OpenAI And Anthropic Seek Cheap Debt Before IPOs

OpenAI and Anthropic are reportedly trying to do something that shows just how expensive the AI race has become. They want investment-grade credit ratings before their expected public listings. The Financial Times reported that bankers for the two AI labs are pushing rating agencies to treat them like future blue-chip borrowers, even though both companies are still loss-making and burning through large amounts of cash.

The reason is simple. AI now requires staggering amounts of money before the revenue fully catches up. Training frontier models, buying chips, leasing data centres, paying cloud partners and hiring technical talent all require capital on a scale that looks more like heavy industry than ordinary software.

An investment-grade rating would matter because it could make borrowing cheaper. It would also open the door to a much wider pool of institutional debt investors. For companies that may need to finance tens or hundreds of billions of dollars in compute over the next decade, even a small difference in borrowing costs becomes meaningful.

The FT says Morgan Stanley and Goldman Sachs have been lobbying rating agencies, arguing that future IPO proceeds could strengthen the balance sheets of OpenAI and Anthropic. That is a bold argument because it asks debt investors to believe not only in current growth, but in future liquidity, future margins and future public-market appetite.

This links directly to the circular AI finance question. We recently wrote about OpenAI’s SB Energy deal and why AI infrastructure is beginning to look like a web of companies financing one another, buying from one another and using future growth expectations to justify present spending. Debt ratings would push that structure into an even more formal market.

The rating question also affects partners. Nvidia, Oracle, Google, Broadcom and other infrastructure players benefit when frontier labs can keep spending. Broadcom’s latest AI chip surge and Nvidia’s Hugging Face deal show that the ecosystem around these labs is already enormous.

But the risk is obvious. Traditional investment-grade borrowers usually have predictable cash flows, profits and mature markets. Frontier AI companies have huge demand, but they also face uncertain margins, intense competition, regulatory pressure, copyright lawsuits, safety costs and a technology cycle that changes quickly.

A top-tier credit rating would be a powerful vote of confidence. It would also be a test of whether financial markets are ready to treat AI labs not as speculative startups, but as the next generation of critical infrastructure companies. That may be where the industry is heading, but it is not a small leap.