Goldman Sachs Says AI Cut U.S. Economic Growth to Near Zero Last Year

Goldman Sachs’ latest macro‑modeling suggests that the $700 billion AI‑related capital outlay slated for 2025 will barely register in the nation’s GDP figures. Chief Economist Jan Hatzius told the Atlantic Council that “AI investment spending has had ‘basically zero’ contribution to U.S. GDP growth in 2025,” emphasizing that the bulk of the hardware—chips, servers and networking gear—is sourced from Taiwan and South Korea, so the value adds to foreign output rather than domestic product (Gizmodo). The firm’s internal calculations, corroborated by analyst Joseph Briggs in a Washington Post interview, show that while U.S. firms such as Meta, Amazon, Google and OpenAI are pouring billions into new data centers, the imported components offset any net increase in domestic investment, a nuance often missed in headline‑grabbing narratives (Gizmodo).

The discrepancy between Wall Street’s optimism and Goldman’s sober assessment mirrors a broader shift in macro forecasts. Reuters reported that Goldman has raised the odds of a U.S. recession to 45% and is the second major bank to hike its recession probability in recent weeks, underscoring growing skepticism about the economy’s resilience (Reuters). By contrast, earlier this year some economists, including Harvard professor Jason Furman, attributed up to 92% of first‑half‑2024 GDP growth to “information processing equipment and software” investments, a claim that now appears overstated in light of Goldman’s data (Gizmodo). The Federal Reserve Bank of St. Louis had also estimated AI‑related spending accounted for 39% of third‑quarter 2025 growth, a figure that Goldman’s analysis calls into question (Gizmodo).

A key factor behind the muted impact is the lack of a reliable metric for AI’s productivity gains. A recent survey of nearly 6,000 executives across the United States, Europe and Australia found that although 70% of firms report active AI usage, roughly 80% see no measurable effect on employment or output (Gizmodo). Without clear evidence that AI is translating into higher labor productivity, the traditional link between capital spending and GDP expansion weakens. Hatzius noted that “there is currently no reliable way to accurately measure how AI use among businesses and consumers contributes to economic growth,” reinforcing the view that the sector’s hype may outpace its real‑world economic contribution (Gizmodo).

The policy implications are immediate. President Donald Trump has invoked AI investment as a justification for a unified federal regulatory framework, arguing that state‑level rules would stifle the “growth engine” (Gizmodo). Yet if the sector’s contribution to GDP is indeed negligible, the case for shielding AI from tighter oversight becomes less compelling. Analysts at Morgan Stanley, cited by Reuters, have already trimmed their 2025 growth outlook on tariff concerns, suggesting that trade frictions could further erode any marginal gains from AI‑related imports (Reuters). In this environment, policymakers may need to reassess whether incentives for AI infrastructure are delivering the promised macroeconomic boost or merely shifting value abroad.

In sum, Goldman Sachs’ assessment reframes the AI‑spending narrative from a driver of robust growth to a largely imported cost center with limited domestic payoff. While the industry continues to pour capital into data centers and model training, the bulk of that spending bolsters foreign manufacturing and does not translate into a measurable lift in U.S. GDP. As recession probabilities climb and the debate over federal versus state regulation intensifies, investors and regulators alike will have to look beyond headline figures and focus on tangible productivity outcomes before treating AI as a panacea for the American economy.