Anthropic’s chip team is using Claude, a report says

Claude is helping a 15-person Anthropic team design the company’s own AI chips, according to Chubby’s 3 November summary of a report by The Information. The summary says the team found the work faster than its estimates for conventional design workflows. It gives no measured speedup.

There is a concrete public trail beyond that report. Anthropic’s Silicon Engineer vacancy describes building a custom-silicon team. Its example projects include using Claude to review RTL—the code describing digital hardware—check specification consistency and investigate gaps in verification coverage. The page does not establish a 15-person headcount or the reported speedup.

The interesting possibility is a shorter path from a hardware specification to a design engineers can validate. That could make Claude useful in developing the infrastructure that runs models. It would still leave verification, fabrication and bringing the chip into service as separate engineering work; faster drafting is not evidence of a working processor.

We inspected the public post and hiring page. The Information’s article body is behind a subscription; only its title, byline and date were accessible to us. The team-size and acceleration claims above therefore remain explicitly attributed to Chubby’s public summary. No chip specification, delivery date or independently measured result was established in this check.