Ramaswamy's AI wealth plan overpromises, underdelivers for workers

Vivek Ramaswamy argues workers can build wealth in the AI era. I’ll be honest — that’s not exactly a radical claim. The harder question is baked into the headline but never really unpacked: which workers, using what tools, on what timeline?

There’s a piece of his thesis that deserves real credit: insisting that AI be discussed in the language of worker upside, not just disruption. That’s a healthier starting point than the usual “robots will take your job” doom loop. But optimism is cheap if it skips the messy plumbing between technological possibility and actual paychecks.

Here’s the thing: Ramaswamy’s argument treats “workers” as one big bucket. Labor markets don’t. Some people walk into this AI cycle with equity, savings, mobility, and useful social networks; others are juggling two jobs and a broken transmission. The column gestures at “opportunity” without much interest in the frictions that separate a software engineer in San Francisco from a clerk in a shrinking manufacturing town. Access to capital, quality retraining, and the ability to move for a job are not vibes; they’re constraints. Ignore them and you turn a plausible promise into a pep talk.

Wealth in the U.S. has historically come less from wages than from ownership — of businesses, real estate, retirement accounts. AI doesn’t change that math. If AI compresses certain job ladders, employers will capture the productivity gains first, and workers will see the upside only if institutions — unions, profit-sharing plans, employee stock ownership programs — translate that productivity into actual stakes. That’s not a morality play; it’s how comp plans work. You can hope markets spontaneously generate new forms of ownership for workers, but markets also like to crown a handful of outsized winners. Ask publishers about who wound up with most of the internet ad revenue.

The policy and institutional scaffolding that turns automation into broad-based wealth is where the real story is — and where the column is thinnest.

Take geography. The column talks about workers as if they’re playing on one big national field, but AI adoption tends to clump. Cities with venture ecosystems, dense supplier networks, and flexible labor markets pull in the new firms and the equity culture. Places that don’t have those features get hollowed out, slowly at first and then all at once. We’ve run this experiment with deindustrialized Midwestern towns and coastal tech hubs. AI is more likely to accelerate that sorting than reverse it unless someone is intentionally paying for the transition.

And “paying” is the quiet part. Who fronts the bill for retraining, relocation, and the time it takes to climb a new skill ladder — public budgets, employers, philanthropy, or the workers themselves? Treating wealth-building as a real project means naming a payer.

Workers also need portable ways to capture ownership. That doesn’t mean every dry cleaner suddenly hands out stock options. It does mean building an ecosystem where non-tech workers can hold meaningful and liquid claims on productive assets. That might look like retooled retirement vehicles that funnel more savings into diversified equity, aggressive encouragement of employee ownership, and profit-sharing models tied explicitly to productivity gains from AI tools inside a firm.

These are design problems, not slogans. They’re solvable, but only if they graduate from applause lines to line items.

The column also skims past coordination costs like they’re a rounding error. Retraining programs already exist; so do community colleges and workforce grants. The issue is quality, speed, and fit. Matching curricula to AI-adjacent skills that are actually in demand is a logistics problem and a political one. Private reskilling efforts often cherry-pick the most “promising” employees. Public programs are reactionary and underfunded. Bridging that gap means unglamorous work: modern apprenticeships, portable credentials, and incentives that reward employers for hiring people who just re-trained instead of people who already look perfect on paper.

None of that fits neatly in a tidy op-ed, but pretending the hard parts are inevitable market magic is how you end up with another generation of disillusioned workers.

Sure, but you could push back: technological shifts in the past created enough winners that opportunity eventually filtered out — so why wouldn’t AI do the same? The problem is that this “eventually” model leans on assumptions about time, mobility, and personal slack that don’t match the lived reality of a lot of households. Trickle-down is less an economic law than a political bet that people can wait out the transition and move freely to where the jobs are. That bet looks worse when whole towns lose their tax base and workers are tied down by caregiving, health issues, or housing that can’t be sold.

History doesn’t say “technology always works out”; it says “technology works out for those who shape the rules early.” Railroads, electrification, the internet — the distribution of benefits followed ownership structures and public choices, not just innovation itself. When Intel or Microsoft minted thousands of middle-class millionaires via stock options, it wasn’t an accident of physics; it was a compensation philosophy, backed by tax and securities rules that made those options worth something.

One sci-fi sidebar, since it’s the law in tech commentary: William Gibson pictured cyberspace as dazzling but unevenly accessible, a neon frontier where some users had godlike reach and others just scavenged in the margins. That’s not a bad sketch of how AI could look if we fixate on the tools and ignore who gets the keys, the maps, and the deed.

Ramaswamy is right to argue that workers can build wealth in the AI era; they absolutely can. But whether that line describes a broad middle or a narrow slice will depend less on model weights and more on ownership rules, institutional plumbing, and how fast we’re willing to turn “worker upside” from a narrative into a negotiable term sheet.