Oracle’s push to become a major player in artificial intelligence has required an extraordinary level of investment. According to a research note by the investment bank TD Cowen, the company faces an estimated $156 billion (₹13 lakh crore) in capital expenditure commitments tied largely to data centers designed to support AI workloads. These facilities, essential for training and running large language models, are central to Oracle’s strategy of competing with hyperscalers such as Microsoft, Amazon and Google.
But the scale of that bet has begun to strain Oracle’s balance sheet. TD Cowen warned that both equity and debt investors have raised concerns about whether the company can realistically finance such a buildout. As a result, Oracle is considering drastic measures: cutting between 20,000 and 30,000 jobs and potentially selling parts of its business, including Cerner, the healthcare software firm it acquired for $28.3 billion in 2022. Analysts estimate that workforce reductions alone could free up $8 billion to $10 billion in cash flow.
Certified Cyber Crime Investigator Course Launched by Centre for Police Technology
Oracle declined to comment on the report, but the deliberations reflect the growing cost of competing in an AI arms race increasingly defined by access to capital as much as by technology.
Banks Pull Back, Borrowing Costs Surge
The immediate catalyst for Oracle’s reassessment has been a retreat by U.S. banks from financing data-center projects linked to the company. TD Cowen said lenders have sharply increased the interest rate premiums they demand for Oracle-related projects, pushing borrowing costs to levels more commonly associated with non-investment-grade borrowers.
Those higher costs have had practical consequences. Several data-center leases that Oracle had been negotiating with private operators stalled when financing could not be secured, preventing construction from moving forward. Without those facilities, Oracle cannot guarantee the capacity its cloud customers expect, creating a bottleneck in its infrastructure rollout.
Oracle has already leaned heavily on debt markets, raising roughly $58 billion in just two months for projects in Texas, Wisconsin and New Mexico. But that funding represents only a fraction of what the company ultimately needs. While some Asian banks have stepped in, willing to lend at premium rates to gain exposure to AI infrastructure growth, their participation does little to resolve constraints in the United States, where most of Oracle’s near-term expansion is planned.
Workforce Cuts and Asset Sales as Pressure Valves
Faced with rising financing costs, Oracle is exploring ways to reduce the amount of capital it must deploy itself. TD Cowen said the company has begun requiring new customers to pay as much as 40 percent of contract value upfront, effectively shifting part of the infrastructure burden onto clients. It is also considering “bring your own chip” arrangements, under which customers would supply their own hardware rather than relying on Oracle to provide it.
Both approaches come with risks. Upfront deposits could deter potential customers, while BYOC models may require renegotiating contracts that assume Oracle supplies the full stack. Large-scale layoffs could further complicate execution, particularly as Oracle races to deliver complex data-center projects.
Any sale of Cerner would mark a significant strategic shift. Since acquiring the healthcare company, Oracle has already cut thousands of jobs there, especially after difficulties with a major Veterans Affairs contract. Divesting Cerner could signal a narrowing of focus around AI-driven infrastructure rather than a broad retreat, but it would also underscore how financial pressures are reshaping priorities.
Customers and the Question of Shared Risk
The financing squeeze is already rippling outward. TD Cowen reported that OpenAI has shifted near-term capacity needs to Microsoft and Amazon, a notable change from earlier plans that relied heavily on Oracle’s data centers. More broadly, Oracle has slowed its long-term U.S. data-center procurement, and private operators are hesitating to commit while questions about financing persist.
Industry analysts are divided on how alarming the situation should be for enterprise customers. Some see the divergence between U.S. and Asian lenders as an early warning sign that Oracle’s hyperscale ambitions are running up against financial limits. Others point to strong underlying performance, including rapid growth in cloud infrastructure and GPU-related services, as evidence that Oracle remains fundamentally healthy.
Still, there is broad agreement on one point: companies relying on large cloud providers can no longer treat infrastructure as an abstract utility. As AI-driven expansion demands unprecedented capital, customers may need to view cloud contracts as shared bets on financial stability as well as technology, and to hedge those bets through multi-cloud strategies.
For Oracle, the coming months may determine whether it can sustain its AI push at the pace it has promised — or whether financial gravity will force a more measured ascent.
About the author — Suvedita Nath is a science student with a growing interest in cybercrime and digital safety. She writes on online activity, cyber threats, and technology-driven risks. Her work focuses on clarity, accuracy, and public awareness.
