AI Chip Startup Etched Hits $10.3B Valuation With $300M Series C Led by Sequoia

AI chip startup Etched announced on July 23 that it has closed a $300 million Series C funding round at a $10.3 billion valuation. The round was led by Sequoia, with participation from Andreessen Horowitz, SK Hynix, Jane Street, and Diffusion Capital. Etched claims this is the highest valuation ever for a Sequoia-led Series C, and it nearly doubles the company's $5 billion valuation from December 2025 in just seven months.

Founded in 2022 by three Harvard dropouts — CEO Gavin Uberti, COO Robert Wachen, and CTO Chris Zhu — Etched now employs 400 people, has booked over $1 billion in orders, and operates its own 2-megawatt data center. Early backers include Peter Thiel, Anthropic researcher Andrej Karpathy, Figma CEO Dylan Field, and Replit CEO Amjad Masad.

Rather than modifying existing GPU architectures, Etched designed two entirely new components from scratch to accelerate AI inference — the compute phase that happens after a user submits a prompt. The first is a "prefill" chip that runs at dramatically lower voltage than any other AI chip, generating less heat and enabling denser transistor packing to handle the math-intensive prompt understanding phase. The second is a cluster-scale memory and interconnect system that lets many chips connect to a shared memory pool with extremely fast, low-latency access for the token-generation "decode" phase.

Etched sells its solution as full rack systems rather than individual chips. Its first silicon batch has been successfully manufactured by TSMC. Notable AI figures including Andrej Karpathy, OpenAI's Noam Brown, and Geoffrey Hinton have tested the hardware. The company claims its chips work with any AI model — from transformer-based models like ChatGPT and Claude, to mixture-of-experts architectures like DeepSeek and Qwen, to state-space models like Mamba.

The massive funding round signals that the AI inference hardware market is becoming a new battleground in semiconductors. As AI shifts from training to mass deployment, inference efficiency and cost have become decisive competitive factors. Etched's bet on fully custom silicon represents a high-risk, high-reward alternative to GPU-centric approaches. Whether the company can deliver on its low-voltage inference claims at scale and compete with NVIDIA's ecosystem in real data center environments will be the key metric to watch going forward.