Is AI the workers' wealth machine or mirage?

If Vivek Ramaswamy is right that workers can build wealth in the AI era, somebody should tell the income statement. His Wall Street Journal piece—“How Workers Can Build Wealth in the AI Era”—has a clean narrative arc, a friendly optimism, and a headline any campaign consultant would frame. He’s not wrong to say there’s upside here. The problem is simpler and more brutal: upside without structure usually flows to capital, not labor.

Right, let’s start with where he has a point. Technological shifts do create new roles, new firms, and new fortunes. That’s not ideology; it’s how capitalism recycles itself. There is real potential for workers to own a slice of the AI upside, whether through equity, profit-sharing, or new forms of participation. But treating that potential as destiny skips the part where power, law, and corporate charters decide who gets what. Opportunity is not an allocation mechanism.

Ownership without teeth is where the fantasy really starts to wobble. “Let workers own more” sounds like a strategy; in practice it’s usually branding. Companies hand out stock options, restricted stock units, or a sliver of an employee stock purchase plan and call it empowerment. Shareholders clap, boards congratulate themselves, and workers are left timing vesting schedules against layoffs and blackout periods. My Goldman days taught me to skip the press release and read the plan documents—buried in the fine print is where you find the real hierarchy.

If the goal is durable wealth for workers, the mechanism matters more than the slogan. Employee ownership that comes with voting rights, board representation, and protections against one-sided dilution is entirely different from equity-as-bonus theater. Without influence over capital allocation—buybacks vs. wages, dividends vs. reinvestment—“ownership” is just a variable component of pay with extra volatility. Equity with no exit path, no liquidity, and no say in governance isn’t a wealth engine; it’s a mirage that happens to vest quarterly.

You can see the difference in how some firms structure things. There are companies that publicize employee stock plans aggressively while keeping all real decision-making locked in dual-class shares or concentrated founder control. Workers hold slivers of upside but no real power over strategy, layoffs, or where automation savings go. It looks inclusive on the cap table; it functions exactly like every other top-down capital structure. The label changed. The distribution didn’t.

Ramaswamy nods toward reskilling and opportunity in the AI era. Again, not wrong. But reskilling isn’t a magic wand; it’s a funnel with leakage at every stage. Training programs vary in quality, completion rates are uneven, and the labor market doesn’t reward credentials equally. Having watched countless “digital transformation” decks in boardrooms, I can tell you the story is always frictionless: workers train, slide into higher-productivity roles, and everyone wins. The actual path is messier and less charitable.

So here’s the more likely sequence inside many firms: automate repeatable tasks, keep a smaller cohort of “strategic” roles, outsource whatever doesn’t fit, and pass the cost savings to investors. Workers then face a constrained menu—fight for a limited number of redefined internal roles, or pay out of pocket to retrain into an adjacent field and hope someone hires them at better wages. If training is voluntary, who eats the opportunity cost of time off? If it’s mandatory, what happens to those who can’t keep up with constant skill churn? These are operational questions, not branding exercises.

There’s also the geographic and sector mismatch nobody likes to talk about in op-eds. Even if AI creates net new jobs, they don’t all appear where displaced workers live or in industries where their experience translates. It’s easy to say “just adapt” when you’re writing from a coastal office; it’s harder when your local employer base is consolidating and the new roles require both relocation and a financial buffer you don’t have. That fraying edge between theory and cash flow is where optimism gets expensive.

Which brings us to policy and incentives, the part of the conversation Ramaswamy mostly waves at. If you’re serious about spreading AI gains, private-market solutions alone won’t do the heavy lifting. Corporate incentives are misaligned by design: boards are bound to shareholders; executives are paid in a currency indexed to stock performance. Without external pressure, the predictable outcome is that AI efficiencies are treated as another margin tool, not a shared surplus.

Policy can at least tilt the playing field. Tax advantages can be tied to more substantive worker ownership structures, not just nominal stock plans. Legal frameworks can support collective bargaining over how automation gains are split—wages, hours, or equity—rather than letting those gains silently accrue to capital. Training funded by employers can be made portable, so workers don’t have to accept golden handcuffs just to keep the value of their own skill upgrades. None of that guarantees broad-based wealth, but it moves outcomes away from “all upside to capital, all risk to labor.”

There’s a popular counter-argument: technological progress eventually lifts many boats, so we just need patience. History does show broad gains over long arcs. But the distributional pattern of those gains has never been automatic. It’s been mediated by unions, antitrust enforcement, labor law, and political bargaining. Treating AI as another inevitable tide that will “work itself out” ignores how institutional settings determine who actually captures productivity growth. I’ll give Ramaswamy credit for seeing the moment as a chance to rethink worker wealth; I push back because possibility without institutional realism is how you end up with nice headlines and flat median net worth.

Practical markers would help turn his thesis into something testable. Which forms of ownership actually give workers enforceable claims on profits? How do we structure training so it’s not just a perk for high performers but a pathway that survives layoffs and reorganizations? What happens to worker stakes when a downturn hits and management reaches for the usual tools—headcount, benefits, and long-term investment? Vague exhortations about building wealth don’t change who sits in the room when those calls are made.

So yes, workers can build wealth in the AI era—but only if ownership means more than a ticker symbol in a benefits portal and training means more than a slide in an investor deck. If Ramaswamy’s right, we’ll see it not in op-eds, but when corporate charters and capital allocation start to look different on paper.