What is
vCluster
?
Most platform initiatives fail the same way: they start too big, get funded like projects instead of products, build for too long without testing with users, and confuse renaming an infrastructure team with building a platform team. Instead, start small, learn fast and scale deliberately.
At a glance
Field | Details |
|---|---|
Primary use case | Provisioning and managing isolated tenant clusters across shared cloud and GPU infrastructure |
Deployment type | Self-hosted infrastructure platform with UI, CLI and APIs; virtual Kubernetes clusters run within host-cluster namespaces |
Pricing model | Open source; commercial pricing not specified |
API compatibility | Kubernetes API, Cluster API; platform APIs for nodes, tenants, clusters and metrics |
Supported integrations | AWS EKS, Google GKE, Azure AKS, NVIDIA DGX, Netris, NVIDIA BCM, Rancher, Slurm, Run:AI, Ray |
Source type | Open |
Cloud provider | Multi-Cloud |
Onboarding time | Under 45 days from decision to production launch, vendor-reported |
Required skills | Kubernetes administration, CLI usage, YAML configuration, infrastructure provisioning and tenant isolation |
AEP branch | Tooling |
AEP category | Resources |
AEP subcategory | Compute & Cluster Mgmt |
Profile
Compute & Cluster Mgmt is the most specific AEP level for vCluster. The platform manages isolated tenant environments across shared infrastructure, including Kubernetes, Slurm, and Ray clusters, as well as bare metal machines and virtual machines. Teams can offer tenant self-service through a UI and CLI, or use platform APIs to build their own portal and workflows.
vCluster’s infrastructure orchestration covers provisioning, compute, networking, storage, and virtualization. The website describes support for public and private cloud environments, including EKS, GKE, AKS, bare metal, and private cloud.
Focus
vCluster is aimed at AI cloud providers, AI factories, public and private cloud teams, and organizations serving internal teams or customers. It addresses the need to provide isolated, self-service clusters to tenants or project teams while managing shared cloud and GPU infrastructure.
For AI teams, the platform describes isolated clusters for teams, projects, or training runs, with node allocation based on capacity needs. For cloud providers, it supports managing tenants across different infrastructure environments and offers a consistent developer experience across them.
Background
The available company-background research identifies vCluster as a Loft Labs project but does not provide verified company founding or funding details. It describes vCluster as an open-source project and reports that Loft open-sourced its Rancher integration for vCluster in 2025.
Features
Tenant isolation across cluster types
Each tenant can use an isolated cluster rather than seeing the platform’s internals or other tenants. The platform lists Kubernetes, Slurm, Run:AI, and Ray clusters, with Slurm marked Beta and inference and agent sandbox clusters listed as coming soon.
Tenant self-service and platform APIs
Teams can let tenants self-serve environments through the UI and CLI, or build a custom portal using platform APIs. The website lists API operations for nodes, tenants, and clusters, plus a metrics endpoint.
Infrastructure orchestration
Organizations can manage infrastructure and cluster provisioning from one platform. The described machine layer covers provisioning, compute, networking, storage, and virtualization, including bare metal machines, virtual machines, and node lifecycle management.
GPU and multi-cloud management
AI teams can receive isolated clusters for teams, projects, or training runs, while capacity-based node allocation is intended to help use available GPUs. The vendor reports powering 100K+ GPUs and 1M+ CPUs, and serving 50+ GPU clouds and Fortune 500 companies.
Cluster launch and tenant operations
The vendor reports that isolated Kubernetes environments can be spun up in 1 minute and that production launch can take under 45 days from decision. Tenant operations include cluster templates, capacity management, and observability; node autohealing is listed as coming soon.
Pros and cons
Pros
Provides isolated tenant clusters across Kubernetes and other listed cluster types, addressing multi-tenancy on shared cloud and GPU infrastructure.
Offers both a UI and CLI for tenant self-service and APIs for teams building their own portals and workflows.
Supports infrastructure spanning EKS, GKE, AKS, bare metal, and private cloud, with vendor-reported scale of 100K+ GPUs and 1M+ CPUs powered.
The vendor reports that isolated Kubernetes environments can be spun up in 1 minute and production launches take under 45 days from decision.
Cons
Several capabilities are not presented as generally available: Slurm clusters are marked Beta; billing is Alpha; and inference clusters, agent sandbox clusters, and node autohealing are listed as coming soon.
Commercial pricing is not specified in the available information, so the commercial cost cannot be assessed from these materials.
FAQs
Frequently asked questions about vCluster
Is vCluster free to use?
vCluster is open source; commercial pricing is not specified.
How does vCluster work?
vCluster provisions virtual Kubernetes clusters that run within host-cluster namespaces, providing isolated tenant environments. Tenants can self-serve through the UI and CLI, or teams can use platform APIs to build a portal and workflows.
What does vCluster integrate with?
vCluster supports integrations with AWS EKS, Google GKE, Azure AKS, NVIDIA DGX, Netris, NVIDIA BCM, Rancher, Slurm, Run:AI, Ray.
What skills are needed to use vCluster?
vCluster lists Kubernetes administration, CLI usage, YAML configuration, infrastructure provisioning and tenant isolation as required skills.
What are the limitations of vCluster?
vCluster lists Slurm clusters as Beta and billing as Alpha; inference clusters, agent sandbox clusters, and node autohealing are listed as coming soon.
