owner.
You can already watch people reach for the wall anyway. Bernie Sanders has called for a moratorium on new AI data centers and a pause on advanced AI development, to protect workers from mass job loss — a wall thrown up against the machine itself. I understand the impulse, but it doesn’t work, and not only because the competitor is already inside. It’s that this particular wall can’t be built by one country. Slow AI down in America and you don’t stop it — you hand the lead to the countries that don’t, and they out-compete us with the very tools we refused to build. You can wall a cheaper worker out at the border. You cannot wall out a technology the rest of the world is racing to develop. Which is the deepest reason to prefer a share to a wall: a share lets ordinary people win from the machine’s advance, instead of staking their floor on a doomed attempt to hold it back. (Sanders, as it happens, has the ownership instinct too — he just builds it the wrong way, which I’ll come to.)
It’s the structural reason a stake beats every alternative on offer. A UBI is a wall made of tax policy — one budget fight from shrinking. A protected job is a wall made of licensing — one disruption from falling. A closed border is a wall made of law — one election from opening. A share is not a wall. It’s a piece of the thing itself. That’s the difference between handing someone a check the government has to keep choosing to write, and handing them an asset they own outright. If your labor isn’t needed, your shares are how you eat — and, once they carry a vote, how you’re heard.
And there’s a benefit here that runs deeper than the income — one I’ll come back to at the end of the book. A check keeps you a recipient, a spectator of an economy that merely happens to you; a share you own and vote makes you a participant in it. People who own begin to think like owners, and that change in how a person stands toward the whole system is a large part of what the system is for.
“But that’s unearned”
The standard objection arrives right here: people should earn their stake through work. You don’t just hand out shares. That’s unearned.
The answer falls straight out of the wall. The well-paid worker’s premium was already unearned — a wall-protected rent, not a reward for scarce value. The Swedish wage paid over the Indian wage for the identical task is a transfer the state is already choosing to make, through immigration and licensing law, in favor of whoever was born inside the wall. Society already redistributes to people whose labor isn’t globally scarce. It just launders the redistribution through a wage and a border so it looks like earnings.
So “unearned” isn’t a line that separates a shareholder stake from the status quo. Unearned is the status quo. Universal ownership doesn’t inject redistribution into a pure meritocracy — it takes a redistribution we’re already running behind a wall and does it in the open, as a share you own and vote, portable and not contingent on holding a protected job. And unlike the wall, the share doesn’t vanish the moment the politics shift or the model shows up. It’s the more honest version of a transfer we’re already making, and the more durable one. I’d rather hand someone a share they own than a wage I have to keep a wall standing to protect.
Keep the machine ownable
Everything so far assumes one thing: that the productive economy is ownable — public, listed, reachable by an index fund and a seed account. That’s the bolt the whole design rests on, and it’s starting to shear, three ways at once, each one pulling the best part of the economy out of ordinary reach.
It’s going private. The number of public U.S. companies is about half what it was in the mid-1990s, and companies now do their steepest growth while private. The frontier AI labs — the sharpest edge of the whole economy — are private. By the time a modern company reaches your index fund, the fast part is over; you’re buying the plateau. And the richest compounding is walled off deliberately: to buy into most private companies you need about $200,000 of income or $1 million in net worth. Read that plainly — you’re allowed to own the fast-growing part of America only after you’re already rich. That’s the same captured-rule mechanism as the zoning that keeps housing expensive, just naked: a wealth test on entry to the best returns.
It’s getting lean, and at the limit, solo. The endpoint everyone flinches from is the founder who builds something at the scale of a major lab using AI agents — raising no outside capital, issuing no shares, hiring no one. Walk that company through both of the ways value normally reaches people and watch them run dry at once: no employees means no wages flowing into any community, and no shares ever sold means nothing on a cap table for a citizen fund or a seed account to buy. It’s the one structure where both levers fail together. The small version already exists — the solo operator with an automation stack. The trillion-dollar version is maybe fifteen or twenty years out. Worth building the framework to survive it before we need it.
