Etched Raises $300M at $10.3B Valuation, Claims Highest-Ever Sequoia-Led Series C
Etched, an AI chip startup founded in 2022 by three Harvard dropouts, closed a $300 million Series C in July 2026 at a $10.3 billion valuation. The round was led by Sequoia, with Andreessen Horowitz, SK Hynix, Jane Street, and Diffusion Capital also participating. The company was valued at $5 billion in December 2025, meaning it doubled in roughly seven months.
What the Company Actually Does
Etched designs chips optimized for AI inference and manufactures them through TSMC. The architecture splits inference into two phases handled differently at the hardware level.
The prefill chip runs at lower voltage than competing AI chips. Lower voltage means more transistors fit in the same power envelope, which Etched says enables faster inference. For the decode phase, the company built a cluster-scale memory interconnect that lets multiple chips share a single memory pool at low latency.
The systems support Mixture of Experts models including DeepSeek and Qwen. They also support non-transformer architectures like Mamba, a state-space model. That last part is worth noting: most AI chip startups have bet entirely on transformers continuing to dominate.
Where Things Stand Operationally
As of June 2026, Etched had booked $1 billion in orders and was shipping first full systems to clients for testing. The company operates a 2 megawatt data center at its San Jose office and recently opened an 80,000 square-foot, 10 megawatt facility in Milpitas, California. It has 400 employees.
Andrej Karpathy, Noam Brown from OpenAI, and Geoffrey Hinton have tested the hardware. Karpathy, Thiel, Dylan Field, and Amjad Masad are among individual backers. The company claims this is the highest valuation ever for a Sequoia-led Series C, which is a specific enough claim that someone will eventually fact-check it.
The Competitive Context
Google is reportedly pursuing a concept called Frozen v2, which involves etching parts of a specific AI model directly into silicon for use with Gemini. The basic idea is similar to Etched's approach: commit to specific model characteristics at manufacturing time to gain performance advantages over general-purpose accelerators.
This could mean the design philosophy Etched is betting on is gaining credibility. It could also mean big players are converging on similar ideas from different directions, which usually gets complicated for startups. The fact that $1 billion in orders are already on the books suggests customers aren't waiting to find out.
The $10.3 billion valuation reflects expectations, not proven at-scale deployment. Client testing started in June. Real-world inference performance across diverse workloads remains to be seen.
Source: Techcrunch