Moody's says 'unprecedented' AI spending threatens credit quality of Amazon, Meta, Alphabet and others

Get the Finance newsletter
Daily finance — markets, central banks, M&A, the prints that move money. Free.
- Moody's Ratings warned that the AI infrastructure buildout is forcing even the world's most cash-rich corporations to abandon decades of asset-light operations, leaning on debt, stock sales and off-balance-sheet moves to fund massive physical buildouts.
- Capital expenditure across the six tracked hyperscalers — Microsoft, Amazon, Alphabet, Meta, Oracle and CoreWeave — is projected to hit $785 billion in 2026 and reach roughly $1 trillion the following year.
- Direct debt across the six hyperscalers has reached approximately $460 billion, while lease commitments have ballooned to $1.2 trillion, with more than $820 billion coming from data center leases that have not yet started.
- Alphabet announced an $85 billion equity sale last month, illustrating how even the cash-rich are tapping public markets to fund AI ambitions.
- Moody's flagged a 'circular AI ecosystem' in which hyperscalers invest billions in pre-IPO labs including OpenAI and Anthropic, which in turn spend heavily on those same companies' cloud services, heightening concentration risk.
- Oracle (Baa2, negative outlook — two notches above junk) and CoreWeave (Ba3, high-yield market) face the most immediate pressure, while Moody's does not view the investment-grade ratings of Microsoft, Alphabet, Amazon and Meta as under imminent downgrade threat.
Why it matters: Oracle sits just two notches above junk with a negative outlook, and CoreWeave already operates in high-yield territory with complex private debt structures — making them the first casualties of the trillion-dollar AI capex binge. Even Alphabet's $85 billion equity sale signals that fortress-balance-sheet companies can no longer self-fund the buildout, and Moody's lease-as-debt treatment means $1.2 trillion in off-balance-sheet commitments will soon bind hyperscalers to heavy rent payments.



