Anthropic model says AI could lift U.S. GDP sharply while shifting gains to capital
Anthropic has published a new study, "Economic Scenarios for Transformative AI," that models how AI could reshape the U.S. economy between 2026 and 2030. The report lays out three scenarios — modest, substantial, and extreme — based on AI capability, adoption speed, automation, and resulting productivity gains. In every case, the economy grows. In the most aggressive scenario, U.S. GDP in 2030 stands 32.4% above a no-AI baseline and annual GDP growth reaches about 15%.
The study also points to a much less even distribution of those gains. Knowledge workers, including software engineers, administrative staff, sales roles, and professional services, face the largest direct hit in the model. Under the extreme case, knowledge-work employment falls 21.5% from mid-2026 levels, unemployment for those workers rises to 17.9%, and the overall unemployment rate reaches 11.9%.
Anthropic’s model suggests labor’s share of income could fall from 60% to 45.2% in the extreme scenario, while capital’s share rises to 54.8%. Even with the economy much larger, aggregate worker income barely increases, with most additional output going to capital owners. A separate survey of 10,980 U.S. adults conducted with Morning Consult found median public expectations closest to the report’s "substantial change" scenario.