Etched Raises $700M and Delivers First Rack
Etched raised $700 million at a $21 billion valuation in a Jane Street-led round announced August 18, while also delivering its first customer rack to the quantitative trading firm. According to Etched's release, Jane Street is actively deploying the system into its workloads. The financing follows Etched's $300 million round at a $10.3 billion valuation on July 23.
Etched raised $700 million at a $21 billion valuation in a funding round led by quantitative trading firm Jane Street, the inference-chip startup announced August 18. The company also reported that it shipped its first customer rack to Jane Street last month and that the firm is actively deploying the hardware in its workloads.
Kleiner Perkins, Sequoia, Andreessen Horowitz, Tiger Global, Bain Capital Ventures, Neo, Primary, Stripes, Positive Sum, and Blackstone also participated, according to Etched's press release. Axios separately reported the same round, valuation, lead investor, and investor list.
The valuation is more than double the $10.3 billion assigned to Etched in its July 23 financing. The Next Web reported that the earlier $300 million Series C was led by Sequoia, placing the two valuation marks 26 days apart.
First production customer
Jane Street both led the new financing and became Etched's first disclosed customer. In a statement published by Etched, 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 data center."
Etched co-founder and CEO Gavin Uberti described the Jane Street deployment as proof that the hardware is running real workloads, according to the company's release. SiliconANGLE reported that Etched had installed an Etched-powered rack at Jane Street's data center last month, while the startup has also built a two-megawatt AI cluster at its San Jose headquarters for prospective customer testing.
According to Etched's release, it has secured more than $1 billion in customer contracts across frontier AI companies and cloud providers.
Hardware claims and production ramp
Etched describes its system around two technologies: Low Voltage Inference, or LVI, and Cluster Scale Memory, or CSM. The company claims LVI increases compute density within a given power envelope, while CSM creates a shared memory pool across a cluster rather than limiting memory to individual chips.
SiliconANGLE reported that Etched's math blocks operate at less than half the voltage of most AI accelerators, a company claim the publication linked to higher FLOPs density. The outlet also noted that Etched did not disclose how its voltage-optimized circuits work. Such architecture choices are relevant to inference because serving large models requires balancing arithmetic throughput, memory capacity, memory bandwidth, latency, and power consumption.
The Next Web reported that Etched's design uses separate prefill and decode approaches, and that the company states its systems support mixture-of-experts models including DeepSeek and Qwen as well as non-transformer architectures such as Mamba. Etched's release states that its clusters are running large MoE and non-transformer designs, but it does not provide workload-level benchmark data or model configurations.
For infrastructure teams, comparable inference-hardware evaluations generally require more than peak compute claims: end-to-end tokens per second, time-to-first-token, interconnect behavior, precision support, model compatibility, rack power, and software maturity determine deployment value. Jane Street's reported production deployment is a concrete customer milestone, while broader technical comparisons will depend on independently reproducible measurements and further customer deployments.
Etched stated that the new capital will support factories, supply chains, and fleet software as it pursues larger-scale inference deployments. The company emerged from stealth in June with a working chip and reported first-pass silicon success on TSMC's N4P process, according to SiliconANGLE and The Next Web.
Key Points
- 1Etched raised $700 million at a $21 billion valuation while completing its first disclosed customer delivery to Jane Street.
- 2Jane Street's role as both lead investor and deployed customer supplies a concrete production-use milestone, though independent performance data remains unavailable.
- 3Inference hardware evaluations across the industry typically depend on latency, memory behavior, power, software maturity, and reproducible workload benchmarks.
Scoring Rationale
The $700 million round and $21 billion valuation make the financing significant for the AI-inference hardware market. Its first disclosed production customer deployment adds relevance for teams tracking alternatives to incumbent accelerator platforms, although public independent benchmarks are limited.
Sources
Primary source and supporting public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
