Goldman Sachs and Morgan Stanley have been talking to Fitch, Moody’s and S&P in recent weeks about investment-grade ratings for OpenAI and Anthropic after those companies list, the Financial Times reported, in an account republished by Folha de S.Paulo.
The banks already sit at the top of both IPO syndicates. Folha, citing people familiar with the matter, said the conversations are aimed at opening the U.S. corporate-bond market — a pool it put at $11.7 trillion — once the shares are public. No rating has been assigned. Anthropic, OpenAI, Goldman Sachs, Morgan Stanley, Moody’s and S&P declined to comment; Fitch did not respond.
Unnamed credit analysts still put both labs in speculative-grade territory. They remain unprofitable and have given little sign of generating positive free cash flow. Bankers’ pitch, according to a senior analyst quoted by the FT, is that the listings will leave the companies “flush with liquidity,” shrinking the credit risk of the debt they will need for data centers and chips. Rating analysts said they want to see the IPO results first.
The rating fight is also about who else is on the hook. Documents cited by Folha show Nvidia has provided $105 billion of credit support for a large OpenAI data center in Ohio, and that the support ends when OpenAI obtains a “satisfactory credit rating.” Nvidia’s August 2026 filing describes that support as residual-value guarantees on leases at SB Energy’s PORTS-Pike campus in Pike County, Ohio, not a $105 billion cash check. Folha also reported large related exposures at Oracle, Google and Broadcom, and said Oracle raised funds for a $300 billion OpenAI-related data-center plan after a recent downgrade that left its own investment-grade standing at risk.
Goldman Sachs and Morgan Stanley were already the two top bookrunners for both planned IPOs as of June 2026, though the lead-left role had not been assigned, according to a Fortune report carried by Yahoo Finance. That report said both companies had filed confidential draft registration statements. Folha said Anthropic’s prospectus was expected soon and that OpenAI was expected to follow with an IPO the next year. Neither company has publicly set a listing date.
If the agencies did grant investment-grade status at listing, OpenAI and Anthropic could borrow more cheaply from a much larger set of bond buyers — pension funds and insurers among them — while still losing money and still funding huge buildouts. The same upgrade could unwind partner backstops, including Nvidia’s Ohio support, and ease pressure on Oracle and chip suppliers that have already extended credit. The agencies have not agreed. The story is Wall Street trying to rewrite the usual credit playbook for two loss-making AI labs, not a rating that already exists.