Oracle’s AI Bet: Debt and Data Centers

Tech giant Oracle (ORCL 10.83%) has racked up remaining performance obligations, which represent future revenue expected to be generated from signed contracts, totaling $523 billion. the company added $68 billion to the backlog in the second quarter of fiscal 2026 alone, signing new deals for AI infrastructure with Meta Platforms, Nvidia, and other customers.

Oracle’s deals to provide AI infrastructure require the company to build expensive data centers filled with powerful GPUs before any revenue is recognized. Oracle entered the AI boom with a debt-heavy balance sheet, and the situation is worsening as the company borrows to fuel its AI ambitions.

Burning cash and piling up debt

For the six-month period ending Nov. 30, Oracle generated $10.2 billion in operating cash flow and used $20.5 billion for capital expenditures. That works out to a negative free cash flow of $10.3 billion. Capital spending more than tripled year-over-year during the period as Oracle plowed ahead with its capital-intensive AI strategy.

Oracle has spent years piling up debt to fund share buybacks, which raises the risk level of adding even more debt to the balance sheet. Thanks to heavy spending on AI infrastructure, Oracle ended the second quarter with about $108 billion in debt. That’s up from $92.6 billion in May. The company completed an $18 billion bond sale in September.

Oracle stock was tumbling on Thursday following the release of its second-quarter report. If the company can successfully convert its AI infrastructure backlog into revenue over the next few years, its revenue growth is expected to accelerate dramatically. However, the upfront cost is enormous. Additionally, there’s a risk related to OpenAI, which reportedly has a $300 billion AI infrastructure contract.