Anthropic Is Building Its Own AI Chip — and Samsung Could Make It

TLDR

• Anthropic is in early talks with Samsung Electronics to manufacture a custom AI chip using its 2nm process
• The project is still in early stages with no design or manufacturing work started yet
• Anthropic hired Clive Chan, an early member of OpenAI’s chip team, signaling a serious engineering push
• Samsung, SK Hynix, and Micron all participated in Anthropic’s $65 billion May fundraising round
• Anthropic says Nvidia, AWS Trainium, and Google TPUs will remain central to its compute strategy

---

Anthropic, the company behind the Claude AI models, has begun early-stage work on a custom AI chip. The company has held talks with Samsung Electronics as a potential manufacturing partner, according to a report from The Information.

Anthropic is reportedly in early talks with Samsung to manufacture a custom AI chip, per The Information.

The Claude maker is still in the early planning stage, deciding what the processor should do and how it would fit into AI servers or clusters.

Samsung is reportedly being… pic.twitter.com/7yLkZlhrbr

— Wall St Engine (@wallstengine) July 2, 2026

The project is still in its earliest phase. No detailed design or manufacturing work has started, and Anthropic may not move forward with it.

To signal intent, Anthropic hired Clive Chan, an early member of OpenAI’s own chip team. The hire points to a deliberate buildout of in-house chip engineering.

The talks center on Samsung’s 2-nanometer manufacturing process and its advanced packaging facilities. That would put Samsung in direct competition with Taiwan Semiconductor Manufacturing, which currently dominates high-end AI chip production.

Samsung Electronics Co., Ltd., SMSD.L

This is a well-worn path for AI companies. Google built its Tensor Processing Units years ago. Amazon has Trainium and Inferentia. OpenAI partnered with Broadcom to design a chip called Jalapeño, an inference chip revealed last month.

Anthropic raised $65 billion in a Series H round in May, reaching a valuation of $965 billion. Samsung Electronics, SK Hynix, and Micron all took part in that round, making Samsung both an investor and a potential manufacturing partner.

Why Samsung Is in the Running

Samsung is the only chip investor in that round that also operates a foundry business. That means it doesn’t just design or make memory — it manufactures other companies’ chip designs in its own fabrication plants.

Custom chips, known as ASICs, let AI companies tune their hardware to specific workloads. That can improve efficiency compared to buying off-the-shelf processors from Nvidia.

Nvidia still holds an estimated 74% share of the AI chip market. That share has not declined despite the push from AI labs to build their own silicon.

Google is separately considering using Samsung for part of a future Tensor Processing Unit. That would be another win for Samsung’s contract manufacturing business if it comes through.

Earlier this week, Samsung Group and SK Group announced a combined $518 billion, decade-long investment to build four memory chip plants in South Korea.

Anthropic Is Not Replacing Its Current Suppliers

Anthropic was clear that this is not a move away from its existing chip partners. In a statement to The Information, the company said AWS Trainium, Google TPUs, and Nvidia GPUs will remain central to how it scales compute.

The company is also in talks to use chips from Microsoft and UK-based startup Fractile. That points to a multi-vendor strategy rather than a single foundry relationship.

For investors, the key question is whether Samsung can turn these early conversations into a confirmed production contract. Any formal agreement would be a direct challenge to TSMC’s near-monopoly on leading-edge AI chip manufacturing.

Samsung has historically struggled with yields at leading-edge process nodes compared to TSMC, a concern analysts have raised repeatedly.