GDP Understates AI’s Early Impact

Updated: 2025.10.03 19D ago 2 sources
Goldman Sachs estimates AI lifted real U.S. activity by about $160B since 2022 (0.7% of GDP), but only ~$45B (0.2% of GDP) appears in official BEA data. Roughly $115B of AI-linked growth is effectively invisible due to national-accounts methods that don’t map company AI revenues cleanly into value added. This creates a visible gap between the corporate AI boom and reported GDP. — If national accounts are undercounting AI, policymakers and commentators may be misreading productivity, inflation, and growth—shaping interest rates, industrial policy, and the AI narrative.

Sources

Valuing free goods
Tyler Cowen 2025.10.03 66% relevant
The post cites an AEJ Macro paper estimating a $2,152 reservation price to give up Facebook (2017 version) and $231B in welfare (2003–2017), exemplifying how major digital benefits aren’t captured in GDP—parallel to the claim that AI-driven gains are undercounted in national accounts.
AI's Economic Boost Isn't Showing Up in US GDP, Goldman Says
msmash 2025.09.15 100% relevant
Goldman Sachs Saturday note: AI infra revenues +$400B since 2022; BEA shows only ~$45B; analysts calculate ~$115B uncounted.
← Back to All Ideas