HomeFull speed ahead. Keep track of advancements in every frontier

TheTechno Optimist.

Mistral AI—Mistral ComputeMistral AI—Mistral ComputeMistral AI—Mistral modelsMistral AI—

Mistral Compute

by Mistral AIFR

PilotStage 3 of 5

A European-controlled AI cloud being built in France with NVIDIA.

Updated 28 May 2026Checked 25 Sep0 updates this week

Milestones

Next · Initial data center capacity in France
  1. Mistral Compute announced at VivaTech with NVIDIA11 Jun 2025Complete.
  2. €830M debt financing for compute30 Mar 2026Complete.
  3. Initial data center capacity in FranceNowCurrent milestone.

Most important updates

  • 28 May 2026
  • 30 Mar 2026
  • 11 Jun 2025

Upcoming

  1. 2027Mistral's target of 200 MW of AI compute capacity across Europe (next)

Current obstacles

  • Power and chips in EuropeBig clusters in Europe face higher power costs and slower grid connections than in the US.

Physics limits

  • Almost all power becomes heatNearly every watt a chip draws ends up as heat. A 72-GPU rack draws about 120 to 130 kW, over 10x an old server rack, so air can't cool it; liquid loops must.
  • Light is too slow for far-apart GPUsSignals in fibre travel about 200 km per millisecond. Training syncs GPUs many times a second, so one tightly coupled cluster must sit on one site or a few nearby ones.
  • At this scale something always breaksMeta logged 419 unexpected interruptions in 54 days on a 16,384-GPU Llama 3 run, about one every 3 hours. Bigger clusters fail more often and must checkpoint constantly.

How it works

4 parts
An NVIDIA DGX GPU server in Ifremer's Datarmor centre in Brest, France: the same class of hardware Mistral Compute fills its halls with
An NVIDIA DGX GPU server in Ifremer's Datarmor centre in Brest, France: the same class of hardware Mistral Compute fills its halls withPhoto: Gregory Rocher, Ifremer · CC BY 4.0 (opens commons.wikimedia.org)
Site

GPUs south of Paris

The first training site at Bruyères-le-Châtel holds about 13,800 NVIDIA GB300 GPUs in a 44 MW data centre run by Eclairion.

Sovereignty

European control

Data and hardware stay under European jurisdiction, the pitch to governments and firms wary of US cloud providers.

Offer

From bare metal to apps

Customers can rent raw GPU servers, a managed platform for training, or Mistral's own models and assistants running on top.

Grid

Low-carbon power

France's grid is mostly nuclear, so the site's electricity is among the lowest-carbon available to a large AI data centre.

Update log

3 updates

Thu 28 May

  • Minor: PressHardware

Mon 30 Mar

  • Major: PressFunding

Wed 11 Jun 2025

  • Major: BlogPartnership

About Mistral AI

The team behind Mistral Compute

Mistral AI

Open-weight models, Le Chat, AI cloud

Mistral AI is a Paris lab founded in 2023 by former Google DeepMind and Meta researchers. It makes efficient open and paid models and the Le Chat assistant. It is Europe's leading independent AI lab.

  • Raised about €3B led by Samsung in September 2026, valued above €21B (about $24B).
  • ASML led its €1.7B round in September 2025 and holds a significant minority stake.
  • The Mistral 3 family (December 2025) is open-weight under a permissive licence.

mistral.ai (opens mistral.ai)Company profile →

Founded
20233 yrs
Headquarters
France
Status
Private
Valuation
$24BprivateSep 2026
Raised
$5.6B3 rounds
Last round
Venture round · $3.5BSep 2026$24B post
Works in
AIOpen-weight modelsComputing
Coverage
2 programs · 11 updateslatest 8 Sep 2026checked 25 Sep
Lead investors
SamsungASMLLightspeed Venture Partners
Partners
NVIDIAMicrosoft
Investors
ASML
People
Arthur MenschCEO and co-founderGuillaume LampleChief Scientist and co-founderTimothée LacroixCTO and co-founder
Primer