S&P Global warns that the top hyperscalers likely carry more AI-linked debt than is currently disclosed.

S&P Global estimates that the world's biggest tech firms are likely to have more AI-linked liabilities than is currently known, although its top analyst for the sector sees no need to change any ratings at present.

Debt issued by AI infrastructure providers and so-called "neocloud" firms, such as Nvidia-backed CoreWeave, has in recent weeks sunk to increasingly stressed levels amid concerns about the aggressive scale of their borrowing.

S&P's Global Head for Digital Infrastructure, Mark Habib, said the biggest risk to hyperscalers' ratings remained whether expected profits fail to materialise. He also highlighted so-called "interconnected financing structures," in which hyperscalers effectively guarantee huge data centre builds or those of smaller AI companies.

"It is safe to assume that there is more (debt) out there that we are not seeing," Habib said during a roundtable in London.

"If it's not reported, then it's not part of the disclosure by the hyperscalers. Also, if we are not rating the transactions where this liability may accrue, then the fact of the matter is that we might not have the visibility."

S&P expects the top six US hyperscalers (Amazon, Microsoft, Alphabet, Oracle, Meta and SpaceX) to spend more than US$7 trillion ($10 trillion) on data centres and other AI-related capital expenditure between 2025 and 2030, and to have negative free cash flow until at least next year.

Habib said there was no estimate of how much additional debt the circular financing might add to firms' balance sheets, although he said S&P was trying to take a "holistic" view overall.

In July, the agency downgraded Oracle to BBB-, the lowest rung of "investment grade," on concerns about its AI spending and what it called "an uncertain path to profitability."

Habib was also asked whether the recent market selloff of bonds linked to the aggressive AI buildout was a sign that investors see them as riskier than their ratings indicate.

"For now, we don't have the view that there is a major disconnect that needs to force our ratings to change at this point," he said, adding that S&P's scores were designed to take a longer-term view.

He said what could change that view was if it became clear that the amount of money being spent was no longer sustainable.

"And that could come from a variety of drivers," Habib said. "Further increases in the investments (by AI firms), or a view that the revenue rate of competition is getting too intense."