Anthropic has hired former Google TPU executive Amir Salek to join its compute team.
Salek helped lead Google’s Tensor Processing Unit programme and delivered its first seven generations of chips.
The Claude maker is expanding its chip and data centre strategy as AI companies race to secure computing capacity.
Anthropic Eyes Its Own AI Chips: Why It Hired a Google Chip Veteran
The Claude maker has hired former Google TPU executive Amir Salek as it expands its compute team and explores building custom chips to support its growing AI infrastructure needs
Summary
Advertisement
Anthropic has hired Amir Salek, a former Google executive who helped build the company’s custom chip programme, as the Claude maker prepares for a possible push into its own semiconductor business, according to a Bloomberg report. Salek will join Anthropic’s compute team and report to James Bradbury.
Salek previously ran Google’s Tensor Processing Unit (TPU) business until 2022 and was involved in delivering the first seven generations of the company’s AI chips. After leaving Google, he became senior managing director at private equity firm Cerberus Capital Management. He has also previously worked at Nvidia.
Anthropic Steps Up Custom Chip Plans
Anthropic currently relies on chips from several suppliers, including Nvidia, Google and Amazon. However, the AI company has recently started hiring for roles linked to an in-house silicon effort, signalling a push to develop technology tailored to its own computing needs.
Custom chips could help Anthropic deal with shortages of AI processors and give the company greater control over the hardware used to run its AI models. The move also comes as leading AI companies compete for data centre capacity and computing power.
Advertisement
Anthropic has been expanding its infrastructure partnerships at a rapid pace. The company has signed an agreement with UK-based chip startup Fractile for an initial order of about $250 million worth of chips, with plans to potentially expand the deal.
AI Chip Race Heats Up
Anthropic is also securing data centre capacity through deals with companies including Riot Platforms and Volta Infra Holdings as it scales its AI operations.
The push towards custom chips mirrors moves by other major AI companies. OpenAI has unveiled Jalapeno, a chip developed with Broadcom, and plans to start using it later this year.
Anthropic’s hiring of Salek adds an experienced chip executive to its compute team at a time when AI companies are looking beyond simply buying processors from established suppliers. His experience leading Google’s TPU programme could support Anthropic’s efforts to develop hardware suited to its growing AI computing requirements.
Advertisement