Etched raises $300 million Series C at a $10.3 billion valuation
The funding round signals strong investor appetite for AI inference hardware, but the company still faces scaling, manufacturing, and market adoption hurdles before mass production proves it can compete broadly.
At a glance
- Etched closed a $300 million Series C led by Sequoia; other participants include Andreessen Horowitz, SK Hynix, Jane Street, and Diffusion Capital
- Valuation post-money is $10.3 billion, up from $5 billion in December, making this the highest Sequoia-led Series C valuation at the time
- The company has manufactured homegrown chips, tested full systems with clients, and reported about $1 billion of booked orders; headcount is around 400 and a 2 megawatt data center is in operation
- Etched touts two new components to speed up AI inference: a prefill chip that runs at low voltage for higher efficiency and a cluster-scale memory interconnect to enable fast, low-latency multi-chip memory pools
The story
Etched, the AI inference chip startup founded by three Harvard dropouts in 2022, has closed a $300 million Series C funding round at a $10.3 billion valuation, co-founder and COO Robert Wachen told TechCrunch. The round was led by Sequoia, with participation from Andreessen Horowitz, SK Hynix, Jane Street, and Diffusion Capital, among others. Earlier investors also joined the round, underscoring strong support from notable backers.
Etched was valued at $5 billion in December after a $500 million round, meaning the company has more than doubled its valuation in roughly seven months. Wachen says the round represents the highest valuation ever for a Sequoia-led Series C.
The startup has been building and testing homegrown chips and full rack systems with client pilots; Etched says it has already booked about $1 billion of orders and has a team of roughly 400 people operating out of a 2 megawatt data center. Founders Gavin Uberti, Wachen, and CTO Chris Zhu have described a path to mass production despite substantial skepticism since the company’s inception.
Etched is marketing an inference architecture built around two new components designed to speed up the two stages of inference: a low-voltage prefill chip that reduces heat and increases packing density, and a memory/interconnect design called cluster-scale memory that enables many chips to connect to a shared memory pool with very low latency. The systems are pitched as capable of running a range of AI models, including Mixture of Experts like DeepSeek and Qwen, as well as non-transformer designs like Mamba, rather than being limited to a single model family.