Vendor Sheet
GPU as a Service with Spectro Cloud and NVIDIA
Spectro Cloud and NVIDIA provide a Kubernetes-based GPU-as-a-Service model that gives application and data science teams on-demand access to GPU infrastructure without requiring deep infrastructure expertise. Palette automates complete environments spanning the OS, Kubernetes, NVIDIA GPU Operator, schedulers such as Kai, Kueue, and Armada, NIM microservices, NeMo, and BlueField with DOCA support. Declarative Cluster Profiles create consistent GPU stacks across cloud, data center, and edge while lifecycle automation manages upgrades, monitoring, policies, and scaling. A technical-preview DPU capability can offload networking, storage, and security functions. The solution supports model training, inference, RAG, agentic AI, regulated workloads, and disconnected edge deployments while improvi
