Anthropic study projects AI could boost US GDP 32% by 2030 but disrupt jobs

Artificial intelligence could expand the United States economy by up to 32% by the end of the decade, according to new economic projections published in September 2026, though the same wave of automation threatens significant labor market disruption and widening income inequality.

The working paper from the Anthropic Institute explores three distinct paths for artificial intelligence between 2026 and 2030, as reported by Qazinform News Agency. Authored by Anton Korinek, Charles I. Jones, Szymon Sacher, Tess Cotter, and Peter McCrory, the paper emphasizes that its models are exploratory scenarios rather than definitive forecasts and assigns no probabilities to them. The framework divides employment into two categories: cognitive occupations — such as management, professional, sales, and office roles, which accounted for 62.4% of U.S. Employment in 2025 — and physical or personal service roles like construction, repair, and transportation, which the model assumes AI does not directly touch.

Media additions

Under the paper's most extreme scenario, artificial intelligence performs nearly half of current cognitive work by 2030, more than doubling task productivity. Gross domestic product surges to 32.4% above the baseline trajectory without AI, and annual growth accelerates to 15.4% in the year leading to 2030. Yet these economic gains carry severe distributional costs. Labor's share of national income drops from 60% to approximately 45%, while capital owners capture an increase in income. Employment in cognitive fields falls 21.5% from mid-2026 levels, driving overall unemployment to 11.9%.

Wages diverge sharply across sectors in this high-stress model. Cognitive workers experience an 11.5% drop in pay compared to the no-AI baseline, leaving earnings slightly below mid-2026 levels. Conversely, wages in non-cognitive occupations rise by 33.6% as labor in those fields becomes relatively scarce. The authors calculate that government transfers equivalent to roughly 9% of GDP, comparable to the combined scale of Social Security and Medicare, would be required to protect cognitive workers' incomes without sacrificing broader economic gains. However, historical precedent suggests such compensation rarely accompanies major technological transitions.

Broader economic analyses place these findings within a wider historical and structural context. A Third Way report examining the debate notes that past technological shifts, such as the digital age and internet expansion, concentrated wealth despite doubling per capita incomes. MIT economist Daron Acemoglu estimates that 50% to 70% of changes in the modern U.S. Wage structure stem directly from digital automation. While optimists point to productivity surges that could unlock trillions in global economic value, pessimists warn that accelerated automation threatens middle- and lower-income workers.

At the enterprise level, workplace studies reveal a complex mix of anxiety and enthusiasm among employees. Workplace data published by Microsoft shows that 64% of workers struggle with time and energy demands due to mounting digital debt, including constant emails, meetings, and notifications. Microsoft's survey of 31,000 people across 31 markets indicates that 70% of workers would delegate as much work as possible to AI to lessen workloads, and 49% worry about job replacement.

The trajectory of the American economy will likely become much clearer within the next one to two years as adoption scales past experimental phases. Policymakers, industry leaders, and economists will monitor whether productivity gains translate into broad-based prosperity or deepen existing economic divides as the decade approaches its close.