He founded Google's TPU program and led the development and production of TPUs from the first to the seventh generation. He also worked at Nvidia and Cerberus Capital Management.

What is the specific target for the custom chips?

The target is to halve the operational cost of inference, which is the step where the AI generates answers.

Is Anthropic replacing its current hardware providers?

Not immediately. The company still uses Nvidia GPUs, Google TPUs, and Amazon Trainium, and the custom chips will be added as part of a multi-chip strategy.

AI Business & IndustryTop Companies' AI MovesLarge Language ModelsTop Companies AI6d agoAug 26, 2026, 06:32 JST2 min read

Anthropic hires Google TPU founder to build own AI chips

Send on LINE

Key takeaway

Anthropic hired the founding lead of Google's TPU chips. He will design custom AI chips to cut inference costs.
The company aims to halve operating costs for running Claude.
This is part of a broader push toward vertical integration in AI hardware.

3 Key Points

What happened
Anthropic has hired Amir Salek, the founder of Google's TPU program, to lead the design of its own AI chips. He will report to James Bradbury, head of the computing infrastructure team.
Why it matters
The goal is to cut inference costs in half by developing custom application-specific integrated circuits optimized for Claude, rather than relying solely on general-purpose chips. This is part of a multi-chip strategy, not an immediate replacement of existing Nvidia, Google, or Amazon hardware.
What to watch
Anthropic has already hired Clive Chang from OpenAI in June 2026 and is in early talks with Samsung to manufacture chips using its 2-nanometer process and advanced packaging technology.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Get news like this every morning — free

Context & Analysis

Anthropic's move to bring in a veteran chip designer signals a deepening commitment to owning its hardware stack. The company is not abandoning its current suppliers but is seeking to optimize its infrastructure for its specific AI model, Claude. This approach could offer cost advantages in the competitive AI market, where inference costs are a significant operational burden.

The hiring is part of a broader trend among major AI companies to vertically integrate, from model development to chip design. OpenAI, for instance, is working with Broadcom on inference-focused chips. Anthropic's early talks with Samsung for advanced manufacturing could give it access to cutting-edge process technology and packaging, potentially improving performance and energy efficiency.

The timing is notable, as the company has already poached talent from OpenAI and has been rapidly strengthening its hardware division this year. This suggests a strategic push to secure compute capacity and reduce costs, which are critical for scaling AI services sustainably.

FAQ

Who is Amir Salek?: He founded Google's TPU program and led the development and production of TPUs from the first to the seventh generation. He also worked at Nvidia and Cerberus Capital Management.
What is the specific target for the custom chips?: The target is to halve the operational cost of inference, which is the step where the AI generates answers.
Is Anthropic replacing its current hardware providers?: Not immediately. The company still uses Nvidia GPUs, Google TPUs, and Amazon Trainium, and the custom chips will be added as part of a multi-chip strategy.

Get the latest AI Business & Industry news every morning

For example, today's edition would include:

DataAgent launches with $10M to auto-fix Kubernetes faultsSiliconANGLE AI · 1h ago
SK Hynix custom HBM boosts inference up to 5.15xDIGITIMES Asia · 1h ago
Taoyuan pitches northern AI data center hubDIGITIMES Asia · 1h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Subscribe free →

Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →