Anthropic is reportedly discussing a possible partnership with Samsung to develop and manufacture a custom artificial intelligence chip.
The talks remain at an early stage. Anthropic has not yet decided what the processor would be used for, how it would fit into its server systems, or how much computing power it would provide.
The reported talks follow an April report that Anthropic was considering designing its own AI processors in response to shortages and rising demand for computing infrastructure.
Developing a custom chip could give Anthropic more control over the hardware used to train and operate its Claude models. However, the company has not committed to producing the processor.
When asked about the potential Samsung partnership, Anthropic said it had nothing further to share.
Anthropic said a diverse hardware system will remain central to its computing strategy.
The company currently trains and operates Claude using Google’s Tensor Processing Units, Amazon’s Trainium processors and Nvidia GPUs. This allows it to assign different workloads to the hardware best suited to each task.
A custom processor would therefore expand Anthropic’s existing hardware options rather than immediately replace chips supplied by Google, Amazon or Nvidia.
Several major AI companies are developing specialized processors to improve efficiency and reduce their reliance on Nvidia, which remains the main supplier of advanced AI accelerators.
Custom chips can be designed around specific workloads, potentially reducing energy consumption and operating costs compared with general-purpose AI hardware.
Anthropic’s reported discussions come shortly after competitor OpenAI unveiled Jalapeño, an inference processor developed with Broadcom. OpenAI said the chip offers improved performance per watt and was designed specifically for large language model inference.
Google and Amazon already operate their own AI processors through their cloud platforms. Google offers TPUs, while Amazon provides Trainium chips.
Samsung already plays a major role in the AI hardware industry by supplying advanced memory and semiconductor technologies.
The company has a close partnership with Nvidia and is building an AI-powered semiconductor factory in South Korea using Nvidia’s accelerated computing and manufacturing software. Samsung also supplies memory products intended for Nvidia’s AI platforms.
Samsung has also reportedly discussed possible work on future Google AI chips as demand strains manufacturing capacity at other foundries.
Neither Anthropic nor Samsung has confirmed an agreement. The processor’s design, purpose, specifications and production plans therefore remain undecided.
Get the latest tech news, telecom insights, and product launches wherever you prefer.
Add ProPakistani to Preferred Sources and see more of our stories in Google Search and Top Stories.