AI infrastructure as a service
GPU capacity at your premises, on an annual rental plan with operations and SLA included. A minimum term of 12 months, paid annually in advance.
Capacity and scope
Annual payment in advance · 12-month minimum · Prices exclude VAT
AI Infrastructure as a Service — Inference S
Monthly equivalent: 2.290 € · not a monthly payment
- Delivery and installation on your premises (mainland Spain)
- Operating system, Kubernetes, NVIDIA GPU Operator and open-source inference stack
- 24x7 monitoring with SPEM Observe
- Free replacement of failed hardware
- Security patching in an agreed window
- Monthly usage and availability report
- Pre-installation site survey
- Certified secure erase on return
36-month terms and additional SLA options require a proposal.
| Specification | S | M | L |
|---|---|---|---|
| Annual price excl. VAT | 27.480 € | 47.880 € | 99.480 € |
| GPU | 2x NVIDIA RTX PRO 6000 Blackwell Server 96 GB (192 GB) | 4x NVIDIA RTX PRO 6000 Blackwell Server 96 GB (384 GB) | 8x NVIDIA RTX PRO 6000 Blackwell Server 96 GB (768 GB) en 2 nodos |
| CPU / RAM | 2x Intel Xeon 6, 32 núcleos / 512 GB DDR5 | 2x Intel Xeon 6, 48 núcleos / 1 TB DDR5 | 4x Intel Xeon 6, 48 núcleos / 2 TB DDR5 |
| Storage | 15 TB NVMe | 30 TB NVMe | 60 TB NVMe |
| Network | 2x 25 GbE | 2x 100 GbE | Switch 100 GbE 32 puertos incluido |
| SLA | 8x5, respuesta siguiente día hábil, 99,5 % | 8x5, respuesta siguiente día hábil, 99,5 % | 24x7, respuesta 4 h, 99,9 % |
| Site requirements | 2U, 2 kW redundantes | 4U, 3,5 kW redundantes | 9U, 7 kW redundantes |
The layers of your infrastructure
Compute
GPU nodes for inference and training. GPU count, memory and form factor depend on the selected configuration.
A service that covers the whole lifecycle
Hardware at your facility
Configured capacity, defined support and direct contact with your technical team.
Operations and monitoring
Configured capacity, defined support and direct contact with your technical team.
Support with SLA
Configured capacity, defined support and direct contact with your technical team.
Capacity that scales
Configured capacity, defined support and direct contact with your technical team.
Colocation, private cloud, Premium SLA and 36-month terms are available on request. HBM XL clusters require a dedicated proposal.
Frequently asked questions
Where does the infrastructure run?
At your premises. Colocation or dedicated private cloud in Spain can also be arranged.
What is the minimum term?
12 months, with the annual plan paid in advance.
What does the price include?
Hardware rental, deployment, operations and the SLA specified for the package. Prices exclude VAT.
Can capacity be increased?
Yes. We assess the workload and provide a proposal before changing your capacity.
Tell us what you want to run. We will tell you what you need.
A concrete response within 24 business hours.