Anthropic recorded net losses of 80.7 billion dollars (71.869 billion euros) in the first half of the year, which implies multiplying by more than six the previous "red numbers" of 13 billion dollars (11.576 billion euros) recorded by the company responsible for advanced AI models like Claude, which is finalizing preparations for its upcoming IPO.
According to the draft of the S-1 prospectus accessed by "The Wall Street Journal", most of these losses, around 72 billion dollars (64.121 billion euros), are due to accounting adjustments related to changes in the value of convertible bonds and other instruments issued by the company to finance itself.
The preliminary document details that Anthropic generated approximately 16.3 billion dollars (14.516 billion euros) between January and June, well above the nearly 1.1 billion dollars (980 million euros) obtained in the same period of the previous fiscal year.
The company specifies that around 20% of its semiannual revenue was concentrated in just two clients, and Anthropic warns that it expects to continue relying heavily on a small group of large clients in the coming years.
In the first six months of 2026, Anthropic allocated nearly 9 billion dollars (8.015 billion euros) to expand its computing capacity and infrastructure, compared to the 2.3 billion dollars (2.048 billion euros) directed to this chapter in the same period of 2025, and also above the 7.3 billion dollars (6.501 billion euros) invested throughout 2025.
The newspaper notes that Anthropic could end up disbursing up to 750 billion dollars (667.931 billion euros) in the coming years to strengthen its processing capacity, including agreements that are still in the negotiation phase.
In this regard, the tech company maintains payment commitments of about 450 billion dollars (400.758 billion euros) in cloud computing contracts for the next seven to ten years.