ai-cloud
AI Cloud Provider Platform for Secure Tenant Clusters
vCluster gives AI cloud providers fully isolated, CNCF-certified tenant clusters on shared GPU infrastructure with near-zero overhead and no physical cluster sprawl.
Trusted by the fastest-growing AI cloud providers
Problem
The AI Cloud Provider Dilemma
Selling raw GPU compute is not enough. Your customers expect more.
Race to the Bottom
Selling bare metal GPUs alone is a race to the bottom. Converging GPU specs mean margins erode unless you offer managed Kubernetes.
Isolation vs. Efficiency
Namespace isolation is too weak. Separate physical clusters are too expensive. Standard Kubernetes forces you to choose.
DIY Takes Too Long
Building a GPU cloud platform in-house requires significant engineering investment.
Solution
One Stack from Bare Metal to Tenant Clusters
vCluster delivers the complete path from GPU racks to managed Kubernetes. Every tenant gets a fully isolated, CNCF-certified cluster with their own API server, etcd, and RBAC, running as a lightweight process on shared hardware. For production deployments, Private Nodes, dedicated worker nodes with per-tenant CNI and storage, deliver hardware-level isolation. Boost Run launched in under 45 days. Lintasarta in 90 days.
Built for AI Cloud Providers at GPU Scale
From bare metal provisioning to kernel-native workload isolation, vCluster covers every layer your AI cloud platform requires.
Tenant Isolation
Isolated Tenant Clusters on Shared GPU Infrastructure
Each tenant gets a real Kubernetes API server, etcd, and scheduler running as a lightweight pod on your control plane cluster. Spin up hundreds of tenant environments without provisioning separate physical clusters.
- Own API server and etcd per tenant
- Spins up in seconds not hours
- Near-zero overhead per tenant cluster
Bare Metal Layer
Zero-Touch GPU Server Provisioning
vMetal handles PXE boot, OS installation, machine registration, and network automation across your GPU fleet. Go from rack to production-ready Kubernetes nodes without manual intervention.
- PXE boot and OS install automated
- Full machine lifecycle management
- Network automation via Netris integration
Workload Security
Kernel-Native Isolation Without VM Overhead
vNode delivers container breakout protection using seccomp, cgroups, namespaces, and AppArmor per workload, preserving bare metal GPU performance with no hypervisor tax.
- No hypervisor overhead on GPUs
- Prevents container breakout attacks
- Defense-in-depth with control plane isolation
AI Platform Readiness
Pre-Validated AI Environments for Tenants
Turn a bare Kubernetes cluster into a production AI platform in minutes. Partner integrations with Run:AI, Ray, and Jupyter let you offer managed AI tooling to tenants without custom configuration work.
- Partner integrations with Run:AI, Ray, and Jupyter
- Cluster to AI platform in minutes
- Tested against tenant isolation layer
Tenant Experience
EKS-Like Portal for Your Customers
Give tenants an AWS-grade self-service experience. Your customers provision their own Kubernetes environments through a portal, without filing support tickets or waiting on your platform team.
- Self-service cluster provisioning
- EKS and GKE comparable experience
- Reduces platform team operational load
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+
GPU Nodes 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 makes vCluster different from namespace-level tenant isolation?
Namespace isolation gives each tenant a partition of the same Kubernetes control plane, meaning they share the API server, etcd, and cluster-wide resources. vCluster gives every tenant their own dedicated API server, etcd, scheduler, and RBAC, running as a lightweight process on your control plane cluster. This means a tenant misconfiguration or compromise cannot affect other tenants, and each tenant gets full cluster-admin access without risk to the shared layer.
How quickly can an AI cloud provider launch a managed Kubernetes offering?
Boost Run launched its managed Kubernetes service in under 45 days using vCluster. Lintasarta launched a GPU cloud in Indonesia in 90 days, running hundreds of tenant clusters. The full stack, from bare metal provisioning through tenant cluster orchestration, is designed to get AI cloud providers to revenue without a lengthy multi-quarter build.
Does vCluster work directly on bare metal GPU servers?
Yes. vCluster Standalone runs as a single binary directly on Linux, with no dependency on k3s, kubeadm, or any external Kubernetes distribution. vMetal handles PXE boot, OS installation, and network automation for your GPU servers, so the entire path from physical rack to tenant-ready Kubernetes clusters is covered in one integrated stack.
How does tenant isolation work across GPU nodes?
vCluster offers a flexible isolation spectrum. Private Nodes are the production default, giving a tenant fully dedicated physical nodes with their own CNI and CSI. Shared Nodes allow multiple tenants on the same physical hardware with resource quota boundaries, scoped to dev, test, CI/CD, and trusted-team workloads. At the workload layer, vNode prevents container breakout and limits blast radius using seccomp, cgroups, namespaces, and AppArmor, preserving near-bare-metal performance without hypervisor overhead.
Is vCluster used by other AI cloud providers in production?
Yes. vCluster powers over 100,000 GPU nodes in production across more than 50 GPU clouds and Fortune 500 customers, including CoreWeave and Nscale. vCluster is named in the NVIDIA DGX SuperPOD reference architecture.
Can vCluster support compliance requirements for AI cloud tenants?
vCluster supports air-gapped and FIPS deployments for environments with strict compliance requirements. Tenant control planes can run as VMs rather than pods to achieve OS-level kernel separation. vNode prevents container breakout and limits blast radius with seccomp, AppArmor, and cgroup enforcement. Network isolation is enforced via hardware-level VLANs, VXLANs, VRFs, and ACLs through the Netris integration, giving each tenant a fully segmented network boundary.
Launch Your AI Cloud Platform Faster
See how AI cloud providers go from GPU racks to tenant-ready Kubernetes in weeks.
