Company Profile
Provider TypeGPU Nodes for AI, ML and HPC
Legal EntityNscale Limited
Infrastructure
GPU FleetNVIDIA DGX Vera Rubin NVL72, GB300 NVL72, H100, H200, and GB200 GPUs
Bare MetalBare metal GPU nodes
enterprises, governments, and mission-critical workloads
Compute & Deployment
Bare Metalbare-metal infrastructure
GPU Hardware
Latest GenNVIDIA DGX Vera Rubin NVL72, GB300 NVL72, H100, H200, and GB200 GPUs
Multi-GPU NodesConnected as one massive GPU fabric for rack-scale performance.
NVLinkUnified by sixth-generation NVLink
Liquid Coolingclosed-loop direct-liquid cooling and rear-door heat-exchangers
Pricing Model
Performance & Scaling
Multi-Node TrainingEngineered for frontier-scale model training and post-training.
Elastic ScalingThis means you can add, remove, scale up, scale down your compute as your needs change.
Developer Experience
Security & Compliance
Uses industry NVIDIA acceleratorsstates specific data center projects and certifications (e.g., state-certified AI Microgrid)partnership on at least one site
Data Center Locations
Coverage
CountriesNorway, U.K., Iceland, Portugal, U.S.
CitiesOslo, Norway, Stavanger, Norway, Narvik, Norway, Glomfjord, Norway, Hayes, U.K., Loughton, U.K., Keflavik, Iceland, Blönduós, Iceland, Sines, Portugal, West Virginia, U.S., North Carolina, U.S., Texas, U.S.
across the world
Compliance Regions
Datacenter Locations
Key Strengths
Bare-metal GPU nodes for raw performance
global network of AI data centers optimized for training and inference
emphasis on low-cost renewable power, modular prefabricated design, and high-density AI infrastructure
Known Limitations
Orchestration, scheduling, or application services are not included by default and are to be layered on later
Additional Information
Green Energy
Some data centers powered by 100% renewable energy
Core Proposition
We built our GPU Nodes service for those who want the raw performance of bare metal GPUs.
Last updated August 2026. Information subject to change.