economic modeling

Anthropic’s new economic modeling tool reveals three possible futures for AI’s impact on the US economy by 2030, with outcomes ranging from modest GDP gains to explosive growth paired with severe worker displacement. The company’s most optimistic growth scenario simultaneously predicts knowledge worker wage cuts exceeding 10%, highlighting the distribution risks inherent in rapid AI advancement.

• Anthropic’s substantial case projects GDP reaching $36.3 trillion by 2030, roughly double normal growth rates, with unemployment near 5%.
• The extreme scenario assumes AI that improves its own capabilities without human intervention, pushing annual growth to 15% and unemployment to historic levels.
• A survey of 10,980 Americans indicated median expectations clustering near the substantial case, with only one in ten anticipating the extreme outcome.

• $44.4T GDP in extreme case scenario, compared to $34.1T in modest case
• 45.2% Labor’s share of national income in extreme case versus 56.1% in substantial case
• 15% Annual growth rate in extreme scenario versus 1.6% in modest case
• 10,980 Survey respondents polled by Anthropic in August 2026

Anthropic released an interactive economic modeling tool on September 9, created by its economics team and presented alongside a technical report titled Economic Scenarios for Transformative AI. The tool invites users to input their own forecasts and compare predictions against responses from over 10,000 Americans surveyed in August. The timing proved significant, coming just hours after researcher Jacob Coxon announced his resignation from Anthropic and accused the AI industry of accelerating toward self-improving superintelligence without adequate safeguards.

The release reflects growing attention within the AI industry to long-term economic consequences of advanced automation. As artificial intelligence capabilities expand rapidly across sectors, economists and policymakers face intensifying questions about labor market disruption, income inequality, and the pace of technological deployment. Anthropic’s modeling represents one of the most comprehensive public efforts by an AI company to quantify and communicate these potential outcomes.

Three Distinct Pathways Differ Sharply on Wage Impact and Distribution

The model presents three cases with markedly different outcomes for workers and the broader economy. In the modest case, AI capabilities would match the internet in scale and economic impact. GDP would reach $34.1 trillion, representing a 1.6% increase over baseline projections, while labor’s share of national income would slip only slightly.

The substantial case assumes AI would handle approximately half of all knowledge work by 2030. This scenario projects GDP at $36.3 trillion, roughly double the normal growth rate, with unemployment settling around 5%. Knowledge worker wages would remain flat across the decade, while labor’s share of income falls to 56.1%. This middle-ground scenario reflects a common expectation among economists who believe AI will drive significant productivity gains without causing catastrophic labor market failure.

The extreme case requires AI capable of self-improvement without human direction. Annual growth would accelerate to 15%, lifting GDP to $44.4 trillion, but unemployment would spike to historic levels and knowledge worker pay would fall by more than 10%, with labor’s share of national income plummeting to 45.2%. This scenario represents a form of artificial general intelligence, or AGI, that operates with minimal human oversight or control.

Across all three scenarios, a consistent pattern emerges: faster economic growth correlates directly with more severe wage compression and income inequality. The modeling thus exposes a fundamental tension between maximizing aggregate economic output and preserving worker compensation levels and labor market stability.

Survey Data Shows Americans Clustering Around Middle-Ground Expectations

Anthropic’s August survey found that the median respondent’s forecasts aligned closely with the substantial case scenario, implying roughly 10% higher GDP by 2030 compared to current trajectories. Approximately one in ten survey participants expected outcomes matching the extreme scenario, suggesting widespread public caution about rapid AI advancement’s downside risks.

This measured public stance reflects earlier polling findings. An Anthropic survey conducted in June identified job loss as Americans’ foremost concern about artificial intelligence, outranking privacy violations and misuse by bad actors. Analysis by Goldman Sachs separately identified entry-level technical roles as the segments facing the sharpest hiring disruption from AI adoption, with potential job losses concentrated in younger workers and less experienced professionals.

The survey methodology also gathered demographic and professional background data, revealing variation in expectations across income groups and sectors. Workers in knowledge-intensive fields expressed higher concerns about wage pressure and job displacement compared to respondents in less automation-susceptible industries.

The Distribution Problem at the Heart of Economic Gains

Anthropic frames the three scenarios not as deterministic forecasts but as choices available to policymakers and industry participants. The framing highlights a tension embedded in the modeling: the scenario generating the most aggregate wealth simultaneously produces the least equitable distribution of gains across the population.

In the extreme scenario, the gains from a rapidly expanding economy are unevenly distributed.

Anthropic Economics Team, Economic Scenarios for Transformative AI report

The tool makes this trade-off explicit by design, forcing users to confront the relationship between economic growth velocity and worker income stability. The harder question, Anthropic notes, concerns who ultimately decides which path emerges.

The modeling exercise also reflects academic and policy debates about whether rapid automation requires new social safety nets, universal basic income programs, or modified tax structures to redistribute gains from AI-driven productivity. Different nations and regulatory regimes may make distinct choices about managing these transitions.

The company has positioned the tool as an open platform for public discourse on AI’s economic future, inviting continued input from survey participants and policymakers. The immediate open question is whether the substantive findings around wage compression and income distribution will influence regulatory or corporate governance decisions as AI deployment accelerates through 2025 and beyond.