Etched has raised $700 million in a funding round that values the US-based AI hardware company at $21 billion. At the same time, it has moved from building and testing its technology to putting its first rack into operation for Jane Street.
Jane Street led the round and is both an Etched investor and the company’s first customer. Etched shipped the rack to Jane Street last month, and the quantitative trading firm is deploying the technology within its own operations.
From Stealth Company to Working Hardware
Etched builds inference clusters for AI workloads, with its systems designed to run large mixture-of-experts models and non-transformer designs. Its clusters rely on two technologies called Low Voltage Inference and Cluster Scale Memory.
The company came out of stealth in June with more than 400 staff and a working chip. It also claimed first-pass silicon success within three years of its seed funding, a milestone that shows how quickly Etched has taken its design from early financing to hardware in a customer’s data centre.
Etched is developing three generations of hardware at the same time. That approach places the company’s current rack, future chip designs and cluster technology on the same development path, while the first customer deployment gives Jane Street a direct role in testing the system in its own operations.
Jane Street said its early testing has produced results that support the deployment. “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 data centre.”
A Large Funding Round With Customers Attached
The financing includes Sequoia Capital, Kleiner Perkins, Andreessen Horowitz, Bain Capital Ventures, Stripes, Positive Sum, Tiger Global, Neo, Primary and Blackstone. Jane Street’s role stands out because it is not only backing Etched financially; it has also taken delivery of the company’s first rack and is using the technology.
Etched has secured more than $1 billion in customer contracts spanning public and private AI firms and cloud providers. Those contracts give the company a commercial base beyond its first deployment, while the $21 billion valuation places a high value on its ability to deliver specialized inference hardware and clusters.
The company’s focus is inference, the part of an AI workload where a trained model produces results. Etched is targeting that work with clusters built for large mixture-of-experts models and non-transformer designs, rather than limiting its systems to one model structure.
That focus also connects the chip to the wider cluster. Low Voltage Inference and Cluster Scale Memory are the two technologies Etched uses to support its systems, and the company is developing three generations of hardware at the same time.
Jane Street’s Two AI Infrastructure Deals
Jane Street’s Etched investment and deployment come alongside a separate agreement with CoreWeave. In April, Jane Street had agreed to spend approximately $6 billion to use CoreWeave’s AI cloud platform.
CoreWeave will provide Jane Street with computing infrastructure across multiple data centre locations, and that infrastructure will incorporate Nvidia’s Vera Rubin technology. The CoreWeave agreement and the Etched rack give Jane Street two different forms of AI computing infrastructure: a cloud platform spread across data centres and a rack from a company whose technology it is deploying within its own operations.
For Etched, the Jane Street deployment adds a working customer site to a funding announcement that already includes a large valuation and more than $1 billion in customer contracts. The company now has a rack operating at its first customer, a chip that Jane Street has tested, and three hardware generations in development.
That combination captures Etched’s position on August 19, 2026: a company that recently emerged from stealth, reached first-pass silicon success within three years of seed funding, and is now turning its inference-cluster design into hardware running in a customer’s data centre.
