Credit rating agency S&P has warned that as big IT companies’ AI investment financing structures grow larger and more complex, their credit ratings could be shaken, Axios reported on Thursday.

Amazon, Microsoft, Alphabet, Oracle, SpaceX and Meta, the top 6 hyperscalers, are expected to spend more than $7 trillion by 2030 on data centres and AI capital expenditure, the report said. S&P analysts said hyperscalers’ credit strength is gradually weakening. They said capital expenditure is rising faster than expected, while financing structures are becoming increasingly complex and opaque.

Naveen Sarma (나빈 사르마), an S&P analyst who co-authored the report, said most hyperscalers except Oracle have high credit ratings and solid cash flow, so risks are not at a level that threatens their survival. But he said it is a problem that these companies lend their reputations and credit ratings to unproven companies such as neoclouds, data centre operators, Anthropic and OpenAI.

Analysts cited the spread of low-cost open-source models, uncertainty over when investments will be recouped, and revenue inflation driven by circular financing as key factors increasing uncertainty in the AI industry. Much of hyperscaler spending comes from debt, and Amazon raised about $100 billion in the bond market this year. These companies are forming what is known as a shadow lending market by using their own credit ratings to back other companies’ borrowing, and they also hold stakes in OpenAI and Anthropic.

Richard De Chazal (리처드 드 샤잘), an analyst at William Blair Equity Research, said if end demand falls short of expectations, these companies could be hit twice by slower revenue growth and declining investment values. S&P warned that the scale of overlap and interconnectedness in the industry is so vast that even top-rated companies could take a significant hit during an economic downturn.