The Anthropic co-founder presented the company's economic scenarios at the IMF and World Bank meetings in Bangkok, with the top number tied to an extreme AI outcome
Jack Clark, co-founder of Anthropic , told a room full of finance ministers and central bankers that AI could push US GDP to about $44 trillion by 2030. For context, the US economy today sits at just over $30 trillion.
He made the case on October 17, 2026, at the IMF and World Bank annual meetings in Bangkok. He paired the forecast with a call for international regulatory cooperation.
The projection comes from Anthropic’s internal economic model. The company released it in September 2026 as an economic scenarios explorer.
Modest: US GDP reaches $34.1 trillion, reflecting 1.6% growth.
Substantial: GDP rises to $36.3 trillion, reflecting 8.3% growth.
The extreme case assumes AI reaches high levels of autonomy and efficiency. It also requires recursively self-improving AI systems to deliver the full effect.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Clark also moved up his own forecast for when that loop might arrive. He now expects full recursive self-improvement capabilities by 2027.
He tied the updated estimate to faster progress in AI technology. That matters because the extreme GDP scenario depends on exactly this capability showing up.
Public opinion, for its part, is not betting on the moonshot. Surveys indicate that views line up most closely with the substantial growth scenario. Respondents estimated a 10% GDP boost alongside a 5% unemployment rate.
Discussions at the meeting reflected wider worries about how financial and employment strategies should adapt to AI.
The model points to real repercussions from heavier automation. These include greater displacement risks for knowledge workers. In extreme cases, it anticipates a surge in unemployment rates.
For policymakers, the message is that planning for one outcome is not enough. Anthropic’s framework gives three very different futures, and each demands a different response.
A modest path to $34.1 trillion looks like steady progress. An extreme path to $44.4 trillion, paired with possible unemployment spikes, would test social safety nets, tax systems and labor policy all at once.
That helps explain Clark’s push for international cooperation. The research is meant to fill holes in what is known about AI’s economic impact and spark debate over what legislative frameworks should look like.
There is also the question of who is doing the forecasting. Anthropic builds frontier AI systems. A model projecting large economic gains from AI comes from a company with a direct stake in how AI is perceived and regulated.
The key variable to watch is the 2027 self-improvement timeline. It is the hinge between the substantial and extreme outcomes. The second thing to watch is labor data. If displacement among knowledge workers starts showing up in employment figures, the debate will move quickly from conference panels to legislation.