Cloud GPU
Cutting-edge Cloud GPU solutions
NVIDIA H100, H200, GH200, A100, L40S, A40 and A16 — on demand as virtual machines or bare metal, for AI and machine learning, HPC, AR/VR, and virtual desktop workloads.
DeploymentVM or bare metal
BillingOn-demand, no minimum
ScalingMulti-GPU · IB on request
Preemptiblelower cost
NVIDIA H100 & H200AVAILABLE
HBM3 · NVLink
Training large models, where per-GPU throughput decides the schedule.
VM or bare metal
Configured to order
Scratch storage included
Quote· on-demand or preemptible
Request availabilityNVIDIA A100 & GH200POPULAR
HBM2e / HBM3
Inference, HPC and data analytics with strong performance per watt.
VM or bare metal
Configured to order
Scratch storage included
Quote· on-demand or preemptible
Request availabilityNVIDIA L40S, A40 & A16AVAILABLE
48 GB / 48 GB / 64 GB
Visualisation, rendering, virtual workstations and VDI at density.
VM or bare metal
Configured to order
Scratch storage included
Quote· on-demand or preemptible
Request availabilityDeployment options by GPU
GPU
Virtual machine
Bare metal
Multi-GPU
NVIDIA H100 / H200
Available
Available
Available
NVIDIA GH200
Available
Available
Limited
NVIDIA A100
Available
Available
Available
NVIDIA L40S / A40
Available
Available
Available
NVIDIA A16
Limited
Available
Available
What people build on it
AI and machine learning
Training and fine-tuning runs where wall-clock time is the constraint, plus low-latency inference serving.
High-performance computing
Simulation, computational chemistry and financial modelling across multi-GPU nodes.
Rendering and visualisation
Offline render farms, real-time visualisation and AR/VR pipelines on L40S and A40.
Virtual desktops
VDI and MIG partitioning on A16 for dense, GPU-accelerated desktop estates.
Tell us the workload, we will size it
Multi-GPU configurations, InfiniBand and managed GPU services are all available on request.