Concentrated ownership doesn’t only decide who gets to buy — it decides what gets built. If demand tracks ownership and ownership is top-heavy, the automated economy optimizes for what the owners want: the yacht, not the cheap house, because the yacht is where the spendable money is. Broaden the ownership and you broaden the demand signal — you keep the machine’s attention on the mass-market goods that reach everyone. Universal ownership isn’t only how the poor get an income. It’s how the economy stays interested in serving the poor at all.
So the fix has to keep the machine ownable without becoming the confiscation I ruled out. Here’s the shape. Any company that grows past some very large scale — think a trillion dollars in value, and maybe far more — has to issue a slice of itself into citizens’ own accounts, carrying a real pass-through vote. The size of that slice could scale with the size of the company. I’m deliberately not naming the threshold or the percentage; those are dials for public debate. What I’m putting forward is the framework, not the settings. It’s a condition of operating at civilization scale, published in advance, in the same category as a listing requirement or a tax bracket. Even a one-person company that never sold a share, once it crosses the line, mints equity into the citizen vehicle. Whether the founder is paid for the slice or simply diluted to make room is exactly the kind of thing public debate should settle — I’d put both on the table.
Why isn’t that the seizure I swore off? Three reasons. It’s prospective and known — I’m setting the price of a new tier of scale before anyone reaches it, so every founder builds knowing the deal; that’s a rule of the game, not a taking. It’s built on the public’s own custom — a company only reaches this scale by serving hundreds of millions of people, and the valuation just is that ongoing custom capitalized into a number. And it’s rent on the commons — nobody builds a colossus on bare ground; they build it on courts that enforce their contracts, on publicly funded research their models descend from, on infrastructure and education and a stability that guards their property while they sleep. The citizen stake is the public collecting on infrastructure it already provided.
It’s worth measuring all this against the most prominent version of the same instinct — because someone got there first, and where we agree matters as much as where we don’t. Bernie Sanders’s 2026 American AI Sovereign Wealth Fund Act would take a one-time 50% stock tax on the largest AI companies, pool the shares in a federal fund, and pay every American a yearly dividend. On the goal, he and I want the same thing: the public owning a real piece of the productive machine, and I take it as good news that the idea has reached the floor of the Senate.
And it isn’t only Sanders, or only the left. The Trump administration has already taken direct equity stakes in a string of companies — about 10% of Intel, 15% of the rare-earth miner MP Materials, pieces of several others — a government portfolio one of its own economic advisers called “a down payment on a sovereign wealth fund.” So the machine passing into public hands is neither hypothetical nor partisan anymore; it’s arriving from both directions at once — seize-it-into-a-fund on the left, take-a-strategic-stake on the right. Which sharpens the whole thing to a single question: it was never whether the public comes to own a piece of the economy, but who holds that piece and who votes it — a Washington fund, or the citizen. And that is where I part from both of them, because the design is everything. (What should be done with the stakes Washington is already piling up, I’ll pick back up when I reach the people the seed can’t help.)
Start with the amount. He takes half in a single move, where I’d start with a sliver at a far higher bar and let it grow slowly, if at all — because 50% at once isn’t broadening ownership, it’s nationalizing an industry. Then the scope, where the difference flips: he socializes only the biggest AI firms, while the shareholder society is about the whole productive economy — every company, reached through the broad index the seed accounts already buy. His is deep but narrow, half of a handful of companies; mine is shallow but broad, a small piece of everything.
Third — and this is the real fault line — who ends up holding the power. Sanders’s bill routes the public’s stake through an Independent Commission for Democratic AI: seven presidential appointees who manage the fund “in the public interest” and vote its shares to block what they judge harmful and push what they judge helpful. Read that plainly. It’s a new arm of government that decides, on your behalf, what the public interest is. That’s more government, not less — the state owning the machine and a handful of appointees steering it. My whole design is the opposite move: don’t build a commission to hold the shares and cast the votes for three hundred million people — give the three hundred million people the shares, and the votes, directly. A stake in your own account. A ballot you cast yourself. He’d call his commission “democratic”; I’d point out that seven appointees voting the nation’s shares is representation at best, and that the more genuinely democratic thing is for every citizen to vote the piece they actually own. Same word. Opposite machine.
