Anthropic assembles team to design its own AI chips
Anthropic
OpenAI
Google DeepMind
Meta
Anthropic is building a team to design custom chips for AI use, aiming to co-design hardware and models for faster, more efficient performance. The move follows rising demand for Claude, and the company has already partnered with AWS, Google, Nvidia, and AMD for AI compute, but now seeks more control.
According to Business Insider, Anthropic, the maker of Claude, is creating a team to design its own custom AI chips. The company plans to co-design hardware and models to improve speed and efficiency. Last month, The Information reported that Anthropic was considering Samsung as a potential manufacturing partner. This decision comes as demand for Claude rises and AI companies compete for AI infrastructure. Anthropic has signed deals with AWS, Google, Nvidia, and AMD for AI computing hardware, but scaling to meet demand requires more than relying on others. Anthropic follows other AI firms: in June, OpenAI unveiled its Broadcom-built Jalapeño chip for inference, Google DeepMind uses Alphabet's TPUs, and Meta is developing its MTIA accelerators. Anthropic is hiring engineers with chip design experience for its 'custom silicon team,' per a job listing. The company did not immediately respond to a request for comment.
- Abbreviations
- TPU = Tensor Processing Unit — тензорный процессор
Source: TechCrunch AI —
original