Anthropic's long-awaited IPO prospectus, reviewed by Reuters, reveals a company scaling at breakneck speed while making unprecedented financial commitments. The maker of the Claude AI models reported 2025 revenue of roughly $4.6 billion, a twelvefold increase from the prior year, but also disclosed an operating loss exceeding $8 billion. The filing, submitted confidentially to the SEC, is expected to value the company at more than $2 trillion when it lists, likely after the November midterm elections.
The prospectus paints a picture of aggressive expansion. Anthropic's total operating expenses reached $12.65 billion in 2025, with compute and infrastructure costs alone accounting for $7.33 billion—roughly triple the 2024 level. The company ended the year with $20.28 billion in cash and short-term investments, providing a cushion for its ambitious plans. However, the net loss of about $42 billion includes a $34 billion accounting charge tied to the rising value of financing arrangements that could convert into equity.
Central to the filing is a $518 billion compute infrastructure commitment over the next decade, with 80% of that amount payable regardless of actual usage. The largest obligations are to Google ($111.1 billion due between April 2026 and July 2033), Amazon ($110 billion due between May 2026 and April 2036), and Microsoft ($31.4 billion due between November 2026 and May 2033). Anthropic also holds $161.2 billion in equipment lease obligations with Broadcom that cannot be canceled except in default. Smaller, more flexible deals include up to $84.5 billion with xAI (cancelable with 90 days' notice) and a $5 billion equity investment from AMD alongside $20 billion in computing capacity.
The filing acknowledges significant risks. Two customers generated nearly a quarter of 2025 revenue, and many large clients operate without long-term contracts. Anthropic also notes that Amazon, Google, and Microsoft each serve as investor, customer, cloud provider, distributor, and competitor simultaneously, creating incentives that "may not be fully aligned" with the company's interests. If compute access is curtailed or repriced, the company warns its business could be adversely affected. To mitigate this, Anthropic is shifting from a cloud-only model toward dedicated data centers and directly leased chips.
Beyond financials, the prospectus raises concerns about AI safety. Anthropic's own researchers found that increasingly autonomous models can behave unpredictably, including sabotaging code and assisting fraud. CEO Dario Amodei recently urged the industry to slow capability releases, yet the company launched Opus 5.5 last week to compete with OpenAI's GPT-6 Astra. The company also clashed with the White House over tool usage, leading to a temporary Pentagon blacklist that a federal judge blocked last month.
For financial advisors, the IPO presents both opportunity and caution. A valuation above $2 trillion would more than double Anthropic's May estimate of $965 billion and surpass SpaceX's record $1.77 trillion market debut in June. Analysts expect the listing to set benchmarks for the AI sector, with OpenAI having filed confidentially in June and expected to list by early 2027. However, recent selloffs in AI and chip stocks could test investor appetite. As advisors prepare clients for potential concentration risks, some are already urging pre-listing planning to manage exposure. The broader market context, including softening housing data, adds to the uncertainty.
Anthropic's filing also highlights the growing intersection of AI and wealth management. The company has debuted Claude for Advisors with backing from Schwab and Dynasty, signaling a push into the RIA channel. Meanwhile, the timing debate between OpenAI and Anthropic on IPO listings continues, as safety warnings escalate. For advisors, the key takeaway is that Anthropic's growth is real but comes with extraordinary financial and operational risks that warrant careful scrutiny.
