Key points
- Anthropic modeled three AI scenarios for the US economy
- Extreme case: GDP 32% above baseline, 14% lose jobs
- Capital's share of income rises in every scenario
- The authors name no most-likely outcome
Anthropic has modeled three ways AI could reshape the US economy by 2030. In the most aggressive scenario, the economy is 32% larger than it would be without AI, but nearly 14% of workers lose their jobs, and fewer than half find new ones. The tool, Scenarios for our Economic Future, doesn't identify a most-likely outcome.
The model breaks jobs into tasks that AI can take over, help with or leave alone. It then estimates how those changes affect economic output and how the income is divided between workers and owners of capital.
The three scenarios
| Scenario | US GDP in 2030 | Change versus baseline | Labor share | Capital share |
|---|
| Modest | $34.1 trillion | +1.6% | 59.4% | 40.6% |
| Substantial | $36.3 trillion | +8.3% | 56.1% | 43.9% |
| Extreme | $44.4 trillion | +32.4% | 45.2% | 54.8% |
Figures as reported by Anthropic's scenario tool, version 1.0, September 2026. The tool compares each scenario with the same economy without AI. All three imply a 2030 baseline near $33.5 trillion by our arithmetic, so the percentages are gains against that counterfactual rather than growth from today's economy. A larger economy does not mean every household gains in proportion.
The modest case treats AI as a technology on the scale of the internet. The substantial case has AI handling half of knowledge work and the economy growing at twice its normal rate. The extreme case has AI beating people at most knowledge work, and NPR reported that GDP grows at more than seven times its current pace while nearly 14% of workers lose their jobs to AI, with less than half of them finding new ones.
The line that matters for investors runs across the last two columns, the split between what gets paid to workers and what accrues to the owners of the machines. Anthropic puts the current split at about 60 cents of every dollar to workers and 40 cents to capital. Capital's share sits above that in all three scenarios, at 40.6% in the modest case and 54.8% in the extreme one.
What Anthropic says about it
Anton Korinek, Anthropic's head of transformative AI economic studies, said the outcome depends on adoption as well as capability. "If the AI can do amazing things but nobody uses it, then it's not going to have an economic impact," he told NPR. Co-founder Jack Clark said the harder part isn't the technology. "I think the technology will keep developing at a very, very fast and sustained rate but diffusion of the technology will likely be more challenging than people think," he said.
Clark also framed the upside as a policy question. "If you end up with this level of GDP growth, you have moves available to you as a policymaker that are unimaginable today," he said. Anthropic paired the model with a survey of nearly 11,000 people, who on average expected both a real productivity gain and real disruption to workers in AI-exposed fields.
What separates the scenarios is how quickly workers and companies take the tools up, how much of a task AI performs rather than assists with, and whether it creates new work to replace what it displaces. The model leaves out policy responses, business cycles, aggregate demand and financial market disruptions, and does not track individual workers, which limits what it can say about the cost of displacement.
Every named voice in NPR's account of the model works for Anthropic, which sells the technology being modeled. The company has separately warned this year that AI systems able to design their own successors could arrive sooner than institutions expect.
