Reality Check: AI-Driven Worker Wealth

If you tell a warehouse picker that AI will make them wealthy, you’ve skipped three pages of the memo. The article argues that workers can build wealth in the AI era. Fine. But between the slogan and the bank account there are missing institutions, missing incentives, and missing bargaining power. Frankly, optimism about broad-based worker wealth needs a plan for who gets equity, who gets pay raises, and who bears the risk when automation shows up at the shift.

I’ll start with the part I actually agree with: pushing wealth-building through ownership instead of just wages is directionally right. Wages pay bills; ownership compounds. Employee equity programs can turn company upside into household balance sheets when the company grows, and that’s better than pretending time-and-a-half is a path to security in an economy reshaped by software.

But ownership without muscle is just a nice word in an HR slide deck.

Equity is usually skewed toward executives and highly paid knowledge workers; options vest slowly; private shares are illiquid until a sale or IPO; and paper gains evaporate when markets reprice. The math doesn’t lie — without structural changes to how ownership is granted and realized, most frontline workers will hold promises, not portable assets.

So, ownership has to come with power. That means enforceable profit-sharing, broad-based equity that can actually be turned into cash, and governance rules that prevent silent dilution of worker stakes. It also means labor institutions that can bargain for those arrangements; voluntary corporate goodwill won’t cut it. From my years at Goldman I saw the same pattern play out repeatedly: the capitalization table is a political document scripted as finance. Governance rules, not mission statements, decide who captures value.

We’ve seen both sides of this movie already. At Costco, the decision to pay higher wages and offer meaningful benefits isn’t charity; it’s encoded in a model that accepts slightly lower margins in exchange for lower turnover and higher productivity. At the other end, gig platforms like Uber built enormous enterprise value while keeping workers as contractors, carefully positioned outside the boundaries of equity and traditional benefits. Two different institutional designs, two very different wealth outcomes, both justified as “market-driven.”

The column’s second comfort blanket is reskilling: workers can “upskill” into AI-adjacent roles and thereby claim a share of productivity gains. Let’s be real: training matters, but training doesn’t equal opportunity. Sectors differ wildly. A machine that optimizes inventory will raise productivity in a fulfillment center; who captures that uplift depends on labor supply, local wages, and the employer’s negotiating power. Health-care aides, truck drivers, retail workers — these roles face structural constraints that make switching into software development or machine-learning ops implausible for most.

Training programs often come with strings attached: unpaid or underpaid “apprenticeships,” selection bias toward people who already have time, savings, and childcare, and a mismatch between credentialing and real employer demand. Public investment in retraining helps, but retraining without portable benefits, relocation support, and clear wage ladders is just a moral pep talk. You can’t tell a worker to become “AI-adjacent” while their health insurance still evaporates the moment a contract ends.

There’s also a historical pattern worth remembering. When automation hit manufacturing, we heard the same story: displaced workers would smoothly transition into higher-skilled roles in services and tech. Some did. Many ended up in lower-paid, less secure work. Regions that lost industrial jobs didn’t magically reconstitute themselves around new training certificates; they struggled for decades. The narrative moved faster than the institutions.

The heart of the problem is institutional design, not slogans. How do you make ownership liquid without forcing distressed sales when workers need cash? How do you make profit-sharing predictable and enforceable instead of discretionary “bonuses” that vanish in a downturn? What tax and regulatory architecture stops firms from offloading risk onto workers while keeping the upside for existing shareholders? These aren’t ideological questions; they’re design problems, and they’re solvable if anyone actually prioritizes them.

Practical levers exist: profit-sharing formulas baked into payroll systems; employee ownership trusts with legal protections against dilution; stronger collective bargaining rights in geographically concentrated sectors; portability of benefits across gigs and employers; disclosure rules that show, in plain language, how AI-driven productivity translates into profits and how much of that flows to labor versus capital. Without these levers, corporate invitations to “share the upside” will sound like marketing copy, not contracts.

A counterpoint I have to my own pessimism is this: markets and entrepreneurship will create new roles and firms where early workers capture equity and get rich. True. Startups and new platforms have made employees wealthy before. But survival bias skews perception: we glorify the handful of successes and ignore the thousands of ventures that flatten out or die quietly. Entrepreneurship shifts risk onto workers and often requires initial capital, networks, and time — exactly what many low- and middle-income workers don’t have.

There is also a quiet geographic filter. Wealth from early-stage equity tends to cluster where the startups are and where investors already circulate. Telling a warehouse worker in a small town that entrepreneurship is their main route to AI-era wealth is like telling them their retirement plan is lottery tickets. Possible, yes. Policy, no.

If the article’s underlying optimism is that smart policy plus flexible markets will let workers share AI’s gains, then I agree with the spirit. But the devil is in institutional plumbing: who writes the rules, who enforces them, and who can walk away from a bad deal.

Right now, AI is on track to look a lot like prior tech waves: big productivity gains, bigger headlines, and a distribution of rewards that skews upward unless someone forces the script to change.