Oracle’s $300 billion megadeal with OpenAI is pushing Wall Street to its limits, straining lender appetite for the massive debt fueling America’s data center boom. Banks, including JPMorgan Chase, have spent months trying to distribute the risk tied to billions of dollars in loans made for data centers built in Texas and Wisconsin and leased to Oracle — and they are running out of room to take on more.
The core problem is straightforward: many lenders that would typically buy these kinds of loans have caps on how much exposure they can keep to a single client. Once those limits are hit, bank balance sheets get clogged, and new projects become harder to finance. That congestion is now rippling across the broader market for AI infrastructure debt.
Why Oracle Faces Tougher Terms
Oracle does not enjoy the same financial standing as its biggest rivals in the cloud computing space. Unlike Google, Microsoft, and Meta — which carry the cleanest credit profiles on Wall Street — Oracle holds a lower investment-grade credit rating, carries more debt, and continues to burn cash. Those factors make lenders more cautious about stacking up additional Oracle-linked exposure.
The cost of insuring Oracle’s bonds against default roughly quadrupled between late September and late March before easing slightly. Over the same period, Oracle shares fell by more than 30%. To ease some of the pressure, Oracle said it plans to raise roughly $50 billion in stock and bonds during 2026 to cover its near-term cash needs. Even so, analysts at Morgan Stanley estimated that additional funding of $100 billion or more could still be needed for 2027 and into 2028.
Morgan Stanley credit analysts have previously written that Oracle’s considerable funding needs over the next three years may test the depths of different fixed-income markets. Big tech overall is expected to self-fund only about half of the projected $3 trillion in AI investment through 2028, leaving the rest dependent on outside financing.
Michigan Campus Wins Big Financing
Despite the broader lending strain, one of the largest Oracle-linked projects crossed a major milestone this week. Related Digital secured financing for a $16 billion data center campus in Michigan, being built for Oracle. The financing package includes equity from Related Digital and funds affiliated with Blackstone, along with fixed-rate, long-term debt anchored by PIMCO-managed funds and accounts. PIMCO reportedly purchased about $10 billion of the bonds, with other investors taking the remainder.
The Michigan campus is part of a larger push announced jointly by Oracle, OpenAI, and Related Digital. The project — a more than one gigawatt facility in Saline Township, Michigan — was announced in October, with construction beginning in February. Oracle said in a post on X that each data center it is developing for OpenAI is progressing on schedule.
“We are proud of the rapid progress that’s been made both in financing and developing our data centres,” an Oracle spokesperson said. “Our partners have diversified their capital sources in ways that enable us to keep construction moving forward on time and as planned.”
Applied Digital’s $7.5 Billion Lease Deal
Oracle-related expansion is taking shape beyond Michigan as well. Applied Digital signed a 15-year lease worth $7.5 billion with an unnamed U.S.-based hyperscaler at its Delta Forge 1 campus, covering 300 megawatts of computing capacity at a 430-megawatt site in the southern United States. The facility spans more than 500 acres, is designed for large AI and high-performance computing workloads, and is expected to begin initial operations by mid-2027. The new agreement lifted Applied Digital’s total contracted lease revenue to more than $23 billion.
Inference Centers Move Closer to Cities
A separate shift is also emerging in how and where AI infrastructure is being built. While large AI training projects in remote locations have dominated investment so far, the next phase of demand is expected to move closer to population centers. Companies like DataBank are building facilities designed to deliver AI model outputs — known as inference — to users more quickly by positioning data centers near where people actually live and work. That trend points to a broadening of AI infrastructure investment beyond the giant campuses that have defined the buildout so far.
Investor Confidence Holds in ORCL Stock
Even as financing concerns mount, institutional investors have not abandoned Oracle stock. Broad Peak Investment Advisers Pte Ltd reported that Oracle accounted for approximately 12.1% of its total holdings in its most recent filing with the Securities and Exchange Commission, making it the firm’s third-largest position with a stake valued at $67.634 million. That level of conviction from institutional investors suggests that, despite the turbulence in the debt markets, confidence in Oracle’s long-term AI positioning has not entirely faded.
