ai-cloud
GPU as a Service Built on Tenant Isolation
Sell GPU capacity with a managed Kubernetes experience. vMetal provisions machines, and vCluster gives customers separate tenant clusters with dedicated workers through Private Nodes.
Trusted by the fastest-growing AI cloud providers
Problem
GPU Services Need More Than Hardware
GPU providers stall when they compete on specs alone while customers want cloud-grade managed Kubernetes.
Raw Compute Is Not Enough
Customers expect self-service environments, managed Kubernetes, and cloud-native tooling on top of the GPUs from day one.
Namespace Isolation Is Too Weak
Namespace isolation leaves tenants exposed to platform internals and each other. Separate physical clusters are too expensive to scale.
Building In-House Takes Too Long
Building a GPU cloud platform in-house takes significant engineering investment.
Solution
One Stack From Bare Metal to Tenant Clusters
vMetal provisions infrastructure below your GPU service. vCluster supplies independent Kubernetes tenant environments and dedicated Private Nodes, while self-service templates expose the offerings you support. vCluster Labs software powers 100K+ GPUs.
Everything to Launch Your GPU Cloud
From zero-touch bare metal provisioning to isolated tenant clusters and partner AI environments, one stack covers it.
Bare Metal
Zero-Touch GPU Server Provisioning
vMetal exposes one Machine API above bare metal and VM provisioning drivers. Registered inventory, images and credentials enable repeatable machine lifecycle workflows for GPU service customers. Network automation uses supported integrations such as Netris.
- Stable Machine API
- Bare metal and VM provisioning
- Driver-specific lifecycle automation
Tenant Isolation
One Isolated Tenant Cluster per Customer
Every tenant gets its own Kubernetes control plane running as a lightweight pod. For customer workloads, Private Nodes give each tenant cluster dedicated worker nodes with per-tenant CNI and storage.
- Independent tenant API and data store
- Tenant-scoped scheduling on Private Nodes
- No dedicated control-plane servers per tenant
Workload Security
Kernel-Level Workload Isolation
vNode uses Linux user namespaces and seccomp to restrict workload privileges and system calls. It adds runtime hardening for GPU service customers without a guest kernel or hypervisor, complementing dedicated workers rather than allocating GPUs.
- Linux user namespaces and seccomp
- No guest kernel or hypervisor
- Helps limit workload breakout
AI Workload Integration
Supported AI Workload Templates
Use supported workload templates for GPU service customers, including the NVIDIA Run:AI partner integration. Configure GPU, network and storage prerequisites, and validate the selected template for Private Nodes before offering it to production customers.
- Native NVIDIA Run:AI catalog
- Validate Private Node configuration
- Release-specific integration prerequisites
Customer Experience
Template Based Self-Service for Customers
Give customers an EKS-like self-service portal to provision their own isolated environments on demand, so your GPU as a service offering matches the cloud experience AI teams already know.
- Approved templates and access policies
- Dedicated production tenant workers
- Readiness depends on available capacity
Why vCluster
This isn’t a side project. Behind every vCluster deployment is 5+ years of deep K8s engineering, security hardening, and battle-tested infrastructure work at massive scale.
100K+
GPUs Powered
50+
GPU Clouds & F500s
<45
Days to Launch
30K
GitHub Stars
Get Started in 3 Steps
1
Schedule a Demo
Talk to our team about your stack
2
Deploy vCluster
Deploy vCluster on your infra in minutes
3
Onboard Your Tenants
Go live with a hyperscaler-grade tenant experience in days
FAQs
What is GPU as a service and how does Kubernetes fit in?
Operators expose approved Kubernetes templates through vCluster Platform interfaces or an integrated customer portal. Templates define supported configurations, while access and capacity policies govern requests. vMetal can provision machines and Private Nodes dedicate production workers. Customers receive their own Kubernetes environment once its control plane, networking, storage and worker capacity are ready; application stacks have additional prerequisites.
How fast can I launch a GPU as a service offering with vCluster?
Boost Run went from decision to production launch in under 45 days with vCluster Platform, and Lintasarta runs 170+ tenant clusters in production. Bare metal provisioning, tenant cluster orchestration, and workload isolation come in one stack, so your engineers can work on what makes your offering different instead of rebuilding infrastructure primitives.
How does vCluster isolate tenants on bare metal GPU hardware?
vCluster gives each tenant an independent Kubernetes API, data store and RBAC scope. Private Nodes dedicate worker capacity to that tenant, including its own CNI and storage configuration. vMetal orchestrates the machines underneath. This separates tenant administration and worker placement; shared infrastructure, application permissions and network policies still require an explicit security design.
Does vCluster support dedicated GPU nodes per tenant?
Yes. Private Nodes are vCluster Platform's recommended production model: dedicated worker nodes joined privately into each tenant cluster, each with its own CNI and CSI and no workloads from other tenants. That suits GPU customers who need hardware-level isolation for performance or compliance, without the cost of separate physical clusters. Shared Nodes are also available for your own internal dev, test, and CI/CD.
Is vCluster compatible with existing AI platforms like NVIDIA Run:AI or Ray?
NVIDIA Run:AI is available through the native Certified Stacks integration, with release-specific prerequisites and templates. Its bundled templates do not automatically supply dedicated Private Nodes. Configure and validate the production worker model, GPU access, networking and storage before offering it to customers. Ray and Jupyter can use compatible Kubernetes deployment tooling and require their own supported configuration.
What bare metal GPU infrastructure does vCluster support?
vMetal handles zero-touch provisioning for GPU servers and VMs, including PXE boot, OS installation, machine registration, and network automation. vCluster Standalone runs as a single CNCF-certified binary directly on Linux, with no k3s, kubeadm, or other distribution as a base layer. vCluster is validated in NVIDIA's DGX reference architecture and powers 100K+ GPUs.
Launch Your GPU as a Service Offering
See how vCluster powers GPU as a service for 50+ GPU clouds & Fortune 500s.
