MINNEAPOLIS
Federal Reserve Bank of Minneapolis President Neel Kashkari said on September 30 that he is skeptical the U.S. economy outside of artificial intelligence construction is contracting. Speaking at a Council on Foreign Relations event, Kashkari challenged the view that growth is primarily sustained by data center investment.
Bloomberg Economics estimates AI-related investment contributed roughly half of the U.S. economy's approximately 2 percent GDP growth over the past year. Other analyses suggest AI capital spending drove about three-quarters of total U.S. GDP growth in the first quarter of 2026.
AI capital expenditure is projected to rise from around 1.8 percent of GDP this year to 2.5 percent in 2027. Kashkari said he does not dispute these figures but questions their broader interpretation regarding the economy's underlying strength.
Kashkari pointed to strong corporate profits across numerous sectors beyond AI and resilient consumer spending as evidence for a sturdier economic foundation. His argument suggests the AI boom is substantial but does not prop up a hollow economy—a contrast to the view that any disruption in hyperscaler spending or a disappointing earnings report from Nvidia could trigger a broader macroeconomic event.
Investor Michael Burry, known for his subprime mortgage short before 2008, has accelerated his timeline for an AI bubble burst. After reviewing an Ares Management report on the fragility of AI financing structures over the weekend of September 27, Burry indicated increased confidence the bubble will burst within the next year, advancing his previous 2028 base case.
Goldman Sachs strategist Peter Oppenheimer reinforced his AI earnings bubble warning with new data. His analysis shows AA-rated tech capital expenditure grew 65 percent year over year in the second quarter. AI borrowers now represent 44 percent of the $135 billion in U.S. convertible bond issuance recorded this year, underscoring concentration risk within the capital markets.
From the bond market's perspective, this debate carries material implications for duration risk and spread compression. If economic growth is broadly distributed, it could mitigate systemic risk from an AI sector slowdown and support credit quality across wider-ranging issuers. A concentrated growth engine, conversely, increases duration risk for portfolios heavily weighted toward tech-linked credit.
