AI capital spending faces higher-rate test as cash flow gap widens
US AI investment has helped offset monetary tightening, but analysts point to equity declines and rising bond reliance as risks to the boom.
By David L. Chen · Senior Columnist
· 3 min read
US nonresidential fixed investment continued to rise through 2022 and 2023 even as the Federal Reserve lifted the federal funds rate, according to an Econbrowser analysis of Bureau of Economic Analysis and Federal Reserve data. The investment strength, tied by the analysis to artificial intelligence-related capital spending, has become a central support for US growth, making its durability a macroeconomic question for investors and policymakers.
The analysis cites BEA data, Federal Reserve figures and the Atlanta Fed’s GDPNow estimate from July 10 for second-quarter 2026 growth. It says a sizable share of estimated US growth in that quarter appears to come from nonresidential fixed investment, although the conclusion depends partly on whether April and May trade data accurately represent computer and semiconductor imports for the full quarter.
The accounting matters because imported technology can lift business investment while subtracting from gross domestic product through the import line. If a US company buys servers, chips or related equipment from abroad, the spending can appear in investment, while the imported content is deducted elsewhere in the GDP calculation. Econbrowser said recent quarters showed nonresidential investment contributions partly offset by imports of computers, computer accessories and semiconductors.
AI spending has filled a gap left by monetary tightening
Because there is no direct official split for AI-specific investment, Econbrowser uses changes since the third quarter of 2022 as a proxy, with the November 2022 release of ChatGPT treated as a marker for the start of the current AI capital spending cycle. The categories examined include software, information equipment, and power and communication structures, based on BEA data and the site’s calculations.
The Bank for International Settlements estimates that US hyperscalers and other AI companies will invest $800 billion in 2026, up from about $750 billion in 2025. That scale helps explain why higher interest rates have not slowed capital formation as much as past monetary cycles might have suggested.
Econbrowser also compares the current period with the 2000 dot-com peak, when imports of telecom and computing equipment served as a leading indicator for nonresidential investment. The analysis notes that information equipment investment data are available only through the first quarter of 2026, while monthly import data for computers, computer accessories and semiconductors run through May. A combined series suggests a possible softening in the second quarter, though the author cautions that the quarter’s estimate is based on only two months of data and may be revised.
Financing mix may make spending more rate-sensitive
The analysis identifies two reasons AI-related capital expenditure could fall short of current expectations. First, Bloomberg data show the Magnificent 7 equity index has declined from its peak. Econbrowser argues that weaker equity values can raise the cost of capital for AI-linked companies and reduce wealth effects that support household consumption.
Second, The Economist reports that investment by the largest technology companies is now running ahead of cash flow, increasing their reliance on external finance through bond markets. Econbrowser says this shift moves firms higher up the financing pecking order: internal cash flow is relatively cheaper and less directly tied to market rates, while bond issuance exposes investment plans to prevailing yields and credit spreads.
The analysis uses Microsoft bond yields as a benchmark for a low-risk corporate rate and notes that some AI-linked firms may face a risk premium above that level. It also cites the investment framework associated with Fazzari and co-authors, under which higher corporate bond rates can reduce investment by raising the required return on projects.
Econbrowser concludes that AI capital spending, which appeared less affected by higher interest rates earlier in the cycle, may become more sensitive as firms depend more on bond financing. It also says uncertainty over future cash flows could weigh more heavily on investment, while policy uncertainty and cost-push inflation may add pressure to interest rates.
This story draws on original reporting from Econbrowser.
