Anthropic Releases 2030 AI Economy Explorer With a Stark Split in Who Gains

Anthropic has released an interactive economic scenario explorer that asks users to set assumptions about AI and see what they imply for the U.S. economy in 2030. Its most consequential finding is not that AI raises output in every scenario. It is that the fastest-growth case also sends a far smaller share of income to workers.

A forecast built from task-level choices

The explorer accompanies Anthropic’s technical report, Economic Scenarios for Transformative AI. It represents occupations as bundles of tasks, then treats AI as something that can leave a task unchanged, help a person do it, automate it, or create new work. The results vary with assumptions about AI capability, adoption, autonomy and productivity.

That structure produces three futures rather than a single prediction. In the modest case, Anthropic compares AI’s effect with the internet’s: GDP in 2030 is 1.6% above its modeled no-AI baseline. The substantial case assumes AI can do half of knowledge work by 2030, mostly autonomously, but is not adopted for all of that work. GDP is 8.3% higher and knowledge-worker wages are essentially flat.

The extreme case makes the trade-off visible

The extreme scenario is a different proposition, not simply more of the same. It assumes AI becomes more productive than people at the vast majority of knowledge-work tasks, performs nearly all of them autonomously, and creates essentially no new knowledge tasks for people. Anthropic says this path would likely require recursively self-improving AI and rapid adoption across knowledge work.

Under those assumptions, the model reaches a $44.4 trillion U.S. economy in 2030, 32.4% above its no-AI baseline. Annual growth reaches roughly 15%, enough to double the economy every 4.5 years. Yet knowledge-worker wages fall by more than 10%, and unemployment rises beyond typical recessionary levels as people displaced from affected occupations take time to find other work.

A larger pie, with a different split

Anthropic’s extreme scenario pairs faster growth with a smaller labor share. The model starts from roughly 60% of economic output going to labor. That falls to 56.1% in the substantial scenario and 45.2% in the extreme one, while capital’s share reaches 54.8%. The point is not that every worker’s pay must fall for labor to lose ground; output can grow while a greater fraction of it flows to capital owners.

The model also distinguishes occupations. In more transformative scenarios, it projects knowledge workers moving into jobs less exposed to AI, while wages rise for workers outside knowledge work. Anthropic’s example is that faster design and permitting could raise demand for physical work such as construction. Such occupational switching is difficult, however, and the model ties that friction to higher unemployment during the transition.

Public expectations sit nearer the middle

Anthropic surveyed 10,980 Americans on expectations for AI capabilities, adoption and job adjustment. The typical respondent’s answers implied an outcome close to the substantial scenario: GDP 10% higher by 2030 than without AI and overall unemployment near 5%. About 10% of respondents gave answers aligned with the extreme scenario.

The explorer is best read as a conditional map, not a settled outlook. Its central divide turns on whether AI mostly augments work or becomes reliable enough to take it over autonomously at speed. Anthropic’s release makes a sharp case that measuring GDP alone would miss the question its own scenarios put in front of workers, companies and policymakers: who receives the gains.