Anthropic Launches Custom Silicon Team Amid Rising Claude Demand

Anthropic has launched a dedicated “custom silicon team” to design proprietary chips for artificial intelligence applications, according to a publicly posted job listing. The move follows reports that the company is evaluating Samsung as a potential manufacturing partner for those chips.

The Claude developer stated its objective is to co-design hardware and large language models in tandem, aiming to improve both speed and energy efficiency of its AI systems. This strategic shift coincides with surging usage of Claude and intensified competition among AI firms for access to high-performance computing infrastructure.

Infrastructure partnerships and scaling limits

Anthropic has secured agreements with Amazon Web Services (AWS), Google, Nvidia, and AMD to obtain AI computing hardware. Yet the company has concluded that dependence on third-party suppliers alone cannot sustain the scale required by current demand levels.

Broader industry trend toward in-house silicon

Anthropic joins several major AI developers pursuing custom chip development. OpenAI unveiled its Jalapeño chip—built by Broadcom—in June; the chip targets inference workloads specifically. Google DeepMind uses Alphabet’s Tensor Processing Units (TPUs) exclusively for AI model deployment. Meta has been advancing its own MTIA accelerators for AI tasks.

The company is actively recruiting engineers with expertise in chip design. Business Insider and The Information reported on the initiative, with the latter noting Anthropic’s outreach to Samsung last month. Anthropic did not respond to a request for comment.