Anthropic models three ways AI could reshape the economy by 2030

Anthropic's economics team published a model of how AI might change the US economy by 2030, built around three scenarios rather than a single forecast. In the modest case, AI lands like the internet did: GDP rises about 1.6 percent to $34.1 trillion, unemployment stays near its historical range, and wages are roughly flat. In the substantial case, AI handles about half of knowledge work, growth runs at double the normal rate to $36.3 trillion, and displaced workers move into other occupations. The extreme case has AI surpassing humans across most knowledge work, with GDP up 32.4 percent to $44.4 trillion and annual growth reaching 15 percent.

The point of the exercise is not the top-line numbers but where the money goes. GDP grows in every scenario, yet labor's share of income falls as AI does more of the work, and capital's share climbs from 40 percent today to 43.9 percent in the substantial case and 54.8 percent in the extreme one. Job losses concentrate in knowledge work while demand for manual labor rises, and in the extreme case knowledge-worker wages drop more than 10 percent even as average wages go up. Anthropic is candid that the model leaves out policy responses, business cycles, and catastrophic risks, calling it a stark simplification.

Why it matters

If your income depends on knowledge work, the scenario that should worry you is the one where the economy grows fastest, because that is also where your bargaining power shrinks most. The model gives policymakers concrete distribution numbers to argue over, rather than a vague sense that AI is good or bad for jobs.