Anthropic, the developer of the Claude language model, has revealed in updated IPO prospectus documents that it has signed purchase and lease agreements for computing capacity totaling $895.4 billion for the coming years. Of that amount, $447.5 billion is scheduled to be paid by the end of 2030.
The disclosure arrives as the company prepares for an initial public offering planned for next month, and its financial figures raise questions for investors about the pace of return on those infrastructure investments.
Financial performance and rising costs
In the third quarter, Anthropic’s revenue grew by nearly 50 percent to $17 billion, a slowdown compared with earlier periods this year. By contrast, spending on infrastructure — including chips, power and other data-center costs — rose 77 percent, from $5.6 billion to $9.9 billion.
Maintaining data centers consumed 58 percent of third-quarter revenue, up from 48 percent in the second quarter. The jump in compute costs pushed operating results into the red: after an operating profit of $974 million in Q2, the three months to the end of September showed an operating loss of $567 million. The company’s reported net loss, however, narrowed from $66.2 billion to $23.3 billion.
The prospectus notes that the company entered large, relatively short-term and expensive contracts to meet surging demand, which hurt operating margins. Anthropic also committed to spending more than $3 billion per month with SpaceX alone.
IPO sizing, valuation and market backdrop
Anthropic’s planned listing could become one of the largest stock market debuts ever: existing shareholders expect a company valuation in the $1.5–2.0 trillion range, potentially exceeding SpaceX’s $86 billion June fundraising. The timing comes amid a difficult market: major U.S. technology shares have fallen in recent weeks, with the Nasdaq 100 and the Philadelphia Semiconductor index posting sharp declines, a backdrop that has increased investor scepticism about the sustainability of the AI industry.
Risks and safety considerations
Beyond the financial data, the company expanded the risk factors in the filing. Management warned that expensive, short-term contracts used to satisfy strong demand have reduced margins. The filing also includes a previously absent passage stating that the development pace of its latest models may have reached a level requiring new internal safety measures to prevent catastrophic risks. It adds that a significant acceleration in development speed might be identified only during training or even after the change has occurred.
Why this matters
Anthropic’s enormous computing commitments and the reliance on costly short-term capacity highlight the infrastructure intensity of scaling advanced AI services. The combination of slowing revenue acceleration, sharply rising operating costs and very large contractual obligations will be closely scrutinized by investors as the company attempts a potentially record-setting IPO in a fragile market.
An AI assistant supported the preparation of this article; the final text was edited and verified by a journalist. This article does not constitute investment advice.
