AI & GPU

NVIDIA GPUs for training, fine-tuning and inference, from a single L4 to multi-node H100 clusters on a fast interconnect.

Plans

Prices exclude VAT. Every plan includes DDoS filtering, IPv4 and IPv6, and 24/7 support.

€349/mo

€899/mo

€1,490/mo

H100

Compare plans

Plan	GPU	CPU	Memory	Storage	Port	Price
L4 NX-G1	1 × NVIDIA L4 24 GB	8 vCPU	64 GB	500 GB NVMe	10 Gbps	€349/mo
L40S NX-G2	1 × NVIDIA L40S 48 GB	16 vCPU	128 GB	1 TB NVMe	10 Gbps	€899/mo
A100 NX-G4	1 × NVIDIA A100 80 GB	24 vCPU	192 GB	2 TB NVMe	25 Gbps	€1,490/mo
H100 NX-G8	1 × NVIDIA H100 80 GB	32 vCPU	256 GB	4 TB NVMe	25 Gbps	€2,190/mo

Ready ML images

Ubuntu with CUDA, cuDNN, PyTorch and Docker with the NVIDIA runtime, ready to run.
Fast local storage

NVMe on every node plus S3 storage in the same data center for datasets and checkpoints.
AI colocation

Bring your own GPU servers: high-density racks with liquid cooling and power sized for your project.
Engineers who know ML

Help with drivers, NCCL, multi-GPU training and serving models in production.

Didn't find your question? Ask an engineer

Can I rent GPUs by the hour?

Monthly plans are listed here. Hourly GPU cloud is available on request for short training runs.

Which GPU should I choose?

L4 suits inference and small models, L40S handles fine-tuning and rendering, A100 and H100 are for training large models.

Do you offer multi-GPU servers?

Yes, servers with 2, 4 and 8 GPUs and NVLink are available. Ask us for a quote.

Where is my data stored?

On the server's local disks and in our own data center. We do not move or inspect customer data.

Tell us about your workload. We will put together a configuration and a quote within one business day.