Anthropic Moves Into Custom Silicon to Scale Claude
AI-generated image · Bay Street Wire
The AI startup is assembling a chip design team to optimize hardware-to-model efficiency as demand surges.
Anthropic is assembling a dedicated team to design its own custom AI chips, as TechCrunch reports, citing reporting from Business Insider and confirmation provided to TechCrunch. The company is currently recruiting engineers for a "custom silicon team" to co-design hardware and models, a move intended to increase the speed and efficiency of its technology.
While Anthropic has secured hardware access through agreements with Nvidia, AMD, Google, and AWS, the company is seeking a way to scale further to meet rising demand for Claude. The Information previously reported that Anthropic has been exploring a potential partnership with Samsung to manufacture these chips.
This shift follows a broader industry trend of AI developers seeking independence from third-party hardware. TechCrunch notes that Meta has developed its MTIA accelerators, and Google DeepMind utilizes Alphabet's TPU chips. Additionally, OpenAI recently debuted the Jalapeño chip, built in partnership with Broadcom, specifically for inference workloads.