The plans are at an early stage and Anthropic may still decide to only buy chips rather than design them. The exploration comes days after the company signed a long-term deal with Google and Broadcom for 3.5 gigawatts of TPU computing starting in 2027. A company spokesperson declined to comment.
anthropic is exploring the possibility of designing its own AI chips, Reuters reported on Thursdayciting three sources familiar with the matter. The effort is in an early stage: the company has not committed to a specific design and has not assembled a team dedicated to the project.
You may still decide to continue purchasing third-party chips instead of building your own. A spokesperson for the San Francisco-based company declined to comment on the report.
The exploration comes at a time when Anthropic’s revenue has accelerated dramatically. The company revealed earlier this week that its annualized revenue run rate surpassed $30 billion, up from about $9 billion at the end of 2025.
That trajectory has created a scale of computing demand that makes the economics of custom silicon increasingly worth examining. Anthropic currently uses Claude on a combination of chips: Tensor processing units designed by Alphabet’s Google, in partnership with Broadcomalong with custom Amazon chips and Nvidia hardware.
The company said it tailors workloads to the chips that best fit them.
Just days before the Reuters report, Anthropic signs long-term agreement with Google and Broadcom That will give it access to about 3.5 gigawatts of TPU-based computing capacity starting in 2027, about three times the roughly gigawatt it consumed in early 2026, according to Broadcom SEC Filing.
The filing noted that the expanded deployment depends on Anthropic’s continued commercial success, unusual coverage for a regulatory filing. The deal builds on Anthropic’s November 2025 commitment to invest $50 billion in U.S. IT infrastructure.
Broadcom is also already a chip design partner for OpenAI and has an undisclosed fifth XPU customer, putting it at the center of the custom AI silicon market that is emerging as an alternative to Nvidia’s general-purpose GPUs.
The possibility of Anthropic developing proprietary silicon mirrors is already underway elsewhere in the industry. Meta has been building its own AI training chips and OpenAI has also been working on custom silicon.
Industry sources cited by Reuters put the development cost of an advanced AI chip at approximately $500 million, reflecting the need to hire specialized engineers and validate the manufacturing process.
That figure is not trivial for a company that, for now, remains unprofitable, but it is more manageable in the face of a revenue base that has more than tripled in four months.
