Anthropic lays out three AI-driven US economic futures

Anthropic’s economics team published a technical report and an interactive model that maps three possible AI-driven outcomes for the US economy through 2030. The tool lets users enter their own assumptions and compare results with survey responses from 10,980 Americans collected in August.

In the modest scenario, AI produces effects similar in scale to the internet and raises US GDP to $34.1 trillion, about a 1.6% gain relative to the baseline projection.

The substantial scenario assumes AI handles roughly half of all knowledge work by 2030. Under that path, GDP reaches $36.3 trillion and growth about doubles the normal rate; unemployment settles near 5% and wages for knowledge workers remain flat.

The extreme scenario requires AI systems that can improve their own capabilities without human help. It projects annual growth near 15%, GDP of $44.4 trillion, historic increases in unemployment and a fall in knowledge-worker pay of more than 10%.

The report also models how income would be distributed across scenarios. Labor’s share of national income slips slightly in the modest case, drops to about 56.1% in the substantial case and falls to roughly 45.2% in the extreme case. The model shows the fastest-growth projection assigns a smaller share of gains to labor.

In the August survey cited in the report, the typical respondent expected GDP to be about 10% higher by 2030, close to the substantial scenario, and roughly one in 10 respondents expected the extreme outcome.

The report and model were posted hours after Jacob Coxon, a researcher at Anthropic, resigned and warned the industry was moving toward self-improving superintelligence.

“In the extreme scenario, the gains from a rapidly expanding economy are unevenly distributed,” the report said. Anthropic framed the scenarios as choices rather than forecasts and invited readers to adjust inputs to see how different technology, labor-market and policy assumptions change the projected outcomes.

The model provides quantified projections for growth, employment, wages and income shares under different assumptions about how quickly and how autonomously AI systems improve and how broadly they substitute for knowledge work.