Anthropic hires Google TPU veteran Amir Salek to accelerate in-house AI chip development
The AMW Read
Anthropic case-study update: named Google TPU lead hire, OpenAI chip alum, and Samsung foundry talks advance custom-silicon staffing; novelty 2, significance 3 for talent+silicon structure.
Anthropic hires Google TPU veteran Amir Salek to accelerate in-house AI chip development
Anthropic is accelerating custom AI silicon after hiring Amir Salek, who led Google's Tensor Processing Unit program from the first through seventh generations, Bloomberg reported on August 21. Salek joins Anthropic's compute organization and will report to James Bradbury, the company's compute lead. Before Google, Salek spent eight years at Nvidia. In June, Anthropic also hired Clive Chan, an early member of OpenAI's custom-chip effort, and is reportedly in talks with Samsung Electronics' foundry business.
The hire lands as Claude's developer races to secure compute for rising model demand without ceding long-term control of its accelerator stack. Salek's TPU pedigree, Chan's OpenAI chip background, and Samsung foundry outreach together show Anthropic treating silicon design as a first-party capability, not a procurement detail. At the same time, the company says Nvidia GPUs, Google TPUs, and Amazon Trainium remain core to its compute strategy—buying capacity now while staffing the rare team that can design chips later.
For builders and investors, the concrete watchpoint is whether Anthropic's custom silicon stays a multi-year hedge or begins to share training and inference load with Trainium and TPU allocations. Talent of this caliber is scarce; each Google-TPU and OpenAI-chip hire tightens the frontier-lab race to reduce single-vendor accelerator dependence.
