IA en un minutoNewsBusiness and products
Anthropic lost $80.7 billion in the first half, according to a draft of its IPO prospectus
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
Anthropic lost a net $80.7 billion from January to June 2026, according to a draft of the prospectus for its stock market debut seen by The Wall Street Journal. Most of it comes from accounting adjustments tied to its financing, not from money spent. In those six months it brought in $16.3 billion, and the listing is planned for November.
Anthropic, the company behind Claude, posted a net loss of $80.7 billion between January and June 2026. The Wall Street Journal reported it on Tuesday, October 20, after seeing a draft of the prospectus the company is preparing for its stock market listing, planned for November.
According to that document, most of the loss is not cash that has left the company but accounting adjustments linked to its financing deals. Business Insider, citing the US newspaper, specifies that $72 billion would correspond to that adjustment, which arises from converting into shares bonds and other debt instruments that had funded the company. A year earlier, in the first half of 2025, the loss was $13 billion.
On revenue, Anthropic billed $16.3 billion in the first six months of the year. Around 20% comes from just two customers, a dependence the draft lists among its risks, according to Expansión. Between April and June it brought in $11.6 billion, and in the third quarter $16.8 billion, according to Business Insider.
Compute is the other side of the books. The draft says Anthropic could spend up to $750 billion on computing capacity over the coming years, a figure that includes deals not yet closed. In the first half it spent close to $9 billion on compute, more than the $7.3 billion of all of 2025, and it has signed infrastructure contracts worth $450 billion through 2036: $110 billion with Amazon Web Services, $111 billion with Google Cloud and $31 billion with Microsoft, all three of them shareholders in the company, according to Business Insider.
The draft also lists other risks: not yet knowing how to stop its AI agents from harming third parties or what that would cost, possible cyberattacks and, according to Business Insider, a serious incident involving its systems causing damage beyond what insurance covers. Expansión notes that the listing could value the company at $2 trillion. The figures come from a draft the newspaper had access to, not from a document published by the company.
Why it matters · analysis and opinion
The figure leading the headlines needs careful reading: it is mostly an accounting entry that does not leave the cash register. What is revealing is the relationship between what Anthropic brings in and what it commits to spend on compute. It sketches a market in which being at the frontier of AI requires long-term infrastructure contracts with the same giants that are its shareholders, and in which losing one or two large customers hurts a lot. For anyone using Claude in their company, the final prospectus will be the best window into the soundness of their provider. Before tying critical processes to a single model, it is worth reading what it says about customer dependence, spending and risks.
Source: Expansión · Written with the help of AI: how we make the news
