Anthropic staff predict 17% GDP growth from AI; broader public already expects 5.4%

Anthropic staff are betting on 17% GDP growth; the broader public is already at 5.4%

Anthropic employees surveyed internally predict 17.1% real GDP growth from AI, with estimates ranging from 6% to 28%. That's a stunning variance, but the takeaway is uniform: no one inside the building expects normal growth rates. The Federal Reserve projects 2% GDP growth. US adults broadly predict 5.4%—already more than double the Fed's forecast.

The gap between Anthropic staff and the general public suggests a calibration problem. Anthropic employees are the most bullish population measured in this informal survey, followed by site visitors at 12.5%, then US adults at 5.4%. The further removed people are from AI labs, the lower their growth expectations fall—though even the general public has priced in an economic boom that dwarfs historical norms.

Anthropic's 17% figure matters because it sits at the intersection of optimism and plausibility. A 17% GDP expansion would dwarf anything the US has experienced in the modern era, unlocking resources to solve infrastructure, healthcare, and other persistent problems simultaneously. It's also the kind of forecast that only makes sense if AI delivers transformative productivity gains across the economy, not marginal improvements.

But the credibility of these predictions depends on specifics that remain murky. The survey did not distinguish between nominal and real GDP growth—a material difference if inflation is material. The survey also offers no breakdown of mechanism: which sectors drive the gains? How does AI translate to measurable productivity? When does this happen?

Anthropic's internal consensus does reflect a broader shift in how AI labs view their own work. People who believe AI is approaching transformative capability tend to expect dramatic economic outcomes. The fact that Anthropic staff cluster around 17% while the general public sits at 5.4% suggests the industry's confidence in near-term AI productivity has not yet propagated into mainstream expectations—or that mainstream forecasters are anchored to historical growth rates and remain skeptical of discontinuous change.