Reality check on Ramaswamy's worker-wealth pitch

Vivek Ramaswamy’s column promises workers a path to wealth in the AI era. Short version: he’s describing a mechanism — ownership — that can work in theory. The piece leans on an appealing narrative: give workers a slice of the upside, and everyone wins. The idea is seductive because it sidesteps politics and points the finger at markets. The math doesn't lie about one thing: ownership transfers returns. But the column skips the harder arithmetic — who gets access, who has bargaining power, and who actually owns the distribution channels.

Let’s start with what the column gets right. Putting ownership at the center is directionally correct. Stock, equity-like instruments, profit sharing, employee ownership trusts — these are all existing tools that convert firm-level gains into household balance-sheet gains. Wage growth alone will not track AI-driven productivity. If the value of labor is being partially replaced by capital (AI systems), then workers need exposure to capital. That’s basic capital-labor dynamics, not ideology.

Now for the part the op-ed glides past: ownership is necessary. It isn’t sufficient. Legal form matters. Vesting schedules, liquidation preferences, and dilution mechanics routinely turn headline-grabbing equity awards into scraps. Option packages that look generous in a press release can be underwater, unexercisable, or wiped out in a down round. Right, this is where the “just give workers stock” slogan collides with capital structure reality.

Then there’s the question of who actually gets the equity. Company-issued options tend to reward founders, early hires, and specialized technologists. The workers likeliest to be hit first and hardest by automation — retail staff, warehouse workers, home health aides — rarely sit inside option pools. When they do, the grants are often too small, too illiquid, or too constrained by employer-specific risk to function as real wealth-building instruments. So the mechanism the column touts is selective by design unless someone deliberately reengineers it for scale, especially in low-margin, high-churn sectors.

Access and distribution mechanics are the real constraints, not the lack of inspirational rhetoric about “democratizing AI.” Who gets the tools? Who gets the training? Who is in the room deciding how AI-generated surplus is carved up? The column nods toward broad access without naming the gatekeepers: investors, boards, labor institutions, and regulators. These are the arenas where it’s decided whether equity grants are meaningful or symbolic, whether profit sharing is recurring or discretionary, and whether training is a line item or a press-release prop.

Geography and sector make the picture even more skewed. AI returns don’t fall like rain; they pool. They concentrate in specific ecosystems with capital, specialized talent, and favorable rules. Workers inside those clusters can negotiate for equity or spin out their own ventures. Workers in thin-margin service sectors or in regions with little tech infrastructure mostly experience AI as something done to them, not with them. That deepens regional and sectoral inequality, even if the national average productivity line looks pretty.

Here’s where history is useful. The last time the U.S. had anything resembling broad-based asset growth tied to a major technological and policy shift, it was the postwar period: strong unions, expanding homeownership, defined-benefit pensions, and a tax code that channeled some returns from capital back into public goods. That wasn’t a “market will handle it” equilibrium; it was a negotiated one. Since then, we’ve already run a real-world experiment in partial worker ownership — think of how many employees at big companies hold stock through retirement plans — and it still hasn’t offset wage stagnation and concentration of corporate control. AI doesn’t reset that history; it compounds it.

From my decade at Goldman I learned to read incentives instead of headlines. Incentives define outcomes. When firms control dilution terms, buyback decisions, and board composition, internal incentives will tilt toward protecting founder and investor stakes unless something external changes the payoff structure. That reality is more powerful than any upbeat narrative about “sharing the AI upside with everyone.”

The standard counter-argument is familiar: competition for talent will solve this. New AI firms will need workers, so they’ll throw equity at them; incumbents will match or lose out. That’s plausible at the high end of the labor market, where engineers walk out the door with competing offers. But market-driven equity distribution is patchy. It barely touches workers whose bargaining power is weak, where automation is mostly about labor cost reduction rather than capability expansion.

Ownership also has a time problem. Equity is volatile and back-loaded; household expenses are stable and front-loaded. Workers who are already financially fragile can’t wait for a liquidity event that might never come. Without protections — minimum cash components, limits on concentration risk, standardized disclosures about downside scenarios — “more equity” just shifts risk onto people least able to absorb it.

A practical worker-ownership regime in an AI economy would look more portable and more standardized than what the column sketches. Portability matters because workers change jobs, sectors, and locations. Ownership that evaporates every time you switch employers is not a wealth strategy; it’s a retention tool for companies. Policy could push in a different direction: portable capital accounts with employer contributions, profit-sharing formulas that follow the worker rather than the firm, and tax preferences tied to equity designs that protect rank-and-file employees instead of just executives.

We’ve also seen what happens when a sector-wide technology shock arrives without structured worker claims on upside. Think of ride-hailing: a glossy tech narrative, significant productivity in matching riders and drivers, and a generation of workers treated as disposable inputs with minimal asset accumulation. That’s the default setting if ownership rules aren’t specific, enforceable, and designed beyond Silicon Valley engineers.

Two operational fixes would move this out of pep-talk territory. First, create simple, standardized equity-like instruments tailored for rank-and-file workers — with clear vesting, basic anti-dilution protections, and portability across employers within an industry or region. Second, tie large-scale AI deployment to mandatory worker-transition investments, such as reskilling pools or contributions into those portable capital accounts. Those are not radical ideas; they just align costs and benefits over time.

Ramaswamy is right to say that workers need a stake in AI-era capital, not just a paycheck. If that column ever gets a sequel, the interesting question won’t be whether ownership matters, but whose term sheet writes the rules for it.