The startup has raised roughly $800 million and secured over $1 billion in customer contracts for its transformer-optimized Sohu chip, with a16z calling inference the most important workload of our era.
A startup most people have never heard of is quietly assembling the ingredients to challenge Nvidia’s stranglehold on AI compute. Etched, founded in 2022, has partnered with Andreessen Horowitz to build purpose-built inference hardware from the ground up, and a16z partners are now publicly declaring that AI inference is “the most important workload of our era.”
Etched has raised approximately $800 million in total funding and locked in over $1 billion in signed customer contracts for its rack-scale inference systems.
What Etched is actually building #
The core product is Sohu, an application-specific integrated circuit (ASIC) designed exclusively to run transformer models. Instead of building a general-purpose chip that can do lots of things pretty well, Etched built a chip that does exactly one thing, run the architecture behind every major large language model, and does it extraordinarily well.
The Sohu chip is fabricated on TSMC’s advanced N4P process node. Etched has already completed what’s known as an A0 tape-out, meaning the first version of the physical chip design has been sent to TSMC for manufacturing.
Etched says Sohu delivers more than 10x performance gains over leading GPUs like Nvidia’s H100. Some reports go further, suggesting that a single Sohu server could replace roughly 160 H100 units.
First racks are planned to ship in summer 2026.
Why inference matters more than training #
Training happens once. Inference happens forever. As AI adoption scales from research labs into consumer products, enterprise software, and autonomous systems, the balance of compute demand is tilting heavily toward inference. That’s the bet a16z is making by backing Etched and framing inference as the defining computational challenge of this period.
The fundraising trajectory tells its own story #
Etched secured an additional $500 million round in late 2025 at a $5 billion post-money valuation. Now the company is reportedly in discussions for a new round that would value it at nearly $20 billion, roughly quadrupling its valuation in a matter of months.
What this means for investors watching the AI hardware race #
ASICs carry inherent risk. They’re optimized for a specific architecture, in this case transformers. If the AI industry moves beyond transformers toward some new model architecture, a chip designed exclusively for transformers becomes significantly less valuable. Etched is making a concentrated bet that transformers remain the dominant paradigm for years to come.
Watch the summer 2026 shipping timeline closely, because if Etched delivers working hardware with the performance it promises, the ripple effects will extend well beyond a single chip company.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our