Financing linked to Oracle’s massive Project Jupiter data center in New Mexico is coming under pressure as concerns grow over the company’s debt, the project’s development and opposition to its potential environmental impact.
Oracle’s aggressive push into AI infrastructure is facing another financial complication.
Around $18 billion in loans linked to a massive Oracle-leased data center project in New Mexico are reportedly trading below their original value, according to the Financial Times.
The loans have recently been quoted at roughly 89 to 91 cents on the dollar by banks involved in the financing, including Santander and Jefferies.
The development comes as investors take a closer look at Oracle’s rapidly expanding debt burden and the risks surrounding the infrastructure projects being built to support the growing demand for artificial intelligence.
A huge AI data center in New Mexico
The project, known as Project Jupiter, covers approximately 1,400 acres in Doña Ana County, New Mexico.
The campus is part of Oracle’s broader agreement with OpenAI to provide computing infrastructure for AI workloads.
The scale of the project reflects the enormous amount of computing capacity that companies now require to train and operate increasingly powerful AI models.
Oracle secured approximately $18 billion in financing from a group of banks late last year to help fund construction, according to previous media reports.
The loans were subsequently expected to be distributed more broadly among investors.
That process has reportedly become more difficult.
Banks are holding more of the debt
According to the Financial Times, efforts to sell portions of the Project Jupiter debt to a wider group of investors have stalled.
One factor is growing concern about Oracle’s financial position.
Oracle has been spending heavily to expand its cloud and AI infrastructure business, while its debt levels have increased.
The company’s credit profile has also attracted greater scrutiny following a downgrade by S&P in July that left Oracle’s rating only one level above junk status.
As a result, banks involved in the Project Jupiter financing may have to keep a larger amount of the debt on their own balance sheets than originally expected.
That creates an additional risk for lenders at a time when investors are becoming increasingly selective about companies taking on large amounts of debt to finance AI infrastructure.
Project Jupiter also faces local opposition
The financial concerns are not the only challenge facing the project.
Project Jupiter has encountered opposition in New Mexico from people concerned about the potential impact of the data center on water supplies and air quality.
Large AI data centers require substantial amounts of electricity and can also require significant water resources for cooling, depending on the technology used.
Those demands have increasingly become a source of controversy as communities around the United States debate where new AI infrastructure should be built and how much pressure it could place on local resources.
For Project Jupiter, those concerns could complicate the development of an already enormous facility.
Natural gas infrastructure adds another complication
The project’s original energy plans have also encountered obstacles.
Project Jupiter was initially expected to receive power from approximately 2.2 gigawatts of gas turbines.
However, the New Mexico State Land Office blocked a request connected to construction of a natural gas pipeline intended to serve the data center, according to the Financial Times report.
The dispute highlights another challenge facing large-scale AI infrastructure: securing enough reliable electricity to operate facilities that can consume enormous amounts of power.
For companies such as Oracle and their cloud and AI customers, access to computing capacity is only part of the equation.
They also need land, electricity, cooling systems, network connections and financing capable of supporting projects that can cost billions of dollars.
Why Oracle’s debt matters to the AI boom
Oracle has become one of the major infrastructure providers supporting the expansion of generative AI.
The company has been investing heavily in data centers and cloud capacity to meet demand from AI developers, including large customers such as OpenAI.
That strategy creates an unusual financial dynamic.
AI infrastructure can require enormous upfront investment, while the revenue generated from those facilities develops over time.
Companies therefore need access to large amounts of capital before the infrastructure begins generating its full economic return.
Oracle’s rising debt has consequently become an important part of the conversation around its AI expansion.
The question for investors is not simply whether demand for AI computing will continue to grow. It is also whether the financial returns from that demand will justify the enormous cost of building the infrastructure needed to support it.
A broader test for AI infrastructure financing
Project Jupiter illustrates a challenge that extends well beyond Oracle.
The AI boom has triggered a wave of investment in data centers, power generation and semiconductor infrastructure. Much of that expansion depends on debt financing, long-term customer agreements and assumptions about continued demand for computing capacity.
If financing becomes more expensive or lenders become less willing to take on large project exposures, some of the industry’s most ambitious infrastructure plans could face additional pressure.
At the same time, local opposition over electricity, water, emissions and land use is becoming an increasingly important factor in determining where new data centers can actually be built.
For Oracle, Project Jupiter sits directly at the intersection of those trends.
The company is betting heavily on the long-term demand for AI computing. But the New Mexico project shows that building the infrastructure behind that boom involves more than securing customers and ordering servers.
It also requires billions of dollars in financing, reliable energy, regulatory approvals and local acceptance.
