Oracle has raised $18 billion in debt, which could help fund massive datacenter investments aimed at meeting surging demand from AI model builders and enterprise customers.
According to Securities and Exchange Commission filings yesterday [PDF], Big Red sold the bonds to raise cash for a number of activities including capital expenditures.
The move follows Oracle’s first quarter results from earlier in the month, when it stunned the stock market by announcing it had bagged cloud contracts, signed but not yet paid for, worth $455 billion.
Later, credit rating agency Moody’s noted that to provide that kind of capacity, Oracle would have to rely on debt. In commentary attached to a research note, Moody’s said it expects the costs to buy equipment and secure real estate and utilities “will be enormous.”
“Whether these will be financed through traditional debt, leases, or highly engineered financing vehicles, the overall growth in balance sheet obligations will also be extremely large,” the analyst said. “We generally treat leases and many structured vehicles as debt-like obligations and include them in debt-related metrics.”
In July, Moody’s changed its outlook on Oracle to negative, although the rating itself stayed the same. The new outlook reflected “the expectation of continuing elevated leverage and increasingly negative free cash flow as Oracle materially ramps up its AI infrastructure business.”
Announcing Oracle’s results, Safra Catz, then CEO, said the erstwhile database and application vendor had lined up cloud deals with OpenAI, xAI, Meta, Nvidia, AMD, “and many others.”
As if to prove the point, reports suggest that OpenAI will pay Oracle $300 billion over five years to fuel the LLM poster child’s ambitions by providing five gigawatts of compute capacity.
Additional deals benefiting Oracle’s cloud ambitions include its arrangement to provide cloud infrastructure and storage to video sharing giant TikTok in the US and its involvement in the US government’s ambitious Stargate AI project. Neither of these projects have revealed the specific pie-slice destined for Big Red.
Where some of Oracle’s customers will get the money needed to service its now huge debt is an open question.
While Meta may have positive cash flow, OpenAI is still reliant on funding. Its exact position remains vague. It might have as much as $60 billion from VCs and other companies, including Microsoft.
Earlier this week, OpenAI and Nvidia signed a letter of intent for the chipmaker to invest up to $100 billion in the GenAI firm, but only if it keeps buying Nvidia chips.
Others have noted a shadow falling over the GenAI funding on which Oracle will rely. Management consultants Bain said the total infrastructure bill to meet industry expectations for AI would be $2 trillion in revenue by 2030.
However, it also said that even if companies shift all of their on-premises IT budget to the cloud, and reinvest any projected savings from AI productivity gains into capital spending on new datacenters, the total amount available would be $800 billion short of what’s needed.
Moody’s said Oracle’s AI infrastructure business has “tremendous potential” but added that the “exponential growth and the already high debt burden at Oracle could result in an extended period of high leverage and negative cash flow.”
As Moody’s noted, Oracle’s assumptions about the financing of its datacenter build program relies “on the AI companies committing to pay for these resources and being able to pay for years to come,” a so-called counterparty risk.
“Given the lack of financial information about the potential counter parties, this risk assessment is subjective at best,” the research commentary said. ®