Imagine the US economy nearly 50% bigger in just four years. That is the eye-catching number Anthropic cofounder Jack Clark put on the table on October 17, 2026, in front of finance ministers and central bankers at the IMF and World Bank annual meetings in Bangkok. One of Anthropic's own internal economic models, he said, shows US economic output could reach $44 trillion by 2030 if AI keeps improving and spreads through the economy fast enough.
Before you get too excited — or too nervous — Clark himself called the figure “outlandish.” And he was talking about a scenario, not a prediction. Still, when one of the world's top AI labs stands up in front of the world's top economists and makes a claim like this, it is worth a closer look.
What the model actually says
Anthropic's economics team built the model to explore how transformative AI could affect economic output, jobs, wages, and the distribution of wealth. The company published it in September 2026 as an interactive tool called “Scenarios for our Economic Future,” based on a technical report titled “Economic Scenarios for Transformative AI.”
The model lays out three paths to 2030, all measured at 2025 price levels:
- Modest: US GDP reaches $34.1 trillion — roughly 1.6% annual growth, close to the status quo.
- Substantial: GDP rises to $36.3 trillion — about 8.3% growth, a major AI-driven boost.
- Extreme: GDP hits $44.4 trillion — a 32% jump that would push annual growth toward 15%.
Today's US economy sits at just over $30 trillion, so the extreme case means adding roughly half the current economy's size in four years.
What would the extreme scenario require?
Clark was blunt about the conditions. In the extreme scenario, AI would perform almost all knowledge-based tasks, businesses would adopt it at an unprecedented pace, and AI systems would reach a level of autonomy and efficiency we have never seen.
As Clark put it, a boom of that size would need entirely new industries and inventions to be created to power it — economic doctrine simply does not allow a jump like that from productivity gains alone. Something genuinely new has to emerge.
One key piece of that puzzle, Clark said, is recursive self-improvement: AI systems that improve themselves faster than humans can. He told the Bangkok audience that he now expects full recursive self-improvement capabilities as early as 2027. The extreme GDP scenario essentially depends on that capability showing up on schedule.
The jobs warning came in the same speech
Clark did not just talk about growth. He told policymakers that AI is already causing job losses in fields with high AI exposure, including call centres and customer service — while, he claims, it is enhancing employment in scientific research.
His core message was aimed squarely at the policymakers in the room: international cooperation is needed to build “unusual and serious rules” for AI, and governments have only “a small number of years” to put safeguards in place before these changes fully arrive. His remarks echo Anthropic CEO Dario Amodei, who has been pushing for stronger coordination among AI companies as model capabilities accelerate.
Why this might not happen — in Anthropic's own numbers
Here is the part that keeps this story honest. The $44.4 trillion figure is the ceiling of the model, not its forecast. Surveys run alongside the scenarios show public opinion landing closest to the substantial path: respondents estimated roughly a 10% GDP boost alongside about 5% unemployment. The extreme path requires everything to go right — faster capability gains, massive adoption, and the recursive self-improvement loop arriving on time.
It is also worth remembering where this number came from: Anthropic's own internal model, presented by Anthropic's own co-founder. That does not make it wrong, but it is a company scenario, not an independent forecast. The IMF's own upgraded 2026 global growth forecast, for context, sits at a modest 3.1% — a very different picture of the near term.
The bottom line
The real story is not really the $44 trillion number. It is that a leading AI lab now treats radical, four-year economic transformation as a scenario worth putting in front of the IMF — and that its co-founder used the occasion to demand international AI rules with urgency. Whether the economy lands at $34 trillion, $36 trillion, or something wilder, the debate about AI's economic footprint has clearly moved from the lab to the boardroom to the central bank.
What do you think — is the extreme scenario inspiring, terrifying, or just bad modeling? Drop a comment and let us know.
Comments
Post a Comment