Anthropic confirmed on August 5 that it is building an in-house silicon team to design custom AI chips for Claude, the first time the company has publicly acknowledged the effort. Business Insider broke the story; Anthropic confirmed to TechCrunch and Reuters that it will co-design hardware and models to make Claude run faster and more efficiently at scale.
The economics and the timeline
Designing an advanced AI chip costs roughly $500 million, according to industry sources cited by Reuters, covering engineering, testing, and fabrication. Anthropic has not disclosed a timeline or whether it will manufacture the chips itself, though The Information reported last month that the company is in early talks with Samsung as a potential manufacturing partner. Job listings for the custom silicon team offer salaries up to $485,000 and specify experience in chip design, verification, and AI-assisted formal validation.
Anthropic emphasized it will continue its multi-chip approach, using hardware from AWS, Google, Nvidia, and AMD alongside any future custom silicon. The company recently signed a multi-billion-dollar deal with Google for access to 1 million TPUs and separately uses 500,000 of Amazon's Trainium2 chips.
Why vertical integration matters now
Anthropic joins OpenAI, which unveiled its Broadcom-built Jalapeño chip in June, and Meta and Google, which have run custom accelerators for years. The move gives Anthropic control over performance optimization, cost structure, and supply — three variables currently dictated by Nvidia and cloud providers. Co-designing chips with models removes inefficiencies from off-the-shelf hardware mismatches, potentially improving inference speed and lowering per-token costs. The chip effort is a bet that owning the stack — model, harness, and silicon — will matter more than renting the fastest commodity GPU.