AINEWS
Anthropic, the AI lab behind Claude, says its own model puts the U.S. on track for $34.1T to $44.4T by 2030, with the adoption of AI as the lever.
Anthropic, the AI lab behind Claude, published an interactive U.S. economic model this week that maps three possible futures for 2030: a modest lift of 1.6% to $34.1 trillion, a substantial lift of 8.3% to $36.3 trillion, or an extreme lift of 32.4% to $44.4 trillion. The single lever that moves the dial between those outcomes is the speed at which AI gets adopted. The product Anthropic sells is on the other end of that lever.
The model, titled "Economic Scenarios for Transformative AI" and built by Anthropic's economics team led by Joseph Korinek, runs on the supply side of the economy only. It asks what happens if AI capabilities continue to scale and a growing share of work tasks are automated over the next four years, and it sizes the GDP impact from labor-augmentation gains alone. The model explicitly excludes policy responses, business cycles, aggregate demand, financial-market disruptions, and possible catastrophic risks. A reader adjusting assumptions in the companion scenario explorer is choosing how fast AI penetrates the labor market, not whether the Federal Reserve cuts rates or whether a recession hits.
In the modest scenario, knowledge workers see modest displacement and most of the GDP lift shows up as broader productivity gains. In the substantial and extreme scenarios, the model's authors write that "coders and call-center agents may have to switch to jobs like electrician and nurse." The model treats those roles as least exposed to current AI capability. That line, which circulated in Tom's Hardware coverage and the Fortune write-up, is an illustrative individual example within Anthropic's model, not a national reallocation plan. Across the scenarios, average wages rise; the model's own framing warns that the gains may not be broadly shared and that the capital share of income can rise.
The variable Anthropic's economics team chose to move is adoption of AI, with adoption of its own product as one input to that lever. That choice is what readers should keep in mind when a vendor publishes a macro range whose top end would lift U.S. output by roughly $10 trillion in four years. The companion survey Anthropic ran alongside the model asked 10,980 members of the general public and 26,173 site visitors to weigh in. The typical respondent's answers lined up with a "substantial change" outcome (GDP +10% by 2030, unemployment around 5%), with about 10% of respondents placing themselves in the extreme scenario.
A useful counterweight comes from economists Ben Moll and Alex Imas, who argue that double-digit AI-driven growth is implausible on historical and growth-accounting grounds. A "10 times richer in 15 years" trajectory implies 16.6% growth per year, and 100 times richer in 30 years, neither of which lines up with how productivity has actually compounded in past technology cycles. Their critique is not a refutation of Anthropic's modest and substantial scenarios, which sit in the single-digit range, but it puts a floor under the upper bound.
The portable lens, then, is the same one a reader would apply to any vendor-published economic forecast: which variable moves the range, and who benefits if it moves up. The model leaves the policy path out. The 200-plus economists and executives who signed a July 2026 open letter urging research into AI's economic effects, among them Joseph Stiglitz, Paul Krugman, Daron Acemoglu, Eric Schmidt, and Reid Hoffman, are trying to fill that gap.