Goldman Sachs chief economist Jan Hatzius stated that AI investments contributed effectively nothing to US GDP growth in 2025. He noted that most essential AI hardware is produced in South Korea and Taiwan, limiting the immediate economic impact of infrastructure spending within the United States.
The bank projects global capital expenditure on AI to reach approximately $1,019 trillion in 2026, with the US accounting for about $581 billion of that total. Analysts suggest that these investments may result in $50 billion in crowding-out effects on other business spending for the year.
Jim Covello, head of global equity research at Goldman Sachs, expressed concerns regarding the unclear returns on planned infrastructure spending. He noted that the economic prospects of AI appear more questionable today than they did two years ago.
During a discussion on August 13, 2026, Hatzius emphasized that the transition to AI is a long-term structural shift spanning decades rather than a short-term growth engine. He also predicted that unemployment rates will remain stable despite the potential for job displacement.
The firm identified two primary risks: high valuations of stocks related to AI and the concentration of capital spending among a small number of major technology companies. Currently, there is limited evidence that AI investments are significantly displacing other business expenditures.
