Anthropic is considering designing its own AI chips to address a shortage necessary for advanced AI systems, according to three sources familiar with the matter. The plans are in early stages, and the company may ultimately decide to continue purchasing chips rather than pursuing design.
The company has not committed to a specific chip design or assembled a dedicated team for the project, as stated by two sources. A spokesperson declined to comment on the reported plans.
Demand for Anthropic’s AI model Claude surged in 2026, with the company’s run-rate revenue exceeding $30 billion, an increase from approximately $9 billion at the end of 2025. This revenue growth highlights the increasing significance of the company within the AI market.
To run its software and the Claude chatbot, Anthropic currently utilizes various chips, including tensor processing units (TPUs) from Google’s Alphabet and chips from Amazon. Recently, Anthropic signed a long-term deal with Google and Broadcom, the latter aiding in TPUs design, as part of a more extensive commitment to invest $50 billion in U.S. computing infrastructure.
Other large technology firms, including Meta and OpenAI, are also exploring efforts to design their own AI chips. The cost to design an advanced AI chip is estimated at around half a billion dollars, as this process requires skilled engineers and precision to ensure defect-free manufacturing.
