Full-stack AI cloud pricing
The best value per GPU-hour across the full AI lifecycle — with the reliability, performance, and support to match
- Full-stack, AI Lab-backed
- Pay-as-you-go, spot, or reserved
- SOC 2 and GDPR compliant
AI Cloud
From rapid prototyping to foundation training to scalable inference
Compute
GB300 NVL72
New1x tray to 2+ racks · NVLink v5
The best value per GPU-hour across the full AI lifecycle — with the reliability, performance, and support to match
1x, 2x, 4x, or 8x NVIDIA GPU instances. NVLink models available.
Pay-as-you-go pricing. Start and end deployments at any time, without long-term commitments.
| Hardware | CPUs | RAM | GPU VRAM | Price | Spot price |
|---|---|---|---|---|---|
| 1x GB300 SXM6 288GB | 32 | 225 GB | 288 GB | $12.55/h | $6.273/h |
| 1x B300 SXM6 268GB | 30 | 255 GB | 268 GB | $10.91/h | $5.455/h |
| 1x B200 SXM6 180GB | 30 | 170 GB | 180 GB | $8.523/h | $4.262/h |
| 1x H200 SXM5 141GB | 44 | 170 GB | 141 GB | $6.006/h | $3.003/h |
| 1x H100 SXM5 80GB | 30 | 120 GB | 80 GB | $4.606/h | $2.303/h |
| 1x H100 SXM5 80GB | 32 | 170 GB | 80 GB | $4.606/h | $2.303/h |
| 1x A100 SXM4 80GB | 22 | 120 GB | 80 GB | $2.160/h | $1.080/h |
| 1x A100 SXM4 40GB | 22 | 120 GB | 40 GB | $1.577/h | $0.7885/h |
| 1x Tesla V100 16GB | 6 | 23 GB | 16 GB | $0.2320/h | $0.1160/h |
| 1x RTX PRO 6000 96GB | 30 | 90 GB | 96 GB | $2.542/h | $1.271/h |
| 1x L40S 48GB | 20 | 60 GB | 48 GB | $1.827/h | $0.9135/h |
| 1x RTX 6000 Ada 48GB | 10 | 60 GB | 48 GB | $1.439/h | $0.7195/h |
| 1x RTX A6000 48GB | 10 | 60 GB | 48 GB | $0.7960/h | $0.3980/h |
Confidential computing on NVIDIA Blackwell GPUs, with your data and model staying encrypted in memory, in transit, and at rest. Additional configurations are available upon request.
| Hardware | CPUs | RAM | GPU VRAM | Price | Spot price |
|---|---|---|---|---|---|
| 1x B300 CC SXM6 268GB | 30 | 255 GB | 268 GB | $11.13/h | $5.565/h |
| 1x B200 CC SXM6 180GB | 30 | 170 GB | 180 GB | $8.695/h | $4.347/h |
| 1x RTX PRO 6000 CC 96GB | 30 | 90 GB | 96 GB | $2.593/h | $1.296/h |
Long-term discounts on self-service deployments. Custom quotes are available on request.
| Commitment | Discount |
|---|---|
| 1 month | 2% off on-demand |
| 3 months | 3% off on-demand |
| 6 months | 4% off on-demand |
| 1 year | 8% off on-demand |
| 2 years | 25% off on-demand |
The latest, most-advanced CPUs on the market — with fast and flexible access.
Pay-as-you-go pricing. Start and end deployments at any time, without long-term commitments.
| Hardware | CPUs | RAM | Price | Spot price |
|---|---|---|---|---|
| CPU-TURIN.4V.16G | 4 | 16 GB | $0.04800/h | $0.01920/h |
| CPU-TURIN.8V.32G | 8 | 32 GB | $0.09600/h | $0.03840/h |
| CPU-TURIN.16V.64G | 16 | 64 GB | $0.1920/h | $0.07680/h |
| CPU-TURIN.32V.128G | 32 | 128 GB | $0.3840/h | $0.1536/h |
| CPU-TURIN.64V.256G | 64 | 256 GB | $0.7680/h | $0.3072/h |
| CPU-TURIN.96V.384G | 96 | 384 GB | $1.152/h | $0.4608/h |
| CPU-TURIN.120V.480G | 120 | 480 GB | $1.440/h | $0.5760/h |
| CPU-TURIN.180V.720G | 180 | 720 GB | $2.160/h | $0.8640/h |
| CPU-TURIN.360V.1440G | 360 | 1440 GB | $4.320/h | $1.728/h |
| CPU.4V.16G | 4 | 16 GB | $0.04800/h | $0.01920/h |
| CPU.8V.32G | 8 | 32 GB | $0.09600/h | $0.03840/h |
| CPU.16V.64G | 16 | 64 GB | $0.1920/h | $0.07680/h |
| CPU.32V.128G | 32 | 128 GB | $0.3840/h | $0.1536/h |
| CPU.64V.256G | 64 | 256 GB | $0.7680/h | $0.3072/h |
| CPU.96V.384G | 96 | 384 GB | $1.152/h | $0.4608/h |
| CPU.120V.480G | 120 | 480 GB | $1.440/h | $0.5760/h |
| CPU.180V.720G | 180 | 720 GB | $2.160/h | $0.8640/h |
| CPU.360V.1440G | 360 | 1440 GB | $4.320/h | $1.728/h |
Confidential computing is also available for CPU workloads, powered by AMD EPYC processors.
