Innovation & Tech

vCluster v5.0 gives AI clouds self-service bare-metal GPU servers

vCluster Labs put Platform v5.0 into limited availability on November 19, adding self-service bare-metal GPU provisioning while operators keep hardware ownership and capacity.

By Grace Kim

3 min read

VCluster Adds Self-Service Private Clouds for Dedicated AI Hardware
VCluster Adds Self-Service Private Clouds for Dedicated AI HardwareAkela999 / pixabay

What's News

  • vCluster Platform v5.0 shipped November 19 in limited availability, adding self-service provisioning of bare-metal GPU servers.
  • Version 5.0 is not generally available; vendor documentation recommends keeping Platform v4.12 and vCluster v0.37 for current deployments.
  • vMetal now maintains physical servers in an inventory, with tenant-level quotas limiting how much capacity a customer can claim.
  • Providers retain hardware ownership and capacity allocation; customers select an OS image and launch Kubernetes, Slurm or inference clusters.
  • The release establishes no cost savings; economics depend on hardware utilization, reservation terms and support load.

vCluster Labs put vCluster Platform v5.0 into limited availability on November 19, giving AI cloud providers a way to sell self-service environments on dedicated physical servers. Customers provision machines from their allocated capacity, pick an operating system image and launch Kubernetes, Slurm or inference clusters without provider staff handling each request.

The release targets a familiar bottleneck. A customer can rent a block of GPU servers and still wait on the provider's operations team to prepare them for a new project. Version 5.0 moves some of those decisions to the customer while leaving hardware ownership and capacity allocation with the provider.

What changes in version 5.0?

Physical servers now sit in an inventory maintained by vMetal, the provisioning component beneath the platform. An operator assigns capacity to customers or leaves it available on demand, then decides which machine types, images and cluster templates each customer can use.

v5.0 adds an explicit customer boundary, called a tenant, above individual projects. Quotas limit how much capacity a customer can claim, and customers cannot see inventory they have not been granted.

For a platform team inside a customer, several internal groups can draw on the same contract without each needing a new allocation from the provider. A research group could set up a scheduled Slurm environment while an application team uses Kubernetes for an inference deployment. The provider decides which configurations it supports; the customer chooses among them. How much control a customer holds depends on what its allocation includes and which templates the provider opens up.

vCluster Labs says providers can use its console or build a branded portal on the same underlying API, so an operator selling infrastructure under its own brand can vary the interface without running a separate inventory and provisioning system for each service.

The provider keeps the hardware; the customer runs the cluster.

What does self-service change about bare metal?

Bare metal gives access to physical servers rather than instances sharing a host with unrelated customers. That appeals to buyers who want a known hardware configuration, dedicated accelerator capacity or control over the software environment. But a physical reservation does not by itself produce a usable cloud service. Someone still has to install the operating system, configure access and prepare the cluster.

The customer workflow in v5.0:

  • Provision machines from allocated capacity.
  • Select an operating system image.
  • Launch a Kubernetes, Slurm or inference cluster.

vCluster's approach pairs dedicated machines and per-customer networking with self-service provisioning. A provider can serve simple server rentals and more managed products from one inventory. The business case rests on fewer provider-handled steps, quicker onboarding and a managed option for buyers who want one. The release establishes no savings figure for either side, and the economics will depend on hardware utilization and how much support customers require.

Idle hardware is the other variable. Reserved capacity gives buyers predictability, but those machines can stay unavailable to other customers between projects. An on-demand pool adds flexibility only when the provider can reliably meet demand. The platform lets providers offer both models, and it does not remove the need to set reservation terms or plan for shortages.

The pitch is speed and choice, not a proven cost saving.

What does limited availability mean for operators?

Version 5.0 is not generally available. The vendor's documentation recommends retaining the existing long-term support releases, Platform v4.12 and vCluster v0.37, for current deployments.

Operators weighing the new version still have three checks to run:

  • Confirm compatibility with existing provisioning and networking tools.
  • Verify that customers can move from a machine allocation to a usable cluster without creating new support bottlenecks.
  • Track how quickly the provider delivers working infrastructure after allocating hardware, and how often customers need help afterward.

If those measures improve without making dedicated capacity harder to manage, self-service could become a real distinction among AI cloud providers selling similar hardware.

Until then, the gating question is operational: whether fewer provider-handled steps translate into faster delivery and fewer support tickets.

Source: unite.ai

Share this article:

More from Grace Kim

Grace Kim

Show full bio

Market editor covering people, launches and day-to-day practice in entrepreneurship, startups and venture funding at Business Bearings.

2,447 articles

Related articles

« Previous article