Etched raised $700 million on Tuesday at a $21 billion valuation. Jane Street led the round after testing the hardware, and the jump from a $10.3 billion valuation in July gives the deal a rare speed for a private AI chip company.
Robert Wachen on two stages
Robert Wachen said, “Inference is built in two stages, prefill and decode.” He described prefill as mathematically and compute-intensive, while decode is memory-intensive, which is why Etched split the system into a prefill chip and a separate memory-and-interconnect design.
Etched said the prefill chip runs at low voltage, while its cluster-scale memory lets many chips connect together and use a shared memory pool at very, very fast, low latency. The company says that setup cuts costs and speeds up frontier inference clusters built to run frontier models.
Jane Street adds a rack
Jane Street said, “We tested the chip and are pleased with the early results. Etched’s unique approach to inference delivers the precision we will need to support our most demanding workloads. We’re excited to now have our own rack running in our datacenter.” That gives the round a rare operational signal: the lead investor is not just funding the company, but using its hardware.
Etched’s prior financing moved fast before this deal. The company was valued at $5 billion in December, then raised a $300 million Series C at a $10.3 billion valuation in July, before Tuesday’s new round took it to $21 billion and added nearly $11 billion in value in about a month.
Etched and frontier models
Etched says its systems can run any frontier model, even though the company originally intended its chips to be custom-designed for one frontier model. That shift matters for buyers because it widens the set of workloads the hardware can serve, while also leaving the company to overcome the older perception tied to its first design goal.
Other backers include Kleiner Perkins, Sequoia Capital, Andreessen Horowitz, Peter Thiel, Tiger Global, Bain Capital Ventures, Neo, Stripes, Primary, Positive Sum, Diffusion, Argo, and Blackstone. A separate profile of a first marriage renovation is unrelated, but the financing here has the cleaner read: a hardware startup built around inference is now being valued like a company with much broader ambition.
