AI chip startup Etched has closed a $300M Series C round at a $10B pre-money valuation, pushing its total fundraising past $1.1B. Sequoia led the round with participation from A16Z, Jane Street, Blackstone, SK Hynix, and Diffusion Capital, the company confirmed July 23.
Etched already holds $1B in customer pre-orders, president Robert Wachen told EE Times. The company is targeting the largest AI operators in the world with hardware designed specifically for inference workloads.
“Our customers are buying many billions of dollars of hardware,” Wachen said. “We expect that the product will be very good for coding, long context, long horizon agents, and a lot of things that inference tokens are going to be used for.”
Early customers are already running workloads remotely inside Etched’s San Jose data center with results that Wachen called “significantly better than anything on the market.” The startup recently opened a new R&D facility in Milpitas, California, that includes a 10 MW data center, lab space, and a quick-turn SMT production line.
Founded by Harvard dropouts Gavin Uberti and Chris Zhu alongside Harvard graduate Robert Wachen, Etched originally pitched itself as building chips that hard-coded the transformer architecture into silicon. The technical strategy has since broadened to support more workload types, including multi-trillion parameter mixture-of-experts models, diffusion models, and state-space models.
Etched plans to begin shipping its rack-scale systems this summer. The company operates a lab in San Jose where it demonstrated large MoE inference on live hardware, though it has not yet released public benchmark figures.
