Questioning the AI Wealth Promise for Workers

Ramaswamy says workers can build wealth in the AI era — and then hands them a pep talk instead of a balance sheet. He names hope as a strategy, optimism as a plan, and faith in individual hustle as a policy. Frankly, that's a political argument dressed up as economic advice; the math doesn't lie about incentives, and Ramaswamy never shows the ledger.

He does, however, put his finger on one thing that actually matters: ownership. Owning capital instead of selling labor is how wealth accumulates. That part is directionally right. The problem is he treats ownership like an attitude — a mindset you can adopt — instead of a design issue baked into contracts, cap tables, and corporate law.

Ownership without the wiring diagram

So let’s start with that. Ramaswamy is correct to say workers should become owners. But who gets equity, how it's priced, and who controls liquidity events are not philosophical questions; they’re term-sheet questions. They live in equity plans, vesting schedules, and board approvals.

Telling workers to “own capital” is useful only if they have actual, durable ways to acquire it: employee stock ownership that isn’t quietly diluted in the next funding round; profit-sharing that doesn’t vanish the first time guidance is missed; equity that doesn’t evaporate when you change jobs or get reorganized into a different entity. None of that shows up in his argument.

From my Goldman days I learned that structure beats intent every single time. Compensation plans are legal instruments; tweak the instrument, and you tweak outcomes. Stock-based pay that’s concentrated in top leadership will reliably funnel AI gains upward. Broad-based plans that survive downturns and restructuring will do the opposite. That’s not ideology. That’s the wiring diagram.

You don’t have to look far for examples. When Microsoft expanded broad stock awards during its cloud buildout, ordinary engineers and product managers saw very different upside than workers at firms that kept equity locked at the top. Contrast that with gig platforms where “partner” is a branding exercise, not an ownership stake. Same technology buzzwords, entirely different balance sheets for workers.

Ramaswamy never engages with this distinction. He celebrates ownership as a virtue and skips ownership as a contract.

Skills are necessary — but they're not a safety net

Where he goes next is the standard sermon: learn new skills and you’ll prosper in the AI economy. Again, partially right. Skills matter. AI will absolutely reward people who can work alongside it instead of being displaced by it.

But treating reskilling as a universal safety net ignores unequal starting points and real-world frictions. A software engineer in a major tech hub has different odds than a home health aide, warehouse worker, or call-center rep in a smaller city. Caregiving responsibilities, access to reliable broadband, the ability to forego income while training — all of that changes who can realistically “retool.”

Saying “get educated” without funded programs, without employer-linked apprenticeships, and without credentials that employers actually honor is telling people to run a marathon in their work clothes. Transition costs are not abstract: lost wages while you reskill, relocation when the local labor market dries up, childcare and eldercare gaps when you add night classes on top of a full-time job. These aren’t edge cases; they’re exactly what determine whether a training promise becomes a raise or just more debt and burnout.

The column never runs that calculus. It treats skill acquisition as an on-off switch, not a constrained optimization problem where time, money, geography, and family structure are hard limits.

Concentration, political economy, and who sets the rules

Ramaswamy also treats market outcomes as if they’re neutral, which is cute given the current AI power structure. Technology markets are already heavily concentrated; the firms that own the data, compute, and distribution rails will decide access, pricing, and partnership terms. If you think worker wealth-building is just about individual hustle in that environment, you’re ignoring who designs the game board.

There are obvious levers he sidesteps: tax rules around capital gains and stock compensation, regulations that shape how and when companies issue equity, tools that strengthen bargaining power for workers in highly concentrated markets, and incentives for genuine profit-sharing tied to verifiable metrics. You can argue government shouldn’t pick winners. Fine. But pretending “don’t interfere” is neutral ignores that it entrenches the people and firms already holding the assets.

This is where historical memory helps. When industrial automation reshaped manufacturing, some firms paired technology adoption with pensions, strong unions, and profit-sharing; others chased short-term margins and shed workers. Same robots, different rules, very different wealth trajectories for workers. AI is heading for the same fork in the road, and telling people to “be owners” without touching the rulebook is essentially choosing the path that favors incumbents.

The entrepreneurship escape hatch

One predictable counter-argument goes like this: some people will do exactly what Ramaswamy prescribes. They’ll start companies, surf the AI wave, and become capital owners. True. Entrepreneurship is a real route to wealth.

But let’s be real: entrepreneurship is selective by design. It rewards risk tolerance, access to networks, and a financial cushion. When that’s your main pathway to worker wealth in an AI-driven economy, you’re betting on selective mobility and survivorship, not on system design. For every founder whose startup exits, there are many who burn savings, stall careers, and end up back in the labor pool without a safety net.

Addressing that gap requires mixing private incentives with public scaffolding. Employers can broaden equity participation and hardwire it to long-term profit-sharing instead of one-off bonuses. Governments can underwrite portable training, create apprenticeships co-designed with industry, and shape tax rules that favor broad-based ownership rather than executive windfalls. None of this is radical; it’s just admitting that who writes the rules decides who gets the AI dividend.

Absent those changes, the gains from AI will predictably stack on top of existing capital and market power. Compounding does what compounding always does.

Ramaswamy’s optimism will sell because it lets everyone feel virtuous — workers for hustling, companies for “empowering,” politicians for cheering from the sidelines. But in an AI economy built on concentrated platforms and opaque cap tables, the people who actually end up wealthier will be the ones already holding the pen when the ownership documents get drafted.