Etched has moved into a new valuation tier with unusual speed. The AI hardware startup on Tuesday announced another $700 million in funding at a $21 billion valuation, with Jane Street leading the round after testing and buying the company’s hardware.
The jump is striking because it follows two earlier valuation milestones in a short span. Etched was valued at $5 billion in December, then raised a $300 million Series C at a $10.3 billion valuation in July. Now its valuation has doubled to $21 billion in a month.
A fast repricing around AI inference
The latest round shows how much investor attention is moving toward AI inference, the computing work that happens after a user submits a prompt. While training often dominates public discussion of AI infrastructure, inference is the part users experience directly when a model produces an answer.
Etched is selling its technology as complete systems it calls “frontier inference clusters.” The company is positioning those systems around speed and cost, two issues that matter as AI workloads become more demanding and are run at larger scale.
The scale of the repricing is central to the story. Moving from a $10.3 billion valuation in July to $21 billion in a month gives Etched an unusually rapid step-up even within the AI market. The company’s earlier $5 billion valuation in December also makes the pace clear: investors are assigning far more value to its approach than they did only months earlier.
What Etched says it built differently
Co-founder and COO Robert Wachen told TechCrunch that investor enthusiasm is tied to two new components Etched designed from scratch. Both are aimed at inference, but they focus on different stages of the process.
Wachen described inference as having two stages: “prefill and decode.” In the prefill phase, the system works through the prompt and its context. In the decode phase, the system generates output tokens, which are the pieces of the answer the user ultimately sees.
Etched’s prefill chip runs at low voltage. According to the company, that lets it fit in more transistors while avoiding the usual heat problems associated with high-end AI chips. The company says this helps it process more tokens faster.
For decode, Etched built a new kind of memory and an interconnect it calls cluster-scale memory. Wachen said it lets many chips connect and use a shared memory pool at very fast, low-latency speeds. Etched says the combined result is higher speeds and lower costs.
Jane Street’s role gives the round extra weight
Jane Street led the new funding round after testing and buying Etched’s AI hardware. In the blog post announcing the round, Jane Street said it had tested the chip and was pleased with the early results.
The investment firm also said Etched’s inference approach delivers the precision needed for its most demanding workloads, and that it now has its own rack running in its datacenter. That matters because the round was not based only on a future promise; the lead investor had already evaluated and purchased the hardware.
Other investors in Etched include Kleiner Perkins, Sequoia Capital, Andreessen Horowitz, Peter Thiel, Tiger Global, Bain Capital Ventures, Neo, Stripes, Primary, Positive Sum, Diffusion, Argo, and Blackstone.
Clearing up an early perception
Etched is also trying to move past an early perception about what its chips do. In its first concept, the company intended to etch a particular model into its chips, which made it sound as if each chip would be custom-designed for one frontier model.
That is no longer the case, according to the source article. Etched’s systems can run any frontier model.
This distinction matters because a system limited to one model would carry a different kind of risk for buyers and investors. If the hardware can support any frontier model, the value proposition is broader: customers are not tied to a single model choice when using the system.
Why the valuation jump matters
Etched’s new $21 billion valuation reflects a bet that AI infrastructure will keep rewarding specialized approaches to inference. The company is not simply presenting a chip, but a full system built around the performance needs of frontier models.
The open question is execution. Etched is promising higher speeds and lower costs through custom components, including its low-voltage prefill chip and cluster-scale memory for decode. The latest funding round suggests major investors believe those design choices are commercially meaningful.
For the AI hardware market, the story is another signal that inference has become a major competitive front. As more users interact with AI systems through prompts and generated answers, the infrastructure behind that response time becomes more important. Etched’s fast valuation rise shows how much value investors see in making that process faster, cheaper and more scalable.