
By burning the transformer architecture into our chips, we’re creating the world’s most powerful servers for transformer inference.

By burning the transformer architecture into our chips, we’re creating the world’s most powerful servers for transformer inference.
What they do: Designs AI chips (Sohu) optimized for transformer-model inference
Stage / funding: Series A (announced Jun 25, 2024), reported ~$120M round; total funding reported $625.4M
Founded: Around 2022
Team size: Approximately 317 employees
| Company |
|---|
AI inference performance and efficiency for transformer models
2022
Computer Hardware Manufacturing
$120,000,000
Round reported to include multiple institutional and angel participants
“Includes institutional VCs and prominent angel/backer participation (e.g., Two Sigma Ventures, Peter Thiel, Thomas Dohmke)”
About Etched Etched is building AI chips that are hard-coded for individual model architectures. Our first product (Sohu) only supports transformers, but has an order of magnitude more throughput and lower latency than a B200. With Etched ASICs, you can build products that would be impossible with GPUs, like real-time video generation models and extremely deep chain-of-thought reasoning.
Mechanical Engineer We are seeking an experienced Mechanical Engineer to join our team and help transform complex designs into reliable, high-power data center products. As a Mechanical Engineer at Etched, you will be integral to the development of high-performance data center products, from initial concept through to production. You will collaborate closely with both internal and external design engineers, including JDM (Joint Design Manufacturer) partners. Your expertise in mechanical design, particularly for data center chassis and high-power systems, will ensure our products meet rigorous performance, reliability, and manufacturability standards.
Representative Projects
You maybe a good fit if you have
Benefits
How We’re Different Etched believes in the Bitter Lesson. We think most of the progress in the AI field has come from using more FLOPs to train and run models, and the best way to get more FLOPs is to build model-specific hardware. Larger and larger training runs encourage companies to consolidate around fewer model architectures, which creates a market for single-model ASICs.
We are a fully in-person team in Cupertino, and greatly value engineering skills. We do not have boundaries between engineering and research, and we expect all of our technical staff to contribute to both as needed.