Claude AI developer Anthropic looks for a way beyond Nvidia

Anthropic is working on its own AI chip and is already in talks with Samsung for that matter. That is according to the US industry publication The Information. Anthropic is the company behind the Claude chatbot. So far, the AI company has relied almost entirely on external hardware. With its own chip, it wants to become more independent, especially from market leader Nvidia. Anthropic has not confirmed the plans.

What Samsung is expected to contribute

Samsung's latest 2nm process is reportedly discussed, along with its advanced packaging technology. This places the main processor closer to the memory chips, accelerating data exchange and reducing bottlenecks. That is precisely what matters for AI accelerators. Samsung recently increased the yield of its 2nm manufacturing process to around 60 percent, making up ground in the process. The company already manufactures Tesla's AI6 driving chip and an inference chip for Nvidia. Anthropic would be another high-profile US customer.

Why Samsung, of all companies

The two companies already know each other. Samsung became a strategic infrastructure partner of Anthropic in May 2026, alongside SK Hynix and Micron, as part of a $65 billion funding round. That close relationship has now led to the possible manufacturing partnership. For Samsung's long-unprofitable foundry division, Anthropic would be another important new customer.

No decision yet

The project is still at an early stage. According to the report, there is no finished design, specific manufacturing steps have not started, and Anthropic could still abandon the plan. However, the company is expanding its staff. In early June, it hired Clive Chan, who previously worked on OpenAI's chip team at an early stage. At the same time, Anthropic told The Information that Amazon's Trainium, Google's TPUs, and Nvidia's GPUs remain central to its own computing capacity. According to the report, Anthropic is also negotiating over chips from Microsoft and the British startup Fractile. In other words, this is about building several pillars, not making a hard break from Nvidia.

Why this matters beyond Anthropic

Nvidia dominates the AI chip market, with The Information estimating its share at around 74 percent. This dependence is expensive and risky, which is why nearly all major AI providers are looking for alternatives. OpenAI is developing its own chip with Broadcom, Google uses its TPUs, and Amazon relies on Trainium. If Anthropic joins them, pressure on Nvidia will continue to grow. For users, the main factor is competition. More competition in AI hardware could lower costs and help the technology advance faster. However, it will likely take years before an Anthropic chip is ready.