Etched doubled its valuation to $21 billion in a month -- with its new lead investor as its first paying customer
The AI inference-chip startup's $700 million Series D was led by Jane Street, which also just took delivery of the company's first production rack -- a funding round and a customer contract announced in the same breath.
Etched, the chip startup building hardware that does nothing but AI inference, announced on August 18, 2026 that it raised $700 million at a $21 billion valuation -- doubling its valuation in roughly a month, since a $300 million Series C at $10.3 billion had closed only weeks earlier. Jane Street led the new round, with Kleiner Perkins, Sequoia, Andreessen Horowitz, Tiger Global, Bain Capital Ventures and Blackstone among the other participants.
The investor is also the customer
The detail that makes this round more than a valuation headline: Jane Street didn't just write the biggest check, it also took delivery of Etched's first production rack of inference-only silicon the same week, becoming the company's first paying customer. Etched co-founder and CEO Gavin Uberti frames the urgency behind that overlap directly: "We've felt the urgency to get our hardware into customers' hands and run real workloads since day one..." It's a real vote of confidence -- a trading firm putting its own compute budget behind the chip it just helped fund -- but it's also a structure worth naming plainly rather than glossing over: the same firm is now both Etched's largest new backer and its first reference customer, which makes Jane Street's own account of how the chip performs harder to treat as fully independent.
The underlying bet is about the economics of inference specifically, separate from training: Etched's chips are built around a low-voltage prefill stage plus new memory and interconnect for the decode stage, the two steps every inference request actually runs through. For any organization tracking what it spends per token rather than per training run, that's the more relevant cost line as usage scales -- and it's the reason a chip startup with no training-hardware ambitions at all just got priced at $21 billion.