AI Wealth Promise Ignores Workers' Real Economic Struggles

“He says workers can build wealth in the AI era.” Fine. But then he tells us how... with broad strokes and no roadmap. Vivek Ramaswamy’s Wall Street Journal opinion is bullish on worker upside as automation advances. I agree with the optimism in spirit. I do not agree that optimism is enough.

Let me start with where he’s right, because the impulse isn’t crazy. AI will create new forms of value. Frankly, the math doesn't lie about creative tools that expand what one person can produce in an hour. When output scales, there is more surplus to divide, and workers absolutely can capture gains through ownership, entrepreneurship, piece-rate premiums for augmented labor, or by repackaging human judgment as a service you can actually bill for. From my Goldman days I watched similar patterns in financial markets: when new systems made certain desks dramatically more productive, the people who did well were the ones who combined some ownership stake with skills the tech couldn’t fully automate.

That’s the part the column gets: technological change doesn’t have to be a one-way extraction that only fattens corporate margins. There is real upside for workers who sit at the intersection of scarce skills and some slice of ownership.

But then the hard questions arrive and the argument starts hiding behind abstractions.

Saying “workers can build wealth” is not a strategy; it's a slogan. Ramaswamy sketches a cultural shift toward grit and agency, hints that workers should become owners and entrepreneurs, and then more or less stops. No details on financing vehicles that would let workers buy into upside. No specifics on governance structures that would give employees any say in how AI gains are allocated. No legal reforms that would change who gets paid when productivity jumps. Let's be real — without naming concrete levers, you’re not giving workers a plan, you’re giving them a pep talk.

Here’s the crux he glances off: wealth creation in an AI-heavy economy sits on three legs — who owns the productive assets, who can actually monetize the new productivity, and what social architecture keeps incomes from collapsing while tasks and roles are redefined. He talks about worker ownership in broad terms, largely outsources monetization to “entrepreneurship,” and barely touches the income shock that hits people whose jobs get restructured before new ones appear. Workers displaced by AI do not teleport into higher-value roles. Frictions like access to capital, geography, and credential gatekeeping decide whether someone rides the wave or gets knocked under it.

We also have to talk about power, not just possibility. Corporations increasingly bundle algorithms with platforms, data, and distribution. When that bundle hardens, economic rents tend to concentrate. Ramaswamy’s argument more or less assumes either enlightened corporate behavior or frictionless capital markets that naturally steer gains into the hands of worker-owners. Neither lines up with how large incumbents usually behave. Look at how big tech platforms handle creators or gig workers: they centralize the value chain and then fine-tune the terms to serve shareholders. That isn’t an indictment of capitalism; it’s just how incentives play out when nobody rewrites the rules.

History isn’t exactly on the “don’t worry, it’ll trickle down” side of this. Think back to earlier automation waves or even the first big outsourcing push in manufacturing. Productivity rose. Aggregate wealth grew. Yet entire regions ended up hollowed out because access to ownership and retraining was tightly constrained and badly timed. The long-run charts looked fine; the local job markets did not. Treating AI as somehow immune to that pattern is faith, not analysis.

If you actually want workers to capture wealth, you have to change incentives in ways you can write into a contract or a statute. Not vibes — structures. Portable benefits that follow workers across gigs so that taking entrepreneurial risk doesn’t mean abandoning healthcare or retirement. Tax and capital rules that make employee ownership more than a branding exercise — think real paths to equity, not tiny grants that never vest or get diluted beyond meaning. Training linked to credentials employers truly pay for, instead of certificates that prove only that you sat through a webinar.

And then there’s competition policy, which he mostly sidesteps. When a handful of platforms control distribution, worker bargaining power is a décor item. Weak competition narrows the space for alternative business models where profit-sharing, co-ops, or meaningful employee stock ownership actually make economic sense. Breaking or regulating entrenched gatekeepers is ugly work — antitrust, data access rules, maybe targeted tax changes — but without that, “worker ownership” risks becoming a niche perk inside a few enlightened firms rather than a broad-based channel for wealth.

One historical analogy here is worth spelling out. When railroads and later telecom networks consolidated, their owners initially captured an outsized share of the gains from connectivity. Only after regulators imposed access rules and new organizational forms (like public utilities and common carriers) did the broader economy share the upside. We remember the growth; we forget how much governance it took to stop the network owners from keeping the whole pie. AI platforms are on track to be the new networks.

Of course, some will argue that markets will self-correct if firms hoard all the AI upside. New entrants will undercut them, workers with initiative will sprint up the skills ladder, capital will flow into worker-led startups. I’ve heard that speech in more than one investment committee. Markets do adjust, but not on a timetable that feels fair to the people caught in the middle. Those adjustments are rarely smooth, and they absolutely do not guarantee that displaced workers end up with a stake in the next wave rather than a severance package and a subscription to a coding boot camp.

Ramaswamy’s piece matters as a counterweight to techno-doom; tone does shape what voters and executives think is politically or socially acceptable. It just stops several steps before the questions that make people uncomfortable: who gives up margin, who shares control, and who pays to bridge workers from one economic role to the next.

My bet: the real test of his thesis won’t be whether AI makes workers more productive, but whether any major employer is willing to hard-code that upside into ownership and governance instead of leaving it in the realm of motivational prose.