Fluidstack
CuratedActiveAI cloud building and running gigawatt-scale GPU clusters for frontier labs
From
—
GPU from
$2.99/GPU-hr
Locations
1
countries
Data centers
0
sites
Networks (ASN)
1
ASNs
Plans & prices
1
plans
Overview
Fluidstack, founded in 2017 out of Oxford, started by aggregating spare data center GPUs and now designs, builds and operates large dedicated GPU clusters for AI labs and governments, including gigawatt-class US campuses. Smaller on-demand and reserved GPU capacity remains available.
Plans & prices
GPU instances1
| Instance | Config | per GPU-hr | Total /hr | Spot | Locations | Source |
|---|---|---|---|---|---|---|
H100_NVLINK_80GB::8 252 vCPU · 1440 GB RAM | 8× H100 SXM80 GB | $2.99 | $23.92 | — |  1 regions | skypilot 4 days ago |
GPU models
Networks (ASN)
Similar providers
HyperstackNexGen Cloud's GPU cloud with per-minute H100/H200/B200 VMs, spot and reserved
CUDO ComputeUK AI infrastructure provider supplying dedicated NVIDIA GPU clusters and cloud GPUs
CivoUK cloud built around fast managed Kubernetes, plus VMs and NVIDIA GPUs
NscaleVertically integrated AI cloud with hydro-powered data centers in Norway and the UK
OriLondon GPU cloud platform that merged into Brookfield-backed Radiant