GPU rental · EU regions
This is the hardware behind Managed GPU and GPU VM — every machine named, owned by AxForge and hosted in the EU. Rent a DGX Spark today, request an RTX tier, or request capacity for the hardware you need next — we buy against it.
Fleet
| System | Availability | Location | Memory | Price | |
|---|---|---|---|---|---|
| NVIDIA DGX Spark (GB10) ×8 | Available | Sweden (eu-se-1) | 128 GB unified each | €495 / month | Reserve |
| NVIDIA DGX Spark (GB10) ×3 | On request · days | — | 128 GB unified each | €495 / month | Request capacity |
| NVIDIA RTX 6000 Pro | On request | — | 96 GB GDDR7 | €1.73 / GPU-hr | Request capacity |
| NVIDIA RTX 5090 ×4 (system) | On request | — | 32 GB GDDR7 each | €0.54 / GPU-hr | Request capacity |
| NVIDIA RTX 3090 ×4 (budget tier) | On request | — | 24 GB GDDR6X each | €0.17 / GPU-hr | Request capacity |
| NVIDIA RTX 3060 ×2 ×4 (starter tier) | Available | Sweden (eu-se-1) | 12 GB GDDR6 each | €0.10 / hour | Reserve |
| Datacenter-scale capacity | Coming | — | — | — | Register interest |
| NVIDIA H100 | On request | — | 80 GB HBM3 | — | Request capacity |
| NVIDIA H200 | On request | — | 141 GB HBM3e | — | Request capacity |
| NVIDIA B200 | Coming | — | 192 GB HBM3e | — | Register interest |
Every system is available as Managed GPU (we deploy and operate) or GPU VM (SSH, your stack). Launch pricing, billed monthly for dedicated systems — an engineer scopes each rental with you. Demand-page GPUs carry no price until the hardware lands.
Honest split
Rentable today
DGX Spark — 8 units live in Sweden (eu-se-1), monthly rental, 128 GB unified memory each, with 7 more units arriving in days. This is the hardware behind our published measured inference numbers.
On request
The RTX tiers — RTX 6000 Pro, RTX 5090 ×4 and RTX 3090 ×4 — are on the way. They can be reserved now; an engineer confirms timing and placement with you.
On request
H100, H200 and B200 are not in the fleet. We buy hardware against registered demand — tell us what you need and we build the business case with you. That is how the Spark fleet grew.
Rather have tokens?
If you want an API rather than a machine, the shared OpenAI-compatible API serves Qwen3.8 27B and more from Sweden — see all models.
Data & privacy
A rented system serves only you: your model, your traffic, our hardware — prompts never persisted. Nothing is logged, retained or used for training; only request metadata (token counts, timestamps, status) is kept for billing and operations. Full policy at axforge.ai/privacy.
FAQ
NVIDIA DGX Spark (GB10, 128 GB unified memory) — 8 units available now in Sweden (eu-se-1), monthly rental, with 3 more arriving in days. RTX 6000 Pro, RTX 5090 and RTX 3090 systems are on the way and can be reserved.
DGX Spark is €495/month; RTX 6000 Pro €1.73, RTX 5090 €0.54 and RTX 3090 €0.17 per GPU-hour — launch pricing, billed monthly for dedicated systems. An engineer scopes the configuration with you; H100, H200 and B200 carry no price until the hardware lands. Talk to an engineer.
Not today — none of these are in the AxForge fleet. We buy hardware against registered demand: tell us what you need and we build the business case with you. That is how the DGX Spark fleet grew. See H100, H200, B200.
The hardware available now runs in Sweden (eu-se-1, Stockholm). AxForge operates its platform from Stockholm and Málaga, with more EU regions in deployment.
Dedicated. A rented system serves only your traffic — your model, your traffic, our hardware. The shared pay-per-token API is a separate product under /models/.
No. Prompts and completions are processed in memory and never persisted, logged or used for training. Only request metadata (token counts, timestamps, status) is kept for billing and operations — see the privacy policy.
Related