Etched

Dropped out of Harvard University, where he was concurrently pursuing a B.S. in mathematics and an M.S. in computer science. Previously worked at OctoML and Xnor.ai, gaining experience in ML compiler and inference optimization. In 2022 he co-founded Etched on the thesis that the AI market would consolidate around transformer models, enabling an ASIC optimized exclusively for transformer inference — the Sohu chip — to dramatically outperform general-purpose GPUs. Etched raised $120 million in a 2024 Series A and manufacturers Sohu on TSMC's 4nm process. Named to MIT Technology Review's Innovators Under 35.

A Harvard dropout who co-founded Etched in 2022 alongside Gavin Uberti, serving as CTO responsible for the hardware architecture and chip design of Sohu, the company's transformer-only ASIC. Zhu and Uberti's core bet — that hardcoding the transformer architecture into silicon rather than supporting general workloads yields orders-of-magnitude improvements in inference throughput and power efficiency — underpins the Sohu chip manufactured on TSMC's 4nm process. Etched raised $120 million in Series A funding in 2024.

Co-founded Etched in 2022 alongside Gavin Uberti and Chris Zhu, serving as COO responsible for operations, go-to-market, and company scaling. Wachen's operational leadership supported the company's $120 million Series A fundraise in 2024 and the production ramp of Sohu, a transformer-only ASIC manufactured on TSMC's 4nm process. Etched's value proposition centers on delivering superior inferencing performance per watt compared to GPU-based systems for transformer model workloads.