| Hardware | CPUs | RAM | Price | Spot price |
|---|---|---|---|---|
| CPU.4V.16G.CC | 4 | 16 GB | $0.04896/h | $0.01958/h |
| CPU.8V.32G.CC | 8 | 32 GB | $0.09792/h | $0.03917/h |
| CPU.16V.64G.CC | 16 | 64 GB | $0.1958/h | $0.07834/h |
| CPU.32V.128G.CC | 32 | 128 GB | $0.3917/h | $0.1567/h |
| CPU.64V.256G.CC | 64 | 256 GB | $0.7834/h | $0.3133/h |
| CPU.96V.384G.CC | 96 | 384 GB | $1.175/h | $0.4700/h |
| CPU.120V.480G.CC | 120 | 480 GB | $1.469/h | $0.5875/h |
| CPU.180V.720G.CC | 180 | 720 GB | $2.203/h | $0.8813/h |
| CPU.360V.1440G.CC | 360 | 1440 GB | $4.406/h | $1.763/h |
Multi-node clusters with InfiniBand interconnect and up to 144x GPUs in self-service access - larger clusters with custom configurations are available on request. Provision and manage with UI console, Terraform, SDK, or API.
True pay-as-you-go pricing on multi-node clusters. Spin up for an hour or months — no commitments, no minimums. Specs and price are shown per GPU.
| Hardware | CPUs | RAM | GPU VRAM | Price (1x GPU) |
|---|---|---|---|---|
| GB300 SXM6 288GB | 32 | 225 GB | 288 GB | $12.55/h |
| B300 SXM6 268GB | 30 | 255 GB | 268 GB | $10.91/h |
| B200 SXM6 180GB | 30 | 170 GB | 180 GB | $8.525/h |
| H200 SXM5 141GB | 22 | 170 GB | 141 GB | $6.006/h |
Discounted capacity starting from 1 month. Get a custom quote.
Talk to salesAuto-scaling containers with pay-per-use and scale-to-zero. Suitable for continuous deployments and batch jobs.
| Hardware | CPUs | RAM | GPU VRAM | Price | Spot price |
|---|---|---|---|---|---|
| 1x B300 SXM6 268GB | 28 | 250 GB | 268 GB | $12.00/h | $6.000/h |
| 1x B200 SXM6 180GB | 28 | 165 GB | 180 GB | $9.380/h | $4.690/h |
| 1x H200 SXM5 141GB | 21 | 160 GB | 141 GB | $6.610/h | $3.300/h |
| 1x H100 SXM5 80GB | 21 | 160 GB | 80 GB | $5.070/h | $2.530/h |
| 1x RTX PRO 6000 96GB | 28 | 85 GB | 96 GB | $2.800/h | $1.400/h |
| 1x L40S 48GB | 20 | 58 GB | 48 GB | $2.010/h | $1.000/h |
| AMD Rome+ | 8 | 32 GB | — | $0.1100/h | $0.04000/h |
Flexible high-speed, low-latency storage solutions — across the full AI lifecycle. Rates are shown per GiB-month.
| Storage type | Price / GiB / month | Price / TiB / month | IOPS | Throughput |
|---|---|---|---|---|
| NVMe | $0.2000 | $200.0 | 100k | 4 GB/s |
| NVMe (shared filesystem) | $0.2000 | $200.0 | 100k | 2 GB/s |
| Container registry | $0.2000 | $200.0 | — | — |
Built in Europe, trusted globally
From rapid prototyping to foundation training and scalable inference — on a single full-stack AI cloud