Anthropic is delaying the marketing phase of its highly anticipated initial public offering, with sources telling Reuters that the AI company now plans to begin its IPO roadshow in mid-October and complete the listing just before the U.S. mid-term elections in November. The shift from an earlier late-September or October timeline raises a key question for public markets: whether investors are willing to accept a valuation near $2 trillion for a frontier AI company.
The company had previously been expected to make its prospectus public shortly after Labor Day, with the IPO planned for late September or October. Under the revised schedule, the public filing is now likely to arrive later in September. Anthropic already confidentially submitted a draft Form S-1 registration statement to the SEC on June 1 and can proceed once regulatory work concludes, subject to market conditions.
Banks and credit backstop: Ahead of analyst meetings, Anthropic is working to close a $15 billion revolving credit facility, up from an earlier target of about $10 billion and six times the $2.5 billion five-year facility secured last year. Morgan Stanley is leading the loan, with Goldman Sachs, JPMorgan Chase and Citigroup in prominent roles. Barclays, Wells Fargo, Bank of America, Deutsche Bank, Royal Bank of Canada and UBS are also involved. The same core banks are expected to lead the IPO.
Valuation and growth: Anthropic’s most recent private financing round valued it at approximately $965 billion, putting it ahead of OpenAI’s $852 billion valuation. Some backers are reportedly weighing a valuation near $2 trillion, which would make this one of the largest IPOs in history and potentially surpass SpaceX’s $1.77 trillion June IPO. Anthropic could seek to raise as much as $100 billion, which would exceed SpaceX’s $86 billion raise.
The valuation debate rests on rapid growth. Anthropic’s annualized revenue run rate topped $65 billion by the end of July, up roughly sixfold from a year earlier. Its second-quarter revenue more than doubled to $11.6 billion, while OpenAI’s rose 18% to $6.7 billion. In July, 43.5% of U.S. companies purchased Anthropic subscriptions or tokens, compared with 39.7% for OpenAI, according to Ramp’s August AI Index.
Compute commitments and cost pressures: To support demand, Anthropic has signed large compute agreements, including 5 gigawatts of computing power with Amazon, another 5 gigawatts of Tensor Processing Unit capacity with Google and Broadcom, a GPU deal with SpaceX, and cloud deals with Nvidia-backed providers Nscale and Lambda worth $45 billion and $35 billion respectively.
Yet the cost side remains a concern. Anthropic’s Fable 5 model is priced at $10 per million input tokens and $50 per million output tokens, roughly 2.5 times higher than OpenAI’s GPT-5.6 Sol at $4 and $20. The model accounted for 6% of Anthropic tokens purchased by companies and 11.4% of model spending in its first month despite its high price. Analysts including Goldman Sachs’ James Covello have called overall AI spending trends unprecedented and unsustainable, while Gartner predicts cost per agentic workflow will increase more than five times through 2028. Morningstar notes that until Anthropic publishes its S-1 with audited financials, investors have limited visibility into gross margins, cash flow and customer concentration.
The IPO is framed as a major test of whether private-market AI valuations can hold in public markets. If Anthropic achieves a near-$2 trillion valuation, it will face immediate comparisons with SpaceX, which went public in June at $1.77 trillion.
