Anthropic, which is developing its own AI chips, may cooperate with Samsung Electronics
- Input
- 2026-07-03 05:51:21
- Updated
- 2026-07-03 05:51:21

[Financial News] A claim has emerged that Anthropic, a U.S. AI developer, is working with Samsung Electronics to produce its own chips for AI applications. However, the plan is still in its early stages and may never materialize.
The Information reported on the 2nd, citing multiple sources, that Anthropic, known for its AI model Claude, is considering Samsung Electronics as its foundry partner. The company is reportedly discussing the use of Samsung Electronics' 2nm manufacturing process and advanced packaging facilities. At a Series H fundraising event in May, Anthropic said the three major memory chip makers, including Samsung Electronics, had joined as strategic infrastructure partners. It then claimed that "their technologies play a key role in the global supply of memory, storage and logic chips." Logic chips are a type of system semiconductor used for computation and control, including CPUs.
Samsung Electronics is the only one of the world's three largest memory chip makers with a foundry division that produces logic chips. If Samsung Electronics wins the final order for Anthropic's in-house AI chip, it could secure a new major customer after Tesla, NVIDIA and Apple. A source said Anthropic is currently in talks with several chip design firms, but detailed discussions on design, testing or manufacturing have not yet begun.
The report comes as major AI companies rush to produce their own chips. OpenAI, Anthropic's rival, began developing its own AI chips with Broadcom Inc. in 2024 and unveiled last month a reasoning chip called "Jalapeño," optimized for large language model inference. Google, Amazon, Meta Platforms and Microsoft are also developing their own AI chips to reduce dependence on outside suppliers.
In Anthropic's case, the company hired Clive Chan last month, one of the early members of OpenAI's custom chip division, and is exploring chip capabilities, performance levels and ways to integrate them with servers.
In response to the report, Anthropic said that a range of AI chips, including NVIDIA's Graphics Processing Unit (GPU), Google's Tensor Processing Unit (TPU) and Amazon Web Services' Amazon Trainium chips, will continue to play a central role in its computing resources.
Meanwhile, The Information also reported that Google is considering assigning part of the production volume for its next-generation TPU to Samsung Electronics.
pjw@fnnews.com Park Jong-won Reporter