The distribution of Oracle's roughly $18 billion syndicated loan for its Project Jupiter data center in New Mexico has stalled, with bank quotes falling to 89-91 cents on the dollar. This pricing puts the loan's market value at approximately $16 billion to $16.4 billion. Despite the broader AI sector's resilience, Oracle's shares dipped 2% overnight, a sign of its recent underperformance relative to the group.
The loan, extended by a banking syndicate late last year, serves to kickstart Project Jupiter's construction. This project falls under the "Stargate" initiative, a $500 billion AI infrastructure endeavor led by OpenAI, SoftBank, and Oracle, with ambitions to build massive AI computing capacity across the United States.
Concerns are mounting over Oracle's persistent borrowing and deteriorating credit profile. In July, S&P downgraded Oracle's corporate credit rating, leaving it just one notch above junk status. Efforts by banks to sell this debt to a wider pool of investors have stalled, forcing them to retain a larger share of Oracle-related exposure on their balance sheets than originally intended, according to the Financial Times.
Cash Flow Strains Emerge for AI Giants
Oracle isn't alone in facing financial pressure. Per the Financial Times, OpenAI revealed in a July internal presentation that it anticipates a cumulative negative free cash flow of $278 billion between 2026 and 2030. Although its revenue is projected to climb from $36 billion in 2026 to $350 billion by 2030, the massive outlays for computing power are expected to far outpace its top-line growth.
Meanwhile, the Wall Street Journal reported on the 18th that Anthropic plans to delay its IPO until November, pushing back from the market's earlier expectation of a record-breaking October offering. Sources indicate that one factor for the postponement is a call from leaders of the biggest AI firms to slow the pace of technological development.
A Trillion-Dollar Funding Appetite Tests Credit Markets
The expansion of AI investment continues, but capital is not an unlimited resource. The five largest U.S. cloud companies have separately guided to combined capital expenditures exceeding $750 billion for 2026, with projections breaking the $1.1 trillion mark in 2027. By 2030, cumulative AI capital spending could reach a staggering $5.5 trillion.
JPMorgan estimates that investment-grade corporate credit markets will provide over $2.1 trillion in financing for data centers over the next five years, with high-yield bonds and leveraged loans contributing an additional $350 billion. The discount and stalled distribution of Oracle's $18 billion loan arrive at the very outset of this financing wave, underscoring the early fault lines forming in the credit market's ability to sustain the AI buildout.