OpenAI and Anthropic are lobbying S&P Global and Moody's for investment-grade BBB- ratings after their IPOs to unlock the $11.7 trillion U.S. corporate bond market, testing whether rating agencies can apply one standard to loss-making trillion-dollar AI labs and profitable incumbents like Oracle.
A quiet campaign is under way to hand two of the most valuable private companies in the world a credit grade that has so far eluded them. OpenAI and Anthropic, together running at an annualized revenue rate above $400 billion but still unprofitable, want investment-grade BBB- ratings from S&P Global and Moody's after their IPOs — the key that would open the $11.7 trillion U.S. corporate bond market to insurers and pension funds.
"We currently still view OpenAI and Anthropic as being deep in speculative-grade territory... they are still losing money," one credit analyst said.
Morgan Stanley and Goldman Sachs have held discussions with the rating agencies on behalf of both companies in recent weeks, according to the Financial Times. Anthropic could publicly file its IPO prospectus as early as next week, with a roadshow planned for late September and a potential listing valuation of $2 trillion — surpassing SpaceX's $1.78 trillion valuation at its June listing.
The outcome matters well beyond the two labs. Oracle, Nvidia, Alphabet and Broadcom have extended credit support or built data center capacity on the assumption that OpenAI and Anthropic can keep paying. An investment-grade rating would let the labs sell bonds to insurers and pension funds, easing their reliance on venture capital, bank credit and partner guarantees.
The two companies present a financial profile with no direct precedent for rating agencies. Their private valuations sit near $1 trillion, giving them a thick equity cushion. Anthropic alone disclosed to shareholders that July revenue reached an annualized run rate of $65 billion, up from $47 billion in May. Yet neither lab has posted a profit: compute, research and talent costs consume the revenue, and neither has shown consistent positive free cash flow.
The grade they are chasing is the same BBB- that S&P assigned Oracle this summer after a downgrade. Oracle is a mature business that earned more than $20 billion in operating profit in the fiscal year ended May 2026 at an operating margin above 30 percent. Its problem is balance-sheet leverage: $168 billion in debt tied to data center construction and $260 billion in off-balance-sheet leasing liabilities, with negative free cash flow.
The two camps sit inside the same credit chain. S&P estimates that roughly half of Oracle's $638 billion in remaining performance obligations — customer prepayments for future services — comes from OpenAI. Sustained demand for AI models is what lets the labs generate revenue to pay Oracle and other infrastructure suppliers, which in turn borrowed against that expectation to build capacity. A slowdown in AI demand would hit the labs and their suppliers' credit profiles at the same time.
Rating agencies face a consistency test. Equity investors can price future growth at a premium, but bond investors judge cash flow and default risk. Loosen standards for the labs and agencies understate the risk to bondholders; hold them to traditional debt-servicing metrics and agencies risk being accused of ignoring how new-economy credit works.
The market has already shown it can disagree with a rating. SpaceX bonds issued days after its June listing traded at spreads wider than the investment-grade grades implied, a sign that investors price innovative tech companies' risk independently of agency opinions. Whether OpenAI and Anthropic win the same market acceptance after an upgrade remains open.
For investors, the rating outcome sets the direction of the AI infrastructure trade. If OpenAI and Anthropic secure investment-grade status, independent access to bond markets would ease refinancing pressure on Oracle and reduce the need for Nvidia's $105 billion in credit support for OpenAI's Ohio data center — an arrangement that terminates once OpenAI obtains a satisfactory rating. If agencies hold the labs at speculative grade, the burden stays with the equity and guarantee providers that have financed the buildout so far.
This article is for informational purposes only and does not constitute investment advice.
