Etched is already eyeing a fresh funding round that will more than double the AI inference chips startup’s valuation at around $21 billion, barely one month after its late July Sequoia-led $300 million Series C round that valued the firm at $10.3 billion.
The new numbers cited in the Wall Street Journal’s August 18 report represent a fourfold expansion of the firm’s $5 billion valuation in December 2025. Etched also closed on Jane Street as its first customer, delivering a server rack filled with AI processors optimized for rapid inference computing to the Wall Street quantitative trading giant.
The scale of the firm’s growth after only coming out of stealth on June 30 signals the rush for alternatives to Nvidia’s big lead at the head of the supply chain that supplies the silicon powering the AI boom.
“This round reflects a growing industry conviction that the challenge demands a new entrant willing to rebuild the stack from first principles,” co-founder and CEO Gavin Uberti said in a statement.
Andreessen Horowitz (a16z), SK Hynix, Jane Street and Diffusion Capital also joined the July 31 financing round.
What do Etched’s chips do?
Etched sells full rack systems optimized for the inference portion of the AI compute stack, which occurs after users submit prompts. The firm believes its inference focus allows it to build specialist chips compared to Nvidia, which builds all-purpose GPUs for both AI training and inference.
Etched splits the work into the prefill and decode phases.
- The prefill phase reads and interprets prompts using what Etched co-founder Robert Wachen calls low-voltage inference. That design choice allows the chip to run cooler with more transistors and a higher clock speed.
- The decode phase, where the model writes its answer one token at a time, uses a Cluster Scale Memory design. The shared memory design allows accelerators inside one rack to draw data from another memory pool instead of having to copy the data themselves.
Etched said its systems are already inside DeepSeek, Qwen, Mamba, and Llama models.
Big money is flowing into inference
Etched is riding the hot hand in a broader market that forecasters back as the fastest-growing slice of a hot AI sector. The company delivered on its promise to start shipping chips by the summer, filling over $1 billion of booked orders, as reported by Cryptopolitan.
Bloomberg Intelligence set a $1.3 trillion target for the inference sector, doubling the size of the AI training market by 2032. Iron Mountain and Structure Research back inference capacity to overtake training capacity this year and continue to grow to account for 80% of AI compute load by 2030.
The rush for AI data centers has created a steady demand route fed by names such as Nvidia, Cerebras, and AMD, which are also building inference-specific systems on the same prefill-and-decode split that Etched describes.
Etched now employs more than 400 people and reported first-pass silicon success on TSMC’s N4P process. It runs a 2-megawatt data center at its San Jose headquarters and has opened a new 80,000-square-foot, 10-megawatt facility in nearby Milpitas, TechCrunch reported.
