S&P Global warns that the debt tied to artificial intelligence among top hyperscale cloud service providers may be even larger than currently known.

Marc Jones

Reuters London, October 19 - S&P Global expects that the world's largest technology companies may face more AI-related liabilities than currently known, although the firm's lead analyst for the sector believes there is no need to adjust any ratings at this time.

Due to market concerns over their aggressive borrowing, debt stress levels have surged in recent weeks among AI infrastructure providers and so-called “new cloud” firms such as CoreWeave, which is backed by Nvidia (link).

Mark Habib, S&P Global's Head of Digital Infrastructure, said the biggest risk for hyperscaler ratings remains whether expected profits can be delivered.

He also emphasized the so-called “interconnected financing structures”—where hyperscalers effectively guarantee vast data center projects or smaller AI companies.

“It is reasonable to surmise there is more (debt) in the market that we haven’t yet detected,” Habib said at a roundtable in London.

“If it isn’t disclosed, then it isn’t part of the hyperscaler’s disclosure obligations. Also, if we haven’t rated a transaction that could give rise to such a liability, we may in fact not have sight of it.”

S&P expects the six largest U.S. hyperscale technology companies—Amazon AMZN.O, Microsoft MSFT.O, Alphabet GOOGL.O, Oracle ORCL.N, Meta META.O and SpaceX SPCX.O—to spend over $7 trillion on data centers and other AI-related capital expenditures between 2025 and 2030, remaining at negative free cash flow at least through next year.

Habib said it is currently impossible to estimate how much extra debt this circular financing could add to company balance sheets, but pointed out that S&P is trying to adopt a “holistic” view overall.

Back in July, S&P downgraded Oracle (link) over concerns about its AI spending and what was called an “uncertain path to profitability,” cutting its rating to “BBB-”, the lowest investment grade.

Habib was also asked whether the recent bond market sell-off linked to aggressive AI expansion signals that investors think these bonds are riskier than their ratings suggest.

“At this point, we do not think there’s a material disconnect requiring us to adjust ratings at this time,” he said, noting that S&P’s rating framework is intended to take a longer-term view.

He said this perspective could change if it becomes clear current investment levels are unsustainable.

“And that could be triggered by a number of things,” Habib said. “For example, if investments by AI firms accelerate further, or if the market judges that competitive returns are becoming too aggressive.”

(To assist non-English speakers, Reuters has automated translation of its reports into several other languages. Because automated translation may have errors or lack necessary context, Reuters does not guarantee the accuracy of translated versions, which are provided solely for reader convenience. Reuters accepts no liability for any damage or loss arising from use of automation translation features.)