As Oracle makes significant investments in AI infrastructure, credit markets have first responded with more cautious pricing. According to Reuters, citing the Financial Times, loans totaling approximately $18 billion related to Oracle’s large-scale Project Jupiter data center initiative in New Mexico are currently trading at 89 to 91 cents on the dollar in the secondary market.
The discount reflects project delays.
This does not mean Oracle is in default. More directly, institutional investors are demanding a lower price to take on this debt, and banks, which originally expected to more easily distribute the project loan, are now less willing to take it on.
Large project loans are typically arranged by banks and then distributed to institutional investors. Reports mention that Santander and Jefferies are among the lenders for Project Jupiter, but due to weakening market demand, some debt related to Oracle remains on the banks' books.
7-month extension lowers bids
Loan prices have dropped to around 89 cents, indicating that the market is beginning to price in execution risk for the project. Reports show that Project Jupiter is at least seven months behind schedule, with permitting approvals, power infrastructure development, and local opposition all slowing progress.
Additionally, an application related to the 2.2 GW gas turbine supply was also denied. For the data center project, construction delays mean financing costs continue to accrue while revenue has not yet been generated, directly reducing project returns.
AI infrastructure financing is being reassessed
From a business perspective, Oracle has not shown any significant slowdown. Reports indicate that the company’s remaining performance obligations are approximately $664 billion, including over $30 billion in new AI cloud contracts. Meanwhile, Oracle’s cloud infrastructure revenue recently increased by 121% year-over-year.
However, the market is more concerned about how much additional capital the company needs to invest before these orders are converted into actual cash flow. In July of this year, S&P downgraded Oracle’s rating from BBB to BBB-, placing it at the lowest tier of investment grade.
The key takeaway from this report is not that Oracle's AI strategy has failed, but that lenders and bond investors are beginning to more clearly assess the associated risks. If this trend continues, financing costs for AI data center projects that rely heavily on asset-intensive expansion may face upward pressure.