I think this post usefully punctures a naïve version of the "AI abundance" story—but it also attacks a position that serious people don't actually hold. Very few thoughtful advocates believe AI firms will altruistically give everyone limitless free goods forever, or that superintelligent AIs will somehow abolish scarcity on their own. That would ignore fiduciary duty, politics, and economics. On that much, we agree. The real question is institutional, not a matter of faith about technology. Within a decade or so, there is a plausible scenario in which AI and robotics dramatically reduce the amount of human labor required to produce goods and services. That does not imply the end of capitalism, private property, or markets. Nor does it imply a socialist revolution. What it does imply is that a labor-income–centered distribution system becomes increasingly misaligned with how value is actually produced. We already live in a mixed economy where this adjustment has partially happened. Roughly $3.8 trillion per year—nearly 20% of US personal income—comes from government transfers (EITC, Social Security, SNAP, SSI, unemployment insurance). This is not charity, communism, or a fantasy of "free stuff forever." It is Keynesian welfare capitalism responding to productivity gains that outpaced wages. An AI-driven economy would extend this logic, not abolish it. When companies replace trillions of dollars in labor costs with machines, that delta does not vanish into the ether. It shows up as higher profits, lower prices, higher output, or all three. That surplus is taxable through the same channels we already use: corporate profits, capital gains, consumption taxes, and resource rents. The question is not where the money comes from; it is how the gains from machine production are distributed. A common objection is that if money isn't earned through labor, it somehow loses meaning—that transfers create a closed loop where value collapses. But that’s a pre-Keynesian error. Money is a claim on goods and services, not a receipt for hours worked. Today, retirees, heirs, and trust-fund beneficiaries spend money they did not earn through labor, and firms are perfectly happy to sell to them. What sustains value is productive capacity and demand, not moral narratives about work. If AI systems can produce abundant goods cheaply, and people have purchasing power—whether from wages or transfers—the economic loop remains intact: people buy goods, firms earn revenue, taxes are collected, and incentives to produce remain. Capitalism does not care why customers have money, only that they do. So I agree that "AI abundance" is a false god if it means magical post-scarcity without politics, taxation, or power struggles. But it is not false to say that AI forces a choice: either adapt distribution mechanisms to a world where machines do more of the work, or tolerate permanent under-consumption amid high productive capacity. That choice won't be resolved by billionaires' goodwill or by slogans. It will be resolved—if at all—by pragmatic political leadership capable of updating mid‑20th‑century welfare‑capitalist institutions for a radically different production regime, working squarely within capitalism to ensure that AI-driven productivity does not translate into extreme and destabilizing inequality.