How the $44 trillion GDP claim happened: Clark cited Anthropic's public scenario explorer at the IMF, then corrected reports calling it an internal model

Reconstruct the room first. On Saturday, October 17, Anthropic co-founder and head of public benefit Jack Clark spoke at the IMF and World Bank annual meetings in Bangkok. According to CNBC, he said rapidly improving AI capabilities would drive much faster change across all fields, and that one of the company's models projected AI could raise US economic output by almost 50% by 2030 under certain conditions; he put potential US GDP at $44 trillion against a current level of “just over $30 trillion.” He also added the line many people skipped: the figure “sounds very outlandish, I know.”

He also spelled out the preconditions: that the technology keeps getting better, and that — if GDP were to grow this much — economic doctrine says new industries and inventions would have to have been created to power an expansion of that size. In the same speech he called for international cooperation to set “unusual and serious rules” for a transformative technology, said policymakers' window to act is shrinking, and argued that growing evidence points to a “singularity” in which AI development outpaces human control. Those lines say more about his intent than the $44 trillion figure: the point was how narrow the governance window is, not how good the growth looks.

Then came the same-day correction. After Reuters and CNBC reported it, a wave of secondary coverage described the figure as an “internal model” projection. On October 17, Clark replied directly under that framing: “this isn't an internal model, it's the public economic scenario explorer we published a while ago - and this is under the 'extreme' parameter setting.” In other words, what reporters got was not a private estimate but a public web tool anyone can tune, and the setting quoted was the most aggressive of three. That detail governs how the story should be read: it is the output of a public tool, not a research team's bet on the future.

The tool lives on Anthropic's site as the Economic Scenario Explorer, built on the technical report “Economic Scenarios for Transformative AI” (Korinek et al., 2026). Its page describes the three scenarios concretely: in the modest scenario AI's impact resembles the internet's and is hard to see in macro data; in the substantial scenario it exceeds the internet or the railroad; in the extreme scenario AI drives a completely transformed, unprecedented economy, likely powered by recursively self-improving systems and faster adoption. The page frames today's US economy as “over $30 trillion of value created over the past year” and invites visitors to enter their own expectations for capability and adoption to see what 2030 might look like.

The three numbers are reported and all at 2025 price levels: $34.1 trillion of US GDP in 2030 under the modest scenario, $36.3 trillion under the substantial one, and $44.4 trillion under the extreme one. So the $44 trillion Clark cited is not a standalone forecast that AI adds $14 trillion, but the extreme setting's output; the other two land 1.6% and 8.3% above a no-AI baseline. Anthropic also published research that same week, “The AI expectations gap,” comparing a weighted sample of 10,980 US adults with its own staff: 39% of US adults and 91% of staff expect AI to make a Nobel-level discovery by 2030, and staff expect AI to be able to do 75% of their current work by 2030 against 28% for US workers overall.

Two more facts belong next to this, and both are easy to drop. First, CNBC's report notes Anthropic is preparing an IPO expected in November that could value the company at up to $2 trillion — reporting, not a company announcement. Second, in the same speech Clark acknowledged AI is causing job losses in fields with high AI exposure such as call centres and customer service, while boosting employment in areas like scientific research, with the big unknown being whether it creates jobs in entirely new industries. Read together with the $44 trillion, that is the complete story: one speech containing both an extreme growth scenario and warnings about displacement and a closing governance window, of which distribution kept only the first half.