In April 2026, Reuters reported that Anthropic was exploring the possibility of developing its own AI chips to address ongoing chip shortages. Now, the company appears to be moving forward with those plans.
According to a Thursday report from The Information, Anthropic has entered discussions with Samsung about a potential collaboration on a custom chip. However, the company has not yet determined the chip’s specific use case, how it will integrate into server infrastructure, or its performance specifications.
When contacted by TechCrunch, Anthropic emphasized that a diversified hardware stack—including chips from Google, Amazon, and Nvidia—will remain central to its compute strategy. Regarding the rumored Samsung partnership, the company declined to comment further.
Anthropic joins a growing list of AI companies pursuing custom silicon solutions. These efforts aim to create specialized hardware for distinct computational tasks and to reduce reliance on Nvidia, which continues to dominate the chip industry.
The timing may also reflect competitive pressure. Just last week, OpenAI unveiled its own custom inference processor, called “Jalapeño,” developed in collaboration with Broadcom. OpenAI claims the chip offers superior performance-per-watt compared to rival products. Additionally, Amazon and Google have long offered custom tensor processing units (TPUs) as part of their cloud platforms.
Samsung is already a key player in the AI hardware ecosystem. The company manufactures chips for Nvidia, which are essential for training and running AI models, and uses Nvidia’s software in its own chip fabrication processes. The two firms are jointly building an AI chip factory in South Korea. Samsung has also explored partnerships with Google for chip manufacturing.
via TechCrunch AI