Back in April, Reuters reported that Anthropic was exploring the possibility of developing its own AI chips to address ongoing supply constraints in the semiconductor market. It now appears that the company is moving more seriously in that direction.

 

According to a report published Thursday by The Information, Anthropic has been in discussions with Samsung about a potential collaboration on a custom AI chip. However, key details remain undecided, including the chip's intended use case, its role within server infrastructure, and its overall performance capabilities.

 

In a statement to TechCrunch, Anthropic emphasized that its computing strategy will continue to rely on a diverse hardware ecosystem that includes chips from Google, Amazon, and Nvidia. The company declined to provide additional details regarding a possible partnership with Samsung.

 

Anthropic is not alone in pursuing custom silicon. Several AI companies have begun developing their own chips to optimize performance for specialized workloads while reducing their dependence on Nvidia, which remains the dominant player in the AI chip market.

 

Anthropic's efforts also come shortly after rival OpenAI announced a collaboration with Broadcom to develop its own custom inference chip, known as "Jalapeño." OpenAI claims the processor delivers improved performance-per-watt efficiency compared with competing solutions. Meanwhile, both Amazon and Google already offer proprietary AI accelerators as part of their cloud platforms.

 

Samsung has become an increasingly important player in the AI ecosystem, serving as a key manufacturing partner for Nvidia by producing chips used to train and run AI models. At the same time, Samsung leverages Nvidia's software technologies in its own chip production processes. The two companies are also collaborating on an AI semiconductor facility in South Korea, while Samsung has held discussions with Google regarding additional chip development partnerships.