Oracle's $18B AI project loan trades at 89–91 cents as delays weigh on demand

Oracle's big push into AI infrastructure is getting a more cautious reception from credit markets. Reuters, citing the Financial Times, reported that about $18 billion of loans tied to Oracle's large-scale Project Jupiter data center development in New Mexico is changing hands in the secondary market at roughly 89 to 91 cents on the dollar. The markdown reflects delays in the project’s execution, not a default by Oracle. Investors are effectively demanding a larger discount to hold the debt, while banks that had expected to syndicate the loan more smoothly are finding demand softer than anticipated. Large project financings are typically underwritten by banks and then distributed to institutional investors. Santander and Jefferies were cited among Project Jupiter's lenders, and some of the exposure is said to remain on banks' balance sheets as appetite has weakened. A reported seven-month slippage has weighed on pricing, with bids falling toward 89 cents as the market begins to factor in execution risk. The project is said to be at least seven months behind schedule, slowed by permitting, power infrastructure buildout, and local opposition. An application connected to a 2.2 GW gas turbine supply was also denied. Construction delays can erode economics for data center developments: financing costs continue to accrue before revenue starts, pressuring expected returns. Operationally, Oracle's underlying business momentum appears intact. Reports put the company's remaining performance obligations at about $664 billion, including more than $30 billion in new AI cloud contracts, and its cloud infrastructure revenue was recently reported up 121% year over year. Credit investors, though, are focused on how much additional capital will be required before contracted demand translates into cash flow. In July, S&P lowered Oracle's rating from BBB+ to BBB, leaving it at the bottom rung of investment grade. The broader signal is not that Oracle's AI strategy has faltered, but that lenders and bond buyers are sharpening their pricing of the risks. If that reassessment persists, funding costs could rise for AI data center projects that depend on heavy, asset-intensive buildouts.