Companies that supply OpenAI with data centers, chips and “computing” processing power have taken on about $96 billion in debt to finance their operations, according to an analysis by the Financial Times. The news highlights the AI sector’s increasing dependence on debt, particularly on loss-making AI startup OpenAI.
Currently, AI companies and many of the data center operators that are rapidly expanding to serve them are not generating enough revenue to cover their construction costs.
OpenAI has committed $1.4 trillion to procure the energy and computing power needed to run its future operations. However, the company previously said it expected revenue to remain at $20 billion this year. And a recent analysis by HSBC concluded that even if the company had earned more than $200 billion by 2030, it would need to find an additional $207 billion to continue operating.
The FT has revealed the following breakdown of the debt assumed by OpenAI’s partners:
- Softbank, Oracle, and Coreweave have already borrowed $30 billion.
- Blue Owl Capital and Crusoe received $28 billion in financing.
- $38 billion is being considered in further talks with Oracle, Vantage and their banks.
- Total debt is $96 billion.
The increased use of debt to fund AI is a relatively new development. Before this year, most AI builds were funded with cash directly from the balance sheets of big tech companies like Microsoft, Alphabet, Amazon, and Meta.
How CoreWeave repays its debt will be of particular interest to investors. In its third quarter earnings report, the company reported current liabilities of $3.7 billion, non-current liabilities of $10.3 billion and future data center lease commitments of $39.1 billion. The company said it expects sales to remain at $5 billion this year, but has $56 billion in “revenue backlog” going forward.
All companies were contacted for comment. CoreWeave declined to comment when contacted next. luck.
Separately, the big five hyperscalers — Amazon, Google, Meta, Microsoft and Oracle — took on $121 billion in new debt this year to fund their AI operations, according to Bank of America. This is more than four times the average level of debt issued by these companies over the past five years ($28 billion).
According to a recent research note by BofA analysts Yuri Seliger and Seohyun Marie Lee, this excess of investment grade (IG) corporate bonds is having a significant impact on credit markets.
“This week (the week before Thanksgiving) is typically the last week of the year with high IG supply, and 2025 supply is ending the year with a bang. We’re tracking about $50 billion this week and about $220 billion over the past four weeks, which is about 70% more than typical volumes for this time of year,” they said.
“This year…hyperscalers added an additional $63 billion, suggesting that this year’s overall supply increase is explained by: [debt-funded M&A deals] and hyperscaler activities. ”
Deutsche Bank says the increased supply of debt from tech companies has widened the credit default swap (CDS) market for “spreads” – the additional interest yield above the notional risk-free rate demanded by bond buyers. CDS act as a type of insurance against corporate debt, paying out holders if creditors default. An increase in CDS yields indicates that the market believes that the probability of default has increased.
“This move is noteworthy. Oracle’s 5-year CDS has expanded by about +60bps since late September to 104bps, and CoreWeave has expanded by about +280bps since September to about 640bps,” Deutsche’s Jim Reid said in a recent note.
“Whether this change will have a meaningful long-term impact remains to be seen, but recent weeks clearly signal a new phase of the AI boom. Investors are increasingly looking to hedge risk, and public credit markets are being asked to fund growing capital investment needs. It’s no longer just free cash flow for hyperscalers,” he said.
