The largest U.S. hyperscalers may carry more AI-related debt exposure than public information shows, Mark Habib, S&P Global Ratings’ global head for digital infrastructure, said at an Oct. 19 roundtable, Reuters reported, complicating credit assessments.
The warning comes during a surge in spending that S&P expects to strain the companies’ cash generation. Its Sept. 3 outlook puts combined data-center and AI-related capital expenditure by Alphabet, Amazon, Microsoft, Meta, Oracle and SpaceX above $7 trillion from 2025 through 2030.
S&P forecasts negative free operating cash flow for each of the six in 2026 and 2027 and does not expect recovery until 2029. Its models generally assume revenue growth will accelerate in 2028 as growth in capital spending slows.
Reuters reported that Habib described financing in which a hyperscaler effectively guarantees a large data-center development or provides backing to a smaller AI company. Some resulting obligations may be absent from company disclosures and from transactions S&P rates, he said. Habib said the agency had no estimate of how much additional debt such arrangements might represent.
In its public research, S&P said companies are financing AI infrastructure through debt, equity issuance, leases and other arrangements. It separately said wider use of joint ventures, special-purpose vehicles and residual-value guarantees makes credit analysis more complex.
Earlier EconoTimes coverage of CoreWeave’s borrowing costs described financing pressure on the AI cloud provider. Habib’s warning brings the financing ties between those projects and major technology companies into focus.
Asked whether a selloff in AI-related bonds showed that investors saw more risk than S&P’s ratings reflected, Habib said the agency saw no major disconnect requiring a ratings change at that point, Reuters reported. He identified profits falling short of expectations as the main risk to hyperscalers’ ratings.
S&P said it is tracking the monetization of AI investments, the durability of demand, potential overcapacity and how contractual commitments resembling debt should be treated.
