S&P warns hyperscale data center operators' AI debt could be higher than known levels.

According to Yonhap, Mark Habib, Global Head of Digital Infrastructure at S&P Global, said on October 19 at a roundtable meeting in London that hyperscale data center operators (hyperscalers) may carry more AI-related debt than currently known, citing the "interlinked financial structures" formed by their massive data center construction and funding guarantees for small AI companies as one reason. S&P Global projects that the six major hyperscale operators—Amazon, Microsoft, Alphabet, Oracle, Meta Platforms, and SpaceX—will invest over $7 trillion in capital expenditures for data centers and AI infrastructure between 2025 and 2030, with negative free cash flow expected to continue at least through next year. In July this year, S&P Global downgraded Oracle's credit rating by one notch to the lowest investment-grade tier of "BBB-", citing excessive AI spending and uncertain profit prospects. Habib stated that S&P credit ratings reflect a long-term outlook and that he currently does not view the divergence between the market and ratings as severe enough to require an immediate adjustment, but warned that this stance could shift if the scale of capital spending is definitively deemed unsustainable.