A new analysis by Moody’s Ratings suggests that the six biggest hyperscalers — Alphabet, Amazon, CoreWeave, Meta, Microsoft, and Oracle — will continue to rack up massive debt. That raises the question of whether their expenditures will ever be matched by the spending that companies and individuals will do to use generative artificial intelligence.
Moody’s wrote that rising capital intensity, leverage, and off-balance-sheet commitments are threatening the credit quality of at least some of the largest tech companies.
With the stature and financial standing of these companies, that might seem an extreme statement, but it’s not.
Debt for any company can be necessary and useful if balanced by the ability to pay the debt service. The level of borrowing this industry is seeing is astoundingly high. Moody’s, which is a major credit rating firm covering corporations, said that the hyperscalers will collectively spend about $785 billion this year and nearly $1 trillion in 2027.
For comparison, Uber raised about $15 billion over years and finally started to make a profit. When Google started in 1998, it raised an estimated $35 million in 12 rounds of funding. That was 12 years ago, but hundreds of billions? Inflation would have put an equivalent value today of $66.75 billion.
The current round of spending by the hyperscalers is aimed at data centers and AI infrastructure. Expensive? Absolutely. It requires land, power systems, construction, and high-performance computing equipment. “Heavy capital spending relative to revenue will lead to declining, and in some cases negative, free cash flow and will hurt leverage ratios, to the extent these expenditures are debt-financed,” they wrote.
The companies in question, again, have money and likely good credit standing, but the current situation isn’t like anything in the past. Moody’s said that a “material shift in the structure of their balance sheets is becoming evident.”
There was a time not long ago when these companies could easily finance what they wished. But the scale is just too high to manage without borrowing. A lot of the debt they’ve taken on has been kept off the balance sheet. Estimated for 2026, there is $821 billion in leases and expenses that haven’t started yet, so the sums don’t appear in their balance sheets. Off-balance-sheet financings are a way to supplement borrowing.
Other companies are tied into these corporations, forming circles of current and expected future investment and sales. For example, Microsoft is a major investor in OpenAI, which in turn leases cloud services from the former. Amazon AWS is a big investor in Anthropic and a cloud provider to it. NVIDIA is supposed to acquire equity in OpenAI, while the software company has committed to use NVIDIA chips in the AI data centers. Advanced Micro Devices agreed to grant equity to OpenAI in return for OpenAI agreeing to use AMD’s chips and products. Then there is some deal between Anthropic, Broadcom, Apollo, and Blackstone. “This AI circularity injects a new dimension of concentration and credit risk for the major tech companies should demand for AI products fail to materialize,” they wrote.
It would be very difficult to pull together enough information to see if the companies could reasonably balance their spending with additional revenue. That is particularly true because it assumes that all the projections of how much companies will spend on AI services. Maybe the projections are realistic, but the inter-company structures depend a lot on the types of numbers that executives often embrace over optimistically.
Read the full article here