AI Chip Startup Etched Raises $300M Series C at $10.3B Valuation

Etched, a San Jose-based AI hardware company, has raised $300 million in Series C funding, bringing the company's valuation to $10.3 billion as it moves closer to commercial deployment of its purpose-built AI inference infrastructure.

The round includes backing from Sequoia, Andreessen Horowitz, Jane Street, Argo, and SK hynix, alongside existing investor Soma Capital.

Building Hardware for AI Inference

Founded by Gavin Uberti, Robert Wachen, and Chris Zhu, Etched is developing hardware specifically designed for running AI models rather than relying on general-purpose GPUs.

Its flagship Sohu chip is designed to improve the speed, cost efficiency, and power consumption of AI inference. The company's broader approach extends beyond the chip itself, with Etched developing the boards, interconnects, and racks required to build complete inference clusters.

That focus comes as AI companies increasingly look beyond model training and toward the infrastructure required to serve AI workloads at scale.

A Bet That Started Before the AI Boom

Etched's story began in 2022, before ChatGPT brought generative AI into the mainstream.

Uberti, Wachen, and Zhu left Harvard to pursue a difficult proposition: building an entire AI inference system from the ground up, rather than adapting existing infrastructure around general-purpose computing hardware.

That meant developing the silicon as well as the surrounding infrastructure needed to operate it at scale.

The bet is now moving closer to reality. Etched says its first racks are scheduled to ship to customers this summer, while a 10-megawatt site is also being developed. Customers are already running workloads remotely on Etched hardware.

Backing for a Different AI Hardware Bet

The $300 million round gives Etched substantial capital to continue scaling its infrastructure and commercial operations.

The company's $10.3 billion valuation also puts it among the more closely watched AI infrastructure startups as investors look for alternatives to the dominant general-purpose GPU model.

For Etched, the thesis is straightforward: as inference becomes a larger part of AI economics, specialized hardware could offer meaningful advantages in performance, cost, and energy efficiency.

Looking Ahead

The next test will be commercial execution. Etched now needs to move from promising architecture and early customer workloads to reliable infrastructure operating at meaningful scale.

With its first racks approaching deployment and a major infrastructure buildout underway, the company is entering a phase where its long-standing bet on purpose-built inference hardware will face the market.

About Etched

Etched is an AI hardware company based in San Jose, California, developing purpose-built infrastructure for AI inference. Founded by Gavin Uberti, Robert Wachen, and Chris Zhu, the company is building specialized chips and inference systems designed to make running AI models faster, more efficient, and less expensive than relying solely on general-purpose GPUs.

Company Snapshot

Company: Etched
Headquarters: San Jose, California
Founders: Gavin Uberti, Robert Wachen, Chris Zhu
Round: Series C
Funding: $300 million
Valuation: $10.3 billion
Investors: Sequoia, Andreessen Horowitz, Jane Street, Argo, SK hynix, Soma Capital

Source & Credits: TechCrunch