S&P Downgrades Oracle, Calling OpenAI a 'Key Credit Risk' — One Notch Above Junk
Rating agency S&P Global cut Oracle's credit rating to BBB-, warning that roughly half of Oracle's $638 billion in contracted business depends on OpenAI. A look at the AI boom's financial plumbing.
S&P Global has downgraded Oracle’s credit rating from BBB to BBB- — one notch above “junk” status, the level at which many institutional investors are no longer allowed to hold a company’s bonds. The reason, in the rating agency’s own words: OpenAI is a “key credit risk” for Oracle.
The numbers behind the warning are striking. Oracle has become one of OpenAI’s biggest infrastructure suppliers, building enormous data-center capacity for the ChatGPT maker. Its capital spending — money poured into buildings, chips, and power — is now projected to hit $95 billion by 2027, up from an earlier estimate of $60 billion. Meanwhile, OpenAI accounts for roughly half of Oracle’s $638 billion in contractual obligations: business that’s signed but not yet paid for. If OpenAI stumbled, Oracle would be left holding massive data-center capacity it couldn’t fill. S&P notes that Amazon, Google, and Microsoft are in a sturdier position — they have their own internal workloads to absorb spare capacity and deeper financial reserves.
What’s behind this? Think of a credit rating as a lender’s trust score: the lower it goes, the more expensive borrowing becomes. S&P isn’t predicting OpenAI’s collapse — it’s pointing out that Oracle has bet its balance sheet on a single, unprofitable, privately held customer whose true value even professional lenders struggle to pin down. There are other wobbles in the same direction: SoftBank reportedly had to shrink a loan backed by OpenAI shares from $10 billion to $6 billion because lenders couldn’t agree on what the shares were worth, and OpenAI has pushed its stock-market debut back to 2027. None of this is a crash; all of it is the financial system quietly repricing risk.
What this means for you: Nothing to do with your chatbot subscription — ChatGPT works today exactly as it did yesterday. But if you follow the “is AI a bubble?” debate, this is what a genuinely informative signal looks like: not a hot take, but a rating agency putting its reputation behind a specific, documented concern. The AI build-out runs on borrowed money and interlocking promises. Watching who gets downgraded — and who doesn’t — tells you more than most headlines.
Sources
Source: https://www.spglobal.com/ratings/en/regulatory/article/-/view/type/HTML/id/3592348
One of the World's Best Mathematicians Just Revived His 1999 Software — With AI, in Hours
Fields Medalist Terence Tao used AI coding agents to resurrect two dozen dead Java applets from 1999 and finally build an app he abandoned 27 years ago. His honest bug report is the best part.