There’s an honesty I owe on my own side, though. His bill simply seizes — “we’re not buying it,” as he put it, “we’re getting it” — half of companies already built, which is the retroactive taking I ruled out and the reason it’s already drawing constitutional fire. My going-forward pieces avoid that by being prospective. But I haven’t solved the mirror problem for the companies that already exist and the adults who already do: my instinct runs the opposite way from his — buy the shares at fair value and hand them to people rather than seize them — but buying costs money, and that lands us right back on the hardest question in this book, the one I keep promising to face. I’d rather name that gap than paper over it.
And that thread — start small, keep it individual, buy rather than seize — is also the answer to the worry that any program like this only ratchets upward over time. Of course it does; the income tax began as a levy on the top 1% and now reaches nearly everyone. But the question was never whether the stake grows, it’s what grows. If a government fund’s share of the economy ratchets up, that’s the state accumulating control. If individual citizens’ ownership ratchets up, that’s ordinary people accumulating power over their own lives. Same direction of travel. Opposite destination.
The vote
I keep saying “a share with a vote,” and I mean it as load-bearing, not decoration. Ownership without governance is just a dividend check, and in an economy increasingly run by machines, whoever votes the shares runs everything. So the vote has to reach the actual citizen.
The mechanics are grade-school arithmetic. If you hold $100,000 in an index fund and that fund keeps 8% of its money in Nvidia, you own $8,000 of Nvidia — so you should vote $8,000 worth of Nvidia shares. Your ownership looks through the fund to the companies underneath, and your vote looks through right along with it, in exact proportion to what you hold. Brokerages already track ownership down to the share; the plumbing exists. Today that vote gets swallowed by a handful of giant asset managers who cast the proxies on everyone’s behalf — a real concentration of power, and the exact thing pass-through voting dissolves. The full case for why the vote matters this much — what it does to a person, and to a democracy — is its own chapter later. For now, hold the principle: your money’s already there; the vote should follow it.
What it costs, and who pays
Pull the money back into view. Seeding every newborn at $18,000 runs about $65 billion a year — under 1% of the budget. But the budget isn’t even the right yardstick. The right comparison is the retirement promise we’ve already made and haven’t funded: Social Security’s unfunded obligation runs to roughly $29 trillion over the next 75 years, its trust fund is projected to run dry around 2033, and benefits get automatically cut by about a quarter when it does. That’s an unfunded, pay-as-you-go promise — today’s taxes paying today’s benefits, the shortfall shoved onto tomorrow’s workers. A seed account is the opposite kind of obligation: pre-funded and compounding, a real asset that grows on its own and eventually needs nobody to tax. It’s the same move private employers already made a generation ago — from the defined-benefit pension, an unfunded promise to pay you later, to the defined-contribution 401(k), a real account you own that compounds on its own. We’d be making that shift for the retirement floor itself: from a defined benefit the country never funded to a defined contribution that funds itself. We’re already standing in a $29 trillion hole; $65 billion a year to build that self-funding stake beside it is a rounding error by comparison.
Which means you’re already asking the obvious question: if it’s so cheap, why not just tax the rich to cover it — or to fund everyone, not only the newborns? Hold that thought, because it’s the whole of the next chapter. The short version is that it doesn’t work, and it’s worth taking slowly.
And I won’t pretend the seed does everything. It rescues the young precisely because it buys them sixty years of compounding — which is exactly why it does almost nothing for the person who’s fifty-five right now, with ten years left and no time for a small stake to grow into a large one. Funding today’s middle-aged is the hard transition problem no seed can solve, and I’ll face it head-on rather than wave it away, later in the book.
But the destination is clean, and it’s cheap for everyone we can still start the clock for. Let the machine do everything. Let one person run a company the size of a nation’s economy if the technology allows it. The fight was never man against machine. It’s over a single question: who owns the machine, and who votes it? Make sure that when it finally needs almost no one to run it, almost everyone still owns a piece — and still gets a say in what it builds.
