Skip to content

Etched - AI supply-chain exposure

PrivateChipsUnited States

Startup designing Sohu, an ASIC hard-wired for transformer inference, claiming large efficiency gains over GPUs; raised $120M+ but pre-volume production.

The quick read

The model reads Etched primarily as a r&d in Chips.

Chain weight
9/100
Parts exposed
1of 80
Layers spanned
1of 5
Close peers
5
Chain position1 of 80 parts1 of 5 layers

Etched touches 1 of the 80 parts of the AI value chain, across 1 of its 5 layers.

  1. Energy14 parts
  2. Chips1 of 24
  3. Infrastructure24 parts
  4. Models12 parts
  5. Applications6 parts
Energy at the top, applications at the bottomBar height: exposure score, 0 to 100

How the model reads it

The structural read · model-generated

All of its mapped exposure sits in Chips.

Chain footprint by layer

  • Chips100%

How it participates

  • R&D100%

The story

In depth · editorial + model · written 2026-07-13

Etched is a chip-design startup building Sohu, an application-specific integrated circuit - an ASIC - hard-wired for one job: running transformer models, the architecture behind nearly every large language model. Rather than the flexible, general-purpose design of a GPU, it bakes the transformer directly into silicon, trading versatility for raw efficiency on that single workload. It designs the chip but does not manufacture it, and remains pre-volume.

Its structural bet is that the transformer is durable enough to be worth freezing into hardware - a wager that inference, not training, becomes the dominant and cost-sensitive workload. The model places it at the periphery because it is young, single-purpose and unproven at scale; if the bet lands, a fixed-function chip could undercut GPUs on the economics of serving models. If the architecture shifts, that same specialisation becomes the risk.

Its one link

Every part Etched touches

The part of the chain it is mapped to, and how much of its book sits there.

Supply concentration

What its parts are made of, and where.

The critical materials and the places the parts it touches depend on most, ranked by how much of its exposure leans on each.

Critical materials it leans on

  1. 1Photoresist
  2. 2Tantalum
  3. 3ABF Substrate (Ajinomoto Build-up Film)

Geographic concentration

  1. 1Israel

Questions

Frequently asked

What is Etched's role in the AI supply chain?

The model reads Etched primarily as a r&d in Chips.

Which parts of the AI value chain is Etched exposed to?

Etched is mapped to 1 part of the AI value chain, most strongly ASIC. It sits primarily in the Chips layer as a r&d.

Does Etched own an AI bottleneck?

Not in the current model - Etched is exposed to constrained parts but sits downstream of them rather than producing them.

Who are Etched's closest peers by AI-chain position?

By shared chain dependencies: Loongson Technology, SOPHGO Technologies, Horizon Robotics, Black Sesame International Holding.

In the app

Go live on Etched

This page is the map. Inside THE ENTITY it becomes live: the full graph around the company, the case for and against, and how today's news moves through its chain.

  • The interactive dependency graph: every supplier and buyer
  • The analyst bull and bear case, and the valuation lens
  • Live signals and the day's movers, read through the chain
  • Your own positions and thesis, tracked against the chain
See pricing

model v2.01 · research, not advice

Chain analytics are illustrative, order-of-magnitude estimates from our model of the AI value chain - not investment advice.

as of 2026-11-09Medium confidence model v2.01
All companies