Europe as a hard requirement
Every machine runs in the EU or the EEA, and the subprocessor list shows where.
04 EU compute H100 to B300
Your own GPUs,
sized for your workload.
Private GPUs in Europe, H100 to B300, sized for response time, generation speed and concurrent users.
GLM
DeepSeek
Qwen
Kimi
Mistral
Llama
gpt-oss
Gemma
Loes
Nemotron
Claude Code
Codex
OpenCode
Cursor
Cline-
Loes
-
OpenAI SDK
Anthropic SDK
What it does
GPUs that work
for you alone.
Some work should not share hardware. EU compute gives your team its own GPUs in European data centres, from a single H100 to several nodes of B300s, with nobody else on the machines.
We size the setup on what your people notice: how fast the first token arrives, how many tokens per second they get and how many of them work at the same time. A plan can mix card types and data centres, and grows or shrinks with the work.
A machine only qualifies if its data centre sits in the EU or the EEA. The United Kingdom and Switzerland are excluded, and so is any location we cannot establish with certainty. The list of subprocessors is public.
Read more
EU compute/ Size your setup
Why teams use it Sovereign
Sovereign
by design.
Sized on first token and speed
You tell us the response time and speed your team needs, and we plan the cards to meet it.
A private connection
With WireGuard on the machine, the model port is closed to the public internet and your traffic goes through your own tunnel.
In the same platform
Your GPUs sit next to the shared models, with the same key, the same team routers and the same Activity view.
Questions
Questions we get
Which GPUs can we get?
H100, H200, B200 and B300, from one card to several nodes of eight. What is free changes by the hour, so we show it when we plan with you.
How is it billed?
A dedicated GPU is billed per minute while it runs. A setup planned for a team gets a fixed price per month.
Can we bring our own GPUs?
Talk to us about running on hardware of your own. Spare GPUs can also join our GPU pool, which runs training and benchmarks, and earn for every second they work.
The platform
One platform, four layers
Start with one layer and add the others when the work asks for it. They share one key, one balance and one Activity view.
Plan your
own GPUs.
Tell us about the work and the people who use it, and we size a setup with you.
