Anthropic is looking into the development of its own AI chips.
Anthropic is in the early stages of exploring the design of its own AI chips, as reported by Reuters, citing three sources familiar with the situation. The company has not yet committed to any specific designs or formed a dedicated team for this initiative. It may still opt to continue purchasing chips from external suppliers rather than developing its own.
This exploration comes shortly after Anthropic entered into a long-term agreement with Google and Broadcom for 3.5 gigawatts of TPU compute capacity starting in 2027. A company spokesperson declined to comment on the matter.
Anthropic's revenue has surged significantly, with the company revealing earlier this week that its annualized revenue run rate has exceeded $30 billion, up from approximately $9 billion at the conclusion of 2025. This growth has generated a demand for compute resources that makes the idea of custom silicon worth considering. Currently, Anthropic operates Claude using a combination of chips, including tensor processing units from Alphabet’s Google in collaboration with Broadcom, along with Amazon's custom chips and Nvidia hardware. The company stated that it allocates workloads to the chips best suited for them.
In addition to the Reuters report, days before, Anthropic signed a long-term agreement with Google and Broadcom that provides roughly 3.5 gigawatts of TPU-based compute capacity from 2027, which is about three times its earlier consumption of one gigawatt in 2026, according to a filing by Broadcom with the SEC. The filing noted that the expanded deployment depends on Anthropic's sustained commercial success, an atypical caveat in a regulatory document. This agreement builds on Anthropic’s pledge made in November 2025 to invest $50 billion in U.S. computing infrastructure.
Broadcom is currently a chip design partner for OpenAI and has an undisclosed fifth customer for its XPU, positioning itself at the forefront of the emerging custom AI silicon market, which is becoming a substitute for Nvidia’s general-purpose GPUs.
Anthropic's potential development of proprietary silicon aligns with trends observed elsewhere in the industry, with Meta constructing its own AI training chips and OpenAI also pursuing custom silicon. Industry sources quoted by Reuters estimate the cost of developing an advanced AI chip to be around $500 million, a figure that reflects the necessity of hiring specialized engineers and validating the manufacturing process. While this cost is considerable for a company that is currently unprofitable, it becomes more feasible given a revenue run rate that has more than tripled in four months.
Published April 10, 2026 - 7:50 am UTC
Other articles
Anthropic is looking into the development of its own AI chips.
Anthropic is looking into the development of custom AI chips. This decision follows Claude’s run-rate revenue exceeding $30 billion, a significant increase from $9 billion at the end of 2025.
