OpenAI partners have racked up $100bn of debt to fund the startup's ambitions - FT
There is virtually no debt on OpenAI's own balance sheet

The aggregate debt accumulated by OpenAI's data center partners is close to $100 billion, according to the Financial Times. OpenAI itself is benefiting from an investment boom financed by borrowing without taking on financial risks, the publication says. According to its sources, a group of banks is discussing a new $38 billion infrastructure loan for the company.
Details
The FT estimates that Japanese conglomerate SoftBank, IT giant Oracle and cloud capacity provider CoreWeave have already borrowed at least $30 billion to invest in OpenAI or build data centers for it. Investment group Blue Owl Capital and data center operators like Crusoe have racked up $28 billion in loans, planning to pay them back from future revenues from contracts with OpenAI. In the coming weeks, Oracle and data center builder Vantage plan to borrow another $38 billion from banks to finance new facilities for OpenAI, the British newspaper's sources said.
These liabilities alone are five times the size of OpenAI's annual revenue - $20 billion in 2025. In fact, the debts associated with OpenAI could be much higher, the FT notes. Many of the startup's partners, including SoftBank and CoreWeave, have borrowed larger sums this year that were not directly tied to OpenAI, the publication says.
The San Francisco-based startup, which has become the world's most valuable non-public company with a valuation of $500 billion, believes it needs even more money to finance the construction of data centers, purchase chips and pay for electricity in an effort to create "general artificial intelligence" (AGI) - a system that surpasses human abilities. FT sources say OpenAI itself has virtually no debt on its balance sheet: in 2024, the company raised a $4 billion line of credit from a group of US banks, but has not yet used it.
Context
Investors express concern about the strategy of Oracle Corporation, whose active expansion into the field of artificial intelligence has led to a growing debt load and customer concentration risks. Against the background of a general cooling of market interest in companies with high costs for AI, Oracle shares showed the strongest decline among technology giants: according to the calculations of The Wall Street Journal, by November 19, quotations fell by 31% from historical peaks. By comparison, Meta's shares lost 25% over the same period, while Nvidia and Microsoft shares slipped 10%.
Analysts record growing concerns about a possible "bubble" in the industry and excessive capital investment, the scale of which alarmed Bank of America respondents for the first time since 2005. According to the forecast of Morgan Stanley, in 2025-2028, the total expenditures of technology companies on data centers may reach $2.9 trillion, but the issuers' own funds will be enough to finance only half of this amount, which will require a significant increase in borrowings.
This article was AI-translated and verified by a human editor
