Products Solutions Pricing Infrastructure Security Compute for equity Company
Talk to our team Get a quote

Sovereign GPU cloud in the EU · NVIDIA BlackwellEurope's AI, on European compute.

Sapience is building Europe's sovereign GPU cloud: dedicated NVIDIA Blackwell clusters, live in Bratislava today, with further EU sites to follow. RTX PRO 6000 capacity is fully reserved through 2026; HGX B300 clusters can be reserved now for delivery from December 2026.

Reserved by the cluster · USD per GPU-hour, excl. VAT
Bratislava, Slovakia Live
RTX PRO 6000 BlackwellFully reserved through 2026
HGX B300 clustersReservations open · from Dec 2026
Minimum8-server cluster · 64 GPUs
Choose your compute

Two Blackwell platforms, sold by the cluster.

Pick the node for the job. Capacity is sold on reservation at cluster level, 8, 24 or 32 servers, priced per GPU-hour for the term in USD, excluding VAT.

Blackwell Server Edition · 8 GPUs per server

NVIDIA RTX PRO 6000

Fully reserved · 2026

For inference, agentic AI, multi-application pipelines and professional visualisation. Models up to 96 GB per GPU, or sharded across the node's 768 GB.

Price on quote per GPU-hour for the term · clusters of 8, 24 or 32 servers
GPU memory
8 × 96 GB GDDR7 · 768 GB per server
Purchase terms
Reservation, from three months. Rate fixed for the term.
Availability
Current capacity is fully reserved through 2026. Reservations for 2027 are open.
Blackwell Ultra · 8 GPUs per server

NVIDIA HGX B300

Reservations open

For large-scale training, foundation models and high-volume inference. 288 GB of HBM3e per GPU with 1.8 TB/s NVLink between GPUs; multi-node clusters on a 1.6 Tb/s fabric.

Price on quote per GPU-hour for the term · clusters of 8, 24 or 32 servers
GPU memory
8 × 288 GB HBM3e · 2.3 TB per server
Purchase terms
Multi-year reservation. Rate fixed in the contract.
Availability
Reservations open. First deliveries from December 2026; your date is confirmed in the contract.
Rates depend on term, cluster size and support scope and are confirmed in a written quote. Clusters are sold whole; single servers and single GPUs are not offered. How reservations and pricing work →
The operation behind the offer

Bratislava first. A European network next.

Our first facility is live in Bratislava, in a colocation data centre where Sapience leases dedicated space, power and cooling: 11 MW identified at the site, 5 MW deploying for the first HGX B300 cluster. Further EU sites follow the same model and are announced as they are contracted.

FacilityColocation data centre · Bratislava
Power5 MW deploying · 11 MW identified
Next sitesFurther EU locations, announced when contracted
OperationsRun by the Sapience team, 24/7
Dotted map of Europe marking Bratislava
Bratislava · Live

What Sapience operates

  • The GPU systems: NVIDIA Blackwell nodes installed and run by the Sapience team
  • The node fabric, storage and tenancy isolation
  • Provisioning, monitoring and technical support
  • Customer contracts, pricing and data-residency commitments

What the colocation partner provides

  • The data hall in Bratislava and the dedicated space Sapience leases in it
  • Redundant power and cooling paths
  • Physical site security and access control
Built with NVIDIA (Partner Network), Supermicro systems, World Wide Technology integration and Rafay platform software.
NVIDIA Supermicro World Wide Technology Rafay
Why European infrastructure matters

Your own nodes, in the EU, run by a European operator.

Three things decide where your data goes and who can touch it: who else shares the hardware, where it physically sits, and who operates it.

Dedicated tenancy

Each customer runs on its own nodes. Compute, storage and network are not shared with other tenants, and no virtualisation layer sits between your workload and the hardware.

Data location

Your data is processed and stored on Sapience systems in Bratislava, Slovakia. Sapience does not replicate customer data outside the EU; backup and replication arrangements are set out in your contract.

Operational control

The infrastructure is owned and operated by Sapience, a European company under EU law, with no dependency on US hyperscalers for compute, storage or networking. ISO 27001 and SOC 2 Type II certification is in progress, not yet achieved.

Security controls, encryption and compliance status →
Getting started

From workload to running nodes.

Four steps. Provisioning and delivery dates are confirmed in your quote and contract, not assumed.

  1. 01

    Tell us about the workload

    Model size, framework, cluster size and timing. Use the quote form if you know what you need, or talk to our team if you want guidance.

  2. 02

    Receive a quote

    Per-GPU-hour rate for the term, cluster size and start date, with availability confirmed in writing.

  3. 03

    Contract and provisioning

    The reservation is signed, the cluster is commissioned to the agreed delivery date and handed over with bare-metal access.

  4. 04

    Run

    Bare-metal access from day one, with our operations team behind the nodes for as long as you run them.

Also from Sapience Compute for equity → Enterprise Hub (in development) → Platform roadmap →

Blackwell clusters in the EU, reserved on your terms.

Node type, cluster size, term and start date: we reply with a written quote and confirmed availability. Not sure yet? Our engineers will size it with you.

Get a quote Talk to our team