Anthropic confidential IPO filing prepared in June shows revenue rising from $400 million in 2024 to $4.6 billion last year alongside an $8 billion operating loss. Sales accelerated further to $11.5 billion in the second quarter of this year. The figures matter because the company plans to go public in November at a valuation of $2 trillion or more.
Revenue growth and terms of the planned listing
Details from the confidential S-1 were published late Monday by Reuters and the Financial Times. The filing outlines a public offering expected to raise as much as $100 billion. Nvidia, which invested $10 billion in Anthropic last November, could join the IPO as an anchor investor. The document positions the listing as one of the largest technology offerings on record.
The loss structure explains how fast scaling consumes cash. Anthropic logged a $42 billion net loss last year, more than five times the 2024 level. That total includes a $34 billion accounting charge tied to funding raised from investors. Stripped of that item, the operating loss reached $8 billion, with computing infrastructure accounting for more than 91% of the sum.
The filing also points to a near-term improvement in operating performance. The Financial Times reported that Anthropic is on track to close the current quarter with an adjusted operating profit. The company prepared the S-1 in June, so the second-quarter sales figure and the current-quarter profit target come from subsequent reporting on the document. Maintaining that posture looks difficult given the scale of fixed infrastructure obligations.
What heavy cloud commitments mean for customers
Anthropic plans to spend $518 billion on computing infrastructure over the next decade. About 80% of that amount sits in contracts that cannot be canceled or require payment even if allocated capacity goes unused. Nearly half of the cloud budget will go to Amazon Web Services, Microsoft and Google. Another large block consists of about $161.2 billion in equipment leases with Broadcom.
The Broadcom relationship connects directly to AI accelerator supply. In April, the chipmaker partnered with Google to provide several gigawatts of TPU capacity to Anthropic. TPUs are custom AI accelerators developed by Google in collaboration with Broadcom. For enterprise buyers, such locked-in capacity means model availability depends less on spot cloud supply and more on how Anthropic allocates reserved compute across products and regions.
The prospectus devotes no fewer than 80 pages to risk factors, including a lengthy discussion of AI safety. It warns that advanced models could pose catastrophic or existential risks, display self-preserving behaviors, conceal or manipulate information, and engage in behavior resembling blackmail. Product development could further increase the risk of harm. Companies selecting vendors should therefore ask about safety testing, deployment controls, incident disclosure, and how contract terms handle model behavior changes. The marker to watch is the November listing process: confirmation of timing, valuation near $2 trillion, and offer size near $100 billion will show whether investors accept this combination of growth, fixed costs, and stated model risk.
