Anthropic is building an in-house team to design its own AI chips. The company confirmed the effort on August 5 after Business Insider first reported it, per TechRepublic, and told TechCrunch it would co-design hardware and models together so Claude runs faster and more efficiently.
The group is called the custom silicon team. Its job posting seeks engineers across "different domains of chip design and verification" and asks for evidence of "direct personal contribution" to semiconductor designs that actually shipped. The listed salary range is $320,000 to $485,000.
What Anthropic did not say
Anthropic did not disclose two things: it has not said when a first custom chip could be ready, and it has not said whether it intends to manufacture processors itself. The Information reported on July 2, 2026 that Anthropic was in talks with Samsung Electronics about manufacturing a custom AI chip, which Anthropic has not publicly confirmed.
The multi-chip strategy stays in place
Anthropic described the team as part of a "multi-chip" strategy that continues to run on hardware from Amazon Web Services, Google, Nvidia and AMD. Its outside compute commitments are still growing. Forbes reported that in April Anthropic expanded its partnership with Google and Broadcom for roughly 3.5 gigawatts of next-generation TPU capacity coming online in 2027, on top of a gigawatt arriving in 2026 under a Google Cloud agreement signed last October. Broadcom develops and supplies the custom TPUs, and separately supplies networking and other components inside Google's next-generation AI racks under a commitment running through 2031. Forbes put the company's revenue run-rate at about $30 billion, up from roughly $9 billion at the end of 2025, with more than 1,000 business customers spending at least $1 million a year.
The chip team was confirmed on the same day Anthropic signed a $10 billion, six-year compute deal with Volta for a Norway data center. In July it committed to 2 gigawatts of AMD Instinct MI400 capacity.
Where it sits among the frontier labs
Anthropic's peers already run silicon programs. OpenAI unveiled its Broadcom-built Jalapeño inference chip in June 2026. Google DeepMind runs on Alphabet's TPUs. Meta develops its own MTIA accelerators. On August 3, Nvidia launched its Open Secure AI Alliance without Anthropic, OpenAI or Google.
The silicon program is one of the long-dated commitments Anthropic carries into an offering it expects to match or exceed SpaceX's record $75 billion raise.
Source: TechCrunch
Comments