Anthropic in talks with Samsung Electronics over AI chip production

By Seo Ji-yeon, The Herald Business

Anthropic, the developer of the generative AI model Claude, is in early discussions with Samsung Electronics over the production of its own AI chips, raising hopes that a potential deal could give Samsung's struggling foundry business a major boost.

The Information reported Wednesday, citing multiple sources, that Anthropic is in talks with Samsung Electronics as a potential manufacturing partner in the early stages of developing its own AI chip.

According to the report, Anthropic is exploring the use of Samsung Electronics' 2-nanometer process and advanced packaging technology.

The 2-nanometer process is among the most advanced semiconductor manufacturing technologies currently available, offering higher integration density for improved power efficiency and performance.

Advanced packaging is considered a key technology that places computing chips and memory chips in close proximity to accelerate data transfer speeds and reduce bottlenecks.

Anthropic has not yet reached the chip design or trial production stage and is conducting technical reviews alongside several semiconductor design firms, sources said.

To advance those efforts, the company last month recruited Clive Chan, an early member of OpenAI's custom AI chip development team, to evaluate chip functionality, performance and server integration options.

Industry observers say the discussions could be a positive signal for Samsung Electronics' foundry business.

When Anthropic raised its Series H funding round in May, it named Samsung Electronics and other major global memory chip makers as "strategic infrastructure partners," highlighting the importance of supply chains for memory, storage and logic chips.

Samsung Electronics was the only company mentioned at the time with a foundry operation capable of producing logic chips, a detail that has drawn industry attention to the prospect of deeper collaboration.

If a deal is finalized, Samsung Electronics would add another major AI customer to its roster alongside Tesla, Nvidia and Apple.

Major AI companies have recently been stepping up development of their own chips to reduce dependence on Nvidia's graphics processing units and cut costs.

Google continues to advance its proprietary tensor processing units, while OpenAI has unveiled its first inference-focused AI chip developed in partnership with Broadcom, signaling that competition in custom semiconductors is intensifying.

In response to a request for comment, Anthropic said Nvidia GPUs, Google TPUs and Amazon Web Services' Trainium chips "will continue to serve as core computing infrastructure," but declined to comment specifically on plans to develop its own chips.

Samsung Electronics also declined to comment on the report.