ai-cloud

Zero-Touch Bare Metal Provisioning for AI Clouds

vMetal turns a Machine request into a provisioned physical server through a stable API and infrastructure driver. Provisioned Machines can register with the target tenant cluster automatically, and released capacity returns to inventory.

Trusted by the fastest-growing AI cloud providers
Problem

Manual Provisioning Kills GPU Revenue

Manual bare metal provisioning creates inconsistent server state and slows capacity delivery.

Provisioning Delays Cost Millions

Unprovisioned GPU servers cannot serve tenant workloads or be offered as available capacity.

Human Error Breaks Deployments

Manual PXE, OS, and network steps introduce variation between servers and data centers.

No Repeatable Path from Rack to Product

Without a stable Machine model, each new infrastructure driver requires different automation and lifecycle handling.

Solution

One API for Zero Touch Bare Metal

vMetal provides one stable Machine API above infrastructure-specific drivers. With Metal3, it automates provisioning and can join completed servers to tenant clusters without a manual kubeadm step.

Built for GPU Clouds That Cannot Wait

vMetal automates the repeatable Machine lifecycle from request through provisioning, cluster attachment, release, and reclaim.

Core Provisioning

Automated Bare Metal Provisioning

With Metal3, vMetal provisions registered servers, installs the selected OS image, applies network configuration, and runs startup automation.

  • PXE boot to production automatically
  • Unified lifecycle from rack to decommission
  • Standardized onboarding per supplier
Dynamic Scaling

Machine Requests on Demand

Auto Nodes can create and delete Machine requests to maintain Private Node capacity for tenant clusters.

  • Provisions bare metal servers automatically
  • Configured capacity management
  • Scales bare metal with workload demand
Network Automation

Optional Hardware Network Isolation

Metal3 and Netris can place separate tenant network environments on hardware-backed L2 boundaries when configured.

  • VLANs and VRFs provisioned automatically
  • Network automation per tenant
  • No manual network configuration required
Cluster Layer

Bare Metal for Tenant Clusters

Provisioned Machines can join a tenant cluster automatically as Private Nodes. vCluster remains responsible for the tenant cluster itself.

  • No external K8s dependency required
  • Runs as binary on bare metal
  • CNCF-certified from day one
Operational Control

Machine Inventory and Lifecycle

vMetal tracks Machine claims, server inventory, provisioning state, deletion, cleaning, and return to the available capacity pool.

  • Single pane for full fleet
  • Quotas and capacity management built in
  • CLI, UI, and API access

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 does zero touch bare metal provisioning mean in practice?

Zero-touch bare metal provisioning automates the repeatable steps after servers and credentials have been registered. With Metal3, vMetal drives provisioning, OS installation, network configuration, and startup automation through the Machine lifecycle.

What provisioning stacks does vMetal support?

vMetal uses interchangeable drivers behind one Machine API. Current documentation lists Metal3, KubeVirt, BCM for existing installations, Terraform or OpenTofu, and Cluster API mechanisms. Driver capabilities and availability should be checked in their documentation.

What sellable products does vMetal produce?

vMetal provisions Bare Metal Machines and Virtual Machines. vCluster can then consume those Machines as Private Nodes for managed tenant cluster products.

How does vMetal connect to Kubernetes cluster delivery?

vMetal supplies the Machine. vCluster supplies the tenant cluster and requests or releases capacity through Private Nodes and Auto Nodes. vCluster Standalone is a separate deployment option for running Kubernetes directly on Linux.

How quickly can a GPU cloud launch using vMetal?

Provisioning time depends on the hardware state, driver, OS image, network setup, and startup automation. vMetal standardizes the workflow and lifecycle model across those inputs.

Does vMetal support network automation as part of provisioning?

Network configuration depends on the selected driver and integration. With Metal3 and Netris configured, separate network environments can provide hardware-backed L2 isolation for tenant infrastructure.

Turn Your GPU Racks Into Revenue

See how vMetal automates bare metal provisioning from rack to production-ready inventory.