ai-cloud
The Scalable Mirantis Alternative for AI Infrastructure
Looking for an alternative to Mirantis for your GPU cloud? vCluster creates fully isolated tenant clusters in seconds on bare metal, with no VM-based performance overhead.
Trusted by the fastest-growing AI cloud providers
Problem
Why Teams Leave Mirantis
GPU cloud operators outgrow Mirantis before they outgrow the problem.
Slow Cluster Provisioning
Provisioning full physical clusters per tenant adds cost and complexity for GPU cloud operators moving at AI speed.
Weak Tenant Isolation
Namespace-level isolation models share control-plane components across tenants, creating a shared blast-radius risk by design.
Operational Overhead at Scale
Managing many separate physical clusters introduces operational overhead that scales with tenant count: a problem vCluster eliminates by design.
Solution
vCluster: The Mirantis Alternative Built for GPU Scale
vCluster virtualizes the Kubernetes control plane itself, giving every tenant a fully isolated, CNCF-certified cluster as a lightweight pod on shared hardware. Proven across 100K+ GPU nodes and 50+ GPU clouds, it delivers the tenant isolation of separate physical clusters without the cost or complexity.
Built for GPU Scale Where Traditional Platforms Struggle
vCluster delivers the full stack from bare metal provisioning to tenant cluster orchestration to kernel-native workload isolation.
Tenant Isolation
Isolated Tenant Clusters in Seconds
Each tenant gets their own API server, etcd, scheduler, and RBAC running as a lightweight pod on shared hardware. Private Nodes, the production default, deliver dedicated worker nodes with per-tenant CNI and storage for hardware-level isolation. No separate physical clusters needed. Spin up hundreds of isolated tenant environments with near-zero marginal cost.
- Own API server, etcd, and RBAC
- Spins up in seconds, not hours
- Near-zero cost per tenant cluster
Bare Metal
Zero-Touch GPU Server Provisioning
PXE boot, OS install, machine registration, and network configuration happen automatically. vMetal manages the full GPU server lifecycle from rack to production, eliminating manual provisioning steps that slow competing platforms down.
- PXE boot and OS install automated
- Full GPU server lifecycle management
- From rack to production, zero-touch
Hardware Isolation
Dedicated Physical Nodes Per Tenant
Tenants requiring complete hardware separation get fully dedicated physical nodes with their own CNI and CSI. No workloads from other tenants, eliminating noisy-neighbor GPU contention and cross-tenant risk.
- Dedicated nodes, no shared workloads
- Own CNI and CSI per tenant
- Eliminates noisy-neighbor GPU contention
Workload Security
Kernel-Native Isolation, No Hypervisor Tax
vNode wraps each workload in its own secure runtime using seccomp, cgroups, namespaces, and AppArmor. Container breakout protection is enforced with no VM overhead, preserving full bare metal GPU performance.
- Seccomp, cgroups, and AppArmor enforced
- No hypervisor overhead on GPU performance
- Prevents container breakout at kernel level
AI Readiness
Pre-Validated AI Environments Built In
Turn a bare Kubernetes cluster into a production AI platform in minutes, not weeks. Partner integrations with Run:AI, Ray, and Jupyter are validated to work with vCluster tenant isolation out of the box.
- Run:AI (partner), Ray, and Jupyter integrations
- AI platform ready in minutes
- Tested with tenant isolation built in
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
How is vCluster different from Mirantis as a Kubernetes platform?
vCluster virtualizes the Kubernetes control plane itself, running each tenant cluster as a lightweight pod inside a control plane cluster. Every tenant gets their own API server, etcd, RBAC, and CRDs without separate physical infrastructure. For production, Private Nodes, dedicated worker nodes with per-tenant CNI and storage, are the default. The result is strong tenant isolation, seconds-fast provisioning, and near-zero marginal cost per new tenant cluster.
Is vCluster production-ready for GPU cloud workloads?
Yes. vCluster powers 100K+ GPU nodes in production across 50+ GPU clouds and Fortune 500 customers including CoreWeave and Nscale. vCluster is named in the NVIDIA DGX SuperPOD reference architecture. Boost Run launched a managed Kubernetes offering in under 45 days, and Lintasarta launched a GPU cloud in Indonesia in 90 days using vCluster.
Can vCluster run directly on bare metal without an existing Kubernetes cluster?
Yes. vCluster Standalone runs as a single binary directly on bare metal Linux with no external Kubernetes dependency. It replaces k3s, kubeadm, or k0s as the base layer. Combined with vMetal for zero-touch bare metal provisioning, vCluster delivers a complete path from GPU racks to fully managed tenant Kubernetes without intermediate dependencies.
What isolation options are available for GPU cloud tenants?
vCluster offers a flexible isolation spectrum. Private Nodes are the production default, giving tenants fully dedicated physical hardware with their own CNI and CSI. Shared Nodes provide namespace and resource quota boundaries for dev, test, CI/CD, and trusted-team workloads. vNode adds kernel-native workload isolation with container breakout protection at every level, all without hypervisor overhead.
Does vCluster support compliance and air-gapped deployments?
Yes. vCluster supports air-gapped and FIPS-compliant deployments for environments with strict connectivity or regulatory requirements. Each tenant cluster is a CNCF-certified Kubernetes control plane with 100% API compatibility, which supports enterprise compliance workflows. Control plane isolation can also be run in dedicated VMs for additional OS-level separation when required.
How quickly can a GPU cloud provider launch managed Kubernetes with vCluster?
Boost Run launched their managed Kubernetes service in under 45 days. Lintasarta deployed a GPU cloud in Indonesia in 90 days with hundreds of isolated tenant clusters. vCluster's full-stack approach from bare metal provisioning to tenant cluster orchestration, demonstrated by Boost Run's 45-day launch and Lintasarta's 90-day launch, compresses timelines that can otherwise stretch significantly.
Ready to Replace Mirantis at GPU Scale
See how vCluster delivers isolated tenant clusters on bare metal without the overhead.
