Anthropic recruits an OpenAI chip expert ahead of its IPO

Clive Chan, who has been a key player in the development of custom chips at OpenAI, has recently joined Anthropic. This change comes as both companies prepare to go public, and as Anthropic considers designing its own chips. Chan, who describes himself as the second hardware employee in OpenAI's chip program, expressed pride in having contributed to a team he considers unmatched in the field. He expects that the chips developed at OpenAI will become "one of the most important engines of AGI."

At OpenAI, Chan worked on creating custom chips and played a role in the partnership with Broadcom, although the latter faced challenges due to production costs and solvency issues. His role at Anthropic remains unclear, but his job description on LinkedIn, "perplexity by picojoule," suggests that he may either design new chips or optimize software for existing hardware.

According to Reuters, Anthropic plans to develop its own AI chips, following a trend initiated by OpenAI and Meta. Although these plans are still in the early stages, Chan could be instrumental in establishing a dedicated team for this project. Currently, Anthropic uses Claude on Google TPUs and Amazon chips, and has recently signed a long-term agreement with Google and Broadcom, planning a $50 billion investment in U.S. computing infrastructure.

Before joining OpenAI in January 2024, Chan spent about two and a half years in Tesla's Autopilot division, according to his LinkedIn. At Tesla, he worked on a custom chip for machine learning training, handling the setup of the software framework, co-designing data centers, and energy-efficient digital formats. Chan's expertise could enable Anthropic to create custom chips, providing a significant financial advantage, particularly for inference.