IA en un minutoNewsBusiness and products
Anthropic's Internal Claude Use Outweighs Its Staff Spending, Its IPO Filing Says
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
Anthropic says in its stock market filing that the value of its own Claude use, calculated at list prices, was multiplied by more than 250 from January to September 2026. Anthropic sees that monthly value above its personnel costs since April, leaving out what is paid in stock. Business Insider points out it is not cash paid out, but consumption valued as if a customer were paying for it.
In the filing for its stock market debut, the Form S-1 it submitted to the SEC on October 26, Anthropic says its internal use of Claude, valued at the list prices it charges customers, now weighs more than what it spends on staff each month. According to the document, that value has grown more than 250 times between January and September 2026, and in April it exceeded monthly personnel spending for the first time, not counting stock-based compensation.
The filing itself and Business Insider, which highlighted it on November 1, make the caveat clear: it is not cash Anthropic pays out, but the value that this Claude consumption would have at list prices. Business Insider adds that more than 90% of the code added to the company's software in September was written by Claude, and that revenue has grown much faster than headcount.
The company explains in the S-1 that each version of Claude is used to help build the next one, and that it believes the compute it devotes to internal use multiplies those gains and speeds up its pace of development. The same document acknowledges that using AI to build more capable AI could also pose a considerable safety risk if it sharply accelerates progress, and places it among the matters covered by its Responsible Scaling Policy.
More AI news, instantly, at iaenunminuto.com
Why it matters · analysis and opinion
The figure is a way of measuring rather than a real outlay, and that is where its interest lies: Anthropic is starting to compare model consumption with the personnel line, which indicates it already treats it as a first-order work resource and not as a side tool. That it says so in a filing aimed at investors suggests it wants it read as a signal of productivity. It is worth reading with caution for two reasons: the value depends on list prices set by the company itself, and the document admits that using AI to create more capable AI can carry safety risks. For any organization, the practical question is how much of its repetitive work already consumes models, and how that consumption is measured next to the cost of people.
Official source: Anthropic (Form S-1 filed with the SEC) · Written with the help of AI: how we make the news
