Anthropic has published the prospectus for its stock market listing, a 614-page document filed on October 26. Analysts, journalists, and commentators have noted details about the company’s revenue, headcount, and governance. The JDN tasked Claude, under editorial supervision, with selecting the most surprising ones.
- Anthropic a déposé son prospectus de 614 pages le 26 octobre.
Revenue, Clients, and Partners
The prospectus indicates that 5 to 7% of revenue for the first nine months of 2026 comes from clients who did not comply with Claude’s usage rules. The company tightened these rules in August and September. For the third quarter alone, the newsletter OnlyCFO estimates this amount at $4 billion, assuming competitors use Claude to train their own models. Anthropic warns that this revenue stream will dry up.
Amazon, Google, and Microsoft provide Anthropic with the computing power it needs and receive in return a commission on Claude sales through their online marketplaces. This commission represents 48% of revenue, compared to 11% in 2023. The S-1 notes that these partners are also developing their own competing models.
The document reports other high ratios. An Anthropic client spends on average nine times more than a year ago, and each employee generates $6.6 million in revenue, compared to $1.4 million a year earlier. However, OnlyCFO tempers these figures: the average masks a very low starting base, and AI has become the top expense line for many clients without their revenue keeping pace.
Governance and Risks
Mark Levine, financial controller for the city of New York, manages municipal employee pensions worth $325 billion, including $7 million in Anthropic shares. In mid-October, he wrote to Dario Amodei requesting a delay of the stock market listing. He cites the prospectus, which mentions “catastrophic or existential risks to humanity.” According to the letter, seen by Reuters, the company should address these risks “rather than rushing headlong into its IPO.” Discussions are ongoing, but no delay has been confirmed.
In his Bloomberg Opinion column, Matt Levine describes Anthropic’s governance as that of a “nice, quirky little worker co-op, guided by its mission.” The seven founders retain majority voting power with a single share, and a trust appoints the majority of the board of directors.
The prospectus also states that employees are “known internally as Ants.” They number 5,300, compared to 940 at the end of 2024. In September, over 90% of the company’s code was written by Claude, a fact the company presents as progress while acknowledging the risk of losing visibility into what happens in its own systems.
Usage and Stock Symbol
According to the prospectus, a professional Claude user who does not use Claude Code, the programming tool, spends less than $75 per month. The top 5% most active users of Claude Code spend twenty times more, or about $1,500 per month.
Asked by Axios about the opportunity to invest, Claude said yes, “at a reasonable price,” adding: “I am Claude, made by Anthropic, weigh my opinion with this conflict of interest in mind.” GPT-6.1, OpenAI’s model, said no, citing spending commitments, the control retained by the founders, and the still-unknown price.
Finally, Anthropic has requested the symbol ANT on the Nasdaq. However, on the Canadian CSE exchange, ANT has designated Anteros Metals since January 2025, a small Newfoundland-based mineral exploration company founded in 2022.
