1. Providers
  2. Custom-silicon cloud

Provider category

Custom-silicon cloud

Sells inference or systems on its own chips instead of GPUs.

Source: https://neocloud.fyi/sources/ Checked 2026-09-18

site's own definition

NameHQWhat it sellsEuropean footprint
Groq US Inference API and systems on its own LPU chips Helsinki and UK, both Equinix colocation
Cerebras US Wafer-scale systems and an inference cloud Mikkeli, Finland (165 MW, phased; Compute Nordic Finland partnership)
SambaNova US Inference cloud and on-premises systems None confirmed

Connects to