Pricing

Predictable Costs.
Unmatched Performance.

Transparent pricing for on-demand GPU access. For reserved capacity, dedicated enterprise clusters, or on-premise deployments, our team will craft a tailored solution for your specific workload.

All prices are indicative per GPU per hour, net, and apply to the initial launch region. HGX configurations approx. 5% less than DGX. Founding Partner pricing upon early commitment. Final pricing upon contract signing.
GPU Model Dev (per-second) On-Demand Reserved (6mo–5yr)
NVIDIA HGX H100 Hopper · 80 GB HBM3 · 16 GPUs · 200 Gbit/s from € 0,0008/s(~€ 2,95/hr) € 2,95 /hr Book a MeetingRequest Access
NVIDIA HGX H200 Hopper · 141 GB HBM3e · 16 GPUs · 200 Gbit/s from € 0,0010/s(~€ 3,50/hr) € 3,50 /hr Book a MeetingRequest Access
NVIDIA GB200 Blackwell · NVL72 from € 0,0013/s(~€ 4,82/hr) € 4,82 /hr Book a MeetingRequest Access
NVIDIA GB300 Blackwell Ultra · NVL72 from € 0,0015/s(~€ 5,45/hr) € 5,45 /hr Book a MeetingRequest Access
Fractional GPU (MIG) 1/8 GPU · Ideal for Dev & Research from € 0,60 /hr - -

All prices indicative · per GPU / hour (or per second for Dev) · Net · DGX reference · HGX approx. 5% less · Reserved pricing requires a meeting

Included

Included in Every Model

No surprises. All essential services are included by default.

GPU, CPU, RAM & NVMe Storage
NVLink & InfiniBand Network
Direct Liquid Cooling
Redundant Power Supply
24/7 Monitoring & Support
Console & API Access
EU Data Residency Guaranteed
GDPR-Compliant Processing
Models

Four Models, Tailored to Your Workload

From per-second developer access to dedicated enterprise infrastructure. Choose the model that fits your AI training, inference, or storage needs.

Pay-as-you-go

Dev

Per-second billing for developers and researchers. Fractional GPUs (1/8 MIG), single GPUs, Docker-native.

  • Per-second billing
  • Fractional GPU access (1/8 MIG)
  • No minimum spend
  • Jupyter & SSH included
Request Access
Flexible

On-Demand

Flexible usage without minimum commitment. Ideal for experiments, prototypes, and variable workloads.

  • No upfront commitment
  • Billed by the hour
  • Capacity subject to availability
  • Maximum flexibility
Request Access
Best Value

Reserved

Guaranteed capacity at reduced prices. 6 months to 5 years. Discuss your requirements in a meeting.

  • Up to 50% savings vs. on-demand
  • Guaranteed GPU capacity
  • Predictable monthly costs
  • SLA-based support
Book a Meeting
Enterprise

Dedicated

Exclusive hardware for maximum isolation. Ideal for regulated industries and large-scale training.

  • Dedicated physical hardware
  • Complete isolation
  • Custom configurations
  • Enterprise SLA included
Book a Meeting
Founding Partners Program

Become a Founding Partner

Your early commitment helps finance the build. Founding Partners receive preferred reserved pricing, guaranteed capacity, and direct access to the team.

Contact Us Explore Platform →