S&P Global Ratings Executive Says Hyperscaler AI Debt May Be Understated

Mark Habib, S&P Global Ratings' global head for digital infrastructure, said large technology companies may have AI-related debt beyond the ratings agency's view, and S&P has not estimated the additional exposure. He made the remarks at a London roundtable on Oct. 19.

“It is safe to assume that there is more (debt) out there that we are not seeing,” Habib said.

Habib cited financing arrangements in which hyperscalers effectively back large data-center construction or smaller AI companies. He did not identify the companies involved or quantify the potential liabilities. At the time, he saw no disconnect that required an immediate ratings change.

Hyperscalers fund AI infrastructure through debt, equity issuance, leases and other arrangements. Joint ventures, special-purpose vehicles and residual-value guarantees can complicate credit analysis.

Combined data-center and AI-related capital spending by Alphabet, Amazon, Microsoft, Meta, Oracle and SpaceX could exceed $7 trillion from 2025 through 2030. An Aug. 27 projection put their combined capital expenditures above $1.3 trillion in 2027 and forecast negative free operating cash flow for all six in 2026 and 2027, with recovery not projected until 2029.