Etched Secures $300m to Scale AI Inference Infrastructure

AI inference chip startup Etched has raised $300 million in a Series C funding round, more than doubling its valuation to $10.3 billion. The latest investment comes less than a month after the company emerged from stealth, reinforcing investor confidence in its specialized AI inference hardware and long-term expansion plans.

The funding round was led by Sequoia, with participation from SK Hynix, Andreessen Horowitz, Jane Street, Diffusion, and Argo, as Etched continues to scale production and prepare its first commercial deployments.

Funding Strengthens Expansion Strategy

Article content — Image Credit: www.etched.com

The latest capital raise marks another significant milestone for the Cupertino-based startup, which previously completed a $500 million funding round in December.

Key funding highlights:

Series C funding: $300 million
New valuation: $10.3 billion
Lead investor: Sequoia
Participating investors: SK Hynix, Andreessen Horowitz, Jane Street, Diffusion, and Argo

In an interview with TechCrunch, Etched COO Robert Wachen said the latest financing increased the company’s valuation to $10.3 billion, more than doubling its value following the December funding round.

Purpose-Built AI Inference Hardware

Founded in 2022 by Harvard University dropouts Gavin Uberti and Chris Zhu, Etched announced last month that it had developed a working AI inference chip and secured more than $1 billion in customer contracts.

The company’s processor is designed specifically for AI inference and separates workloads into two dedicated stages:

Compute-intensive prefill (prompt processing)
Memory bandwidth-intensive decode (output generation)

Etched says the chip operates at a significantly lower voltage than other AI chips, reducing heat generation while allowing more transistors to be integrated. The architecture also incorporates a shared low-latency memory pool and a proprietary high-bandwidth interconnect designed to accelerate memory access across multiple chips.

Scaling Manufacturing and Infrastructure

Manufactured using TSMC’s N4P process technology, the chip is currently undergoing customer validation in a rack-scale configuration, with first shipments planned for this summer.

To support production, Etched revealed it has begun fabricating hundreds of millions of dollars’ worth of inference clusters while opening a new 10MW, 80,000-square-foot laboratory in Milpitas. The company said the facility will support continuous deployment of first-generation hardware and rapid prototyping of future products.

Highlighting its broader vision, Etched said: “Our mission is to run the world’s inference,” adding that expanding access to advanced AI models will require more efficient computing capable of delivering higher performance per watt at gigawatt scale.

Looking Ahead

With fresh funding, a rapidly growing valuation, and production infrastructure now expanding, Etched is positioning itself to accelerate deployment of dedicated AI inference systems.

The company’s focus on specialized hardware, large-scale manufacturing, and next-generation inference clusters reflects its strategy to meet growing demand for efficient AI computing.