Synthetic intelligence lab Anthropic is exploring the potential of designing its personal chips, three sources stated, as the corporate and its rivals reply to a scarcity of AI chips wanted to energy and develop extra superior AI methods.
Samuel Boivin | Nurphoto | Getty Pictures
Synthetic intelligence lab Anthropic is exploring the potential of designing its personal chips, three sources stated, as the corporate and its rivals reply to a scarcity of AI chips wanted to energy and develop extra superior AI methods.
The plans are in early levels and the corporate should still resolve to solely purchase AI chips and never design any, based on two individuals with information of the matter and one individual briefed on Anthropic’s plans. The corporate has but to decide to a selected design or put collectively a devoted workforce to work on the mission, one of many sources stated.
A spokesperson for the San Francisco-based firm declined to touch upon the article.
Demand for its AI mannequin Claude has accelerated in 2026, with the startup’s run-rate income now surpassing $30 billion, up from about $9 billion on the finish of 2025, Anthropic stated earlier this week.
Anthropic makes use of a variety of chips, together with tensor processing items (TPUs) designed by Alphabet’s Google and Amazon’s chips to develop and run its AI software program and chatbot Claude.
Earlier this week, Anthropic signed a long-term cope with Google and Broadcom which helps design the TPUs. That deal builds on the corporate’s dedication to speculate $50 billion in strengthening U.S. computing infrastructure.
Anthropic’s discussions mirror comparable efforts underway at giant tech firms which can be searching for to design their personal AI chips, together with Meta and OpenAI.
Designing a complicated AI chip can price roughly half a billion {dollars}, based on trade sources, as firms must make use of expert engineers and spend to verify the manufacturing course of has no defects.