Etched builds computer chips that do one thing: run AI models fast and cheap. Founded in 2022 by three Harvard dropouts, the San Jose company spent years as an industry punchline for betting its entire existence on specialized silicon while Nvidia's general-purpose GPUs ruled the market. In July 2026 the punchline closed a $300 million Series C at a $10.3 billion valuation, led by Sequoia.
The round, which we covered in our report on Etched's Series C, doubled the company's value in roughly seven months and followed news that booked orders for its inference systems had crossed $1 billion.
A bet that looked reckless in 2022
CEO Gavin Uberti, COO Robert Wachen, and CTO Chris Zhu left Harvard to build silicon specialized for transformer models, the architecture behind systems like ChatGPT and Claude. At the time, wagering a chip company on a single model architecture struck most of the industry as reckless, and custom silicon startups had a long history of ending badly. CNBC chronicled the early skepticism when the company raised its $120 million Series A in 2024.
Four years on, the wager looks prescient. Etched's first chips came back successfully manufactured by TSMC, complete systems entered testing with customers, and demand now outpaces supply as clients move from evaluation to deployment.
What the hardware actually does
Etched sells full rack-scale systems rather than bare chips, all aimed at inference, the computing that happens every time a user submits a prompt. Inference splits into a compute-hungry prefill phase and a memory-hungry decode phase, and Etched built dedicated components for each: a prefill chip that runs at unusually low voltage, and an interconnect design called cluster scale memory that lets many chips share one memory pool at very low latency.
The company has pushed back on the perception that its hardware only runs specific large language models. According to Wachen, the systems handle any modern architecture, including Mixture of Experts models like DeepSeek and non-transformer designs like Mamba.
The money and the milestones
Etched raised its $120 million Series A in June 2024, a $500 million round at a $5 billion valuation in December 2025, and the $300 million Series C at $10.3 billion in July 2026, which TechCrunch reports is the highest valuation ever for a Sequoia-led Series C. Total funding tops $1 billion, with Andreessen Horowitz, SK Hynix, Jane Street, and Diffusion joining the latest round and individual backers including Peter Thiel, Andrej Karpathy, Dylan Field, and Amjad Masad.
The company employs about 400 people, recently opened an 80,000-square-foot production and prototyping facility near its San Jose headquarters, and runs a 2 megawatt data center. The remaining challenge is the hard one: mass-producing rack systems at scale, a step that has broken plenty of well-funded hardware startups before.
Why it matters
Inference is where AI's economics get decided. Training a frontier model happens a few times a year at a few companies, but inference runs billions of times a day at every company shipping an AI feature. As usage scales, the cost of serving tokens becomes the line item that determines whether AI products have real margins.
Cheaper inference at the silicon layer points one direction for founders: more products become economically viable to build. The story of how a 25-year-old CEO carried this bet from a dorm room to a $10.3 billion company is on Gavin Uberti's profile.
Frequently asked questions
What does Etched build?
Etched designs AI inference hardware sold as complete rack-scale systems, including a low-voltage prefill chip and a shared-memory interconnect called cluster scale memory. The systems are designed to run AI models faster and cheaper than general-purpose GPUs.
Who founded Etched?
Gavin Uberti (CEO), Robert Wachen (COO), and Chris Zhu (CTO) founded Etched in 2022 after dropping out of Harvard to build transformer-specialized silicon.
How much is Etched worth?
Etched was valued at $10.3 billion in its July 2026 Series C, a $300 million round led by Sequoia. That is roughly double its $5 billion valuation from December 2025.
Does Etched hardware only run transformer models?
No. The company says its systems run any modern AI model, including Mixture of Experts designs like DeepSeek and Qwen and non-transformer architectures like Mamba.
How is Etched different from Nvidia?
Nvidia sells general-purpose GPUs adapted for AI work. Etched builds silicon specialized for inference, with dedicated hardware for the prefill and decode phases, and sells complete rack-scale systems rather than individual chips.