One Customer, Half the Order Book

Ten months ago Larry Ellison had the single best morning in the history of business. Oracle reported a backlog stuffed with contracted AI cloud revenue, the stock rose as much as forty three percent in a day, and Ellison gained one hundred one billion dollars overnight, the largest single day wealth gain ever recorded on the Bloomberg Billionaires Index. The customer behind that backlog was OpenAI, and the contract was three hundred billion dollars of computing capacity over roughly five years, sold to a company that has never earned a profit. 

Read that again, because it is the entire risk register. One counterparty. Unprofitable. Five years. Priced by the market as though the cash were already in the vault.

Here is the ten month update. On July 9 S&P cut Oracle from BBB to BBB minus, one notch above junk, and named the concentration plainly: OpenAI accounts for roughly half of the company’s six hundred thirty eight billion dollar backlog. Credit default swaps on Oracle now trade at levels last seen in 2008. Total debt sits near one hundred sixty billion with free cash flow projected at negative forty two billion, and Moody’s holds a negative outlook. The stock has fallen from three hundred forty five dollars to a fifty two week low under one hundred fifteen, erasing roughly six hundred billion in market value. 

The part that should interest anyone who does this for a living is the collateral chain. Ellison personally and irrevocably guaranteed forty point four billion dollars to back his son’s bid for Warner Bros Discovery. He does not hold that in cash. He holds Oracle paper, and about thirty percent of his stake was already pledged as collateral before the bid. On July 20 a federal judge issued a temporary restraining order blocking the close, and twelve state attorneys general are suing. The guarantee is signed, the assets are undelivered, and the collateral reprices every week the courts deliberate. Oracle fell thirty four percent in July even as it landed a seven billion dollar Pentagon contract. When credit sours, good news stops counting.

Mike Brock walked the whole structure here: https://lnkd.in/gikuKhet He discloses that he is short Oracle, so weigh him accordingly, but the sourcing holds.

None of this is a market opinion. It is a control failure. One customer at half the order book is a concentration finding in every framework I have ever tested against, and at a regulated institution it would have been written up as one. What kept it invisible is that everyone in the loop was booking growth off the same dollar, and nobody independent was ever asked to sign.

If your risk function would flag this in a vendor, why does it pass in your index fund?