Anthropic has confirmed, for the first time, that it is assembling an in-house team to design custom chips for its Claude models, according to newly posted job listings and a company spokesperson. The move follows months of speculation about Anthropic’s hardware ambitions and puts it alongside Google, Meta, and OpenAI in building proprietary silicon.
A job posting for a Silicon Engineer on Anthropic’s own careers page says the company runs “some of the largest AI training and inference workloads in the world” and is “deepening that investment by building a custom silicon team.” A separate listing for the Technical Program Manager overseeing the effort states that “the hardware our models run on is one of the most direct levers we have on capability, cost, and reliability” — and that Anthropic has decided to take design authority over that hardware itself.
Still a multi-chip strategy
Anthropic told reporters this week that it will keep relying on outside hardware even as it builds its own chips, continuing what it calls a “multi-chip approach” spanning processors from Amazon Web Services, Google, Nvidia, and AMD. The new team’s job is to co-design chips and models together so Claude runs faster and more efficiently at the scale enterprise customers now demand.
The listings describe a lean, senior operation: candidates must have personally contributed to an ASIC design that “taped out and shipped,” and be comfortable managing outside foundries, IP vendors, and packaging partners rather than working inside a large in-house fab. Pay for the roles ranges from $320,000 to $485,000 a year.
The confirmation follows a string of earlier signals: Anthropic was first reported to be exploring custom silicon in April 2026, OpenAI’s own custom-chip program lead left to join Anthropic in June, and reports emerged in July that Anthropic was in early talks with Samsung as a potential manufacturing partner. Rival labs have already moved down this path — OpenAI unveiled its own Broadcom-built inference chip in June, while Google and Meta have run in-house accelerators for years.
Anthropic has not said when a first chip might be ready or who would manufacture it. The push comes weeks after AMD agreed to supply Anthropic with up to 2 gigawatts of AI chips as part of a multibillion-dollar deal, underscoring how tight compute supply has become across the industry even as Anthropic works toward chips of its own.