Anthropic is forming an in-house team to design chips for AI. TechCrunch reported on Aug. 5 that Anthropic is seeking experienced chip design engineers to work on a “custom silicon team,” according to a job posting.
Anthropic plans to design hardware and models together so its technology runs faster and more efficiently. It is seen as moving into in-house chip design as demand for Claude rises and competition among AI companies to secure infrastructure intensifies.
Anthropic currently has contracts with AWS, Google, Nvidia and AMD to secure AI computing hardware. It appears to have concluded that relying only on external supply is not enough to scale up in line with rising demand.
Last month, there was also a report that Anthropic was considering Samsung Electronics as a potential partner to make chips.
Anthropic is not the first AI company to move into in-house chip development. OpenAI unveiled an inference chip developed with Broadcom in June, and Google DeepMind uses Alphabet's TPU chips. Meta is also developing the MTIA accelerator for AI workloads.
Chi-gyu Hwang (황치규)
delight@d-today.co.kr