vCluster Labs today added a bare metal machine management layer that automates the lifecycle of bare metal servers configured with graphical processor units (GPUs).
Announced at the NVIDIA GTC 2026 conference, vMetal enables IT teams to discover, provision, assign, upgrade, and repurpose physical machines via a centralized control plane. Any server infrastructure connected to the network can be automatically discovered and provisioned as a dedicated bare metal machine or attached to Kubernetes clusters.
Tom Brightbill, head of product for vCluster Labs, said the company expects vMetal will mainly be adopted by so-called neoclouds to optimize consumption of bare metal servers running Kubernetes clusters, but internal IT teams also optimize consumption of infrastructure resources across a multi-tenant environment.
Certified Stacks that VCluster Labs supports include Run:ai from NVIDIA and Slinky, with additional AI frameworks such as Jupyter and SkyPilot planned for future releases. Certified Stacks enable platform teams to deliver repeatable AI environments with governance and guardrails built in
The overall goal is to provide IT teams with an ability to flexibly manage server infrastructure in addition to the virtual instances of Kubernetes clusters that can be configured using the company’s vCluster platform, said Brightbill. In effect, bare metal servers are now another infrastructure resource that IT teams can programmably provision, he added.
As AI workloads are deployed in production environments internal IT teams are assuming more responsibility for managing the underlying infrastructure. While IT teams generally have the tools needed to programmatically provision virtual machines, more of them now also need to once again programmatically provision bare metal servers to deploy AI applications in ways that eliminate the processing overhead that might be added by a virtual machine.
Regardless of motivation, the overall IT infrastructure environment is becoming much more diverse in the age of AI. The challenge is determining which control plane provides the greatest amount of flexibility.
Historically, vCluster Labs has been making a case for Virtual Kubernetes clusters that are used most widely in pre-production environments to reduce the total number of physical Kubernetes servers an organization needs to deploy. However, as the number of instances of virtual clusters being used in production environments grows, more IT teams are looking to deploy virtual clusters to reduce the total cost of IT.
vCluster Labs has also previously made available an Infrastructure Tenancy Platform for AI reference architecture will make it simpler for IT teams to deploy AI inference workloads on platforms based on Kubernetes clusters configured with GPUs.
Each IT team depending on the workload will need to determine to what degree to continue to rely on virtual machines versus bare metal servers. In the meantime, IT teams should assume that in addition to provisioning virtual machines they will be expected to provision bare-metal servers, a skillset that many may have lost after years of relying solely on virtual machines. Less clear, of course, is to what degree artificial intelligence (AI) agents might one day soon automate the provisioning of all types of IT infrastructure.
SHARE THIS STORY
Originally published by Techstrong.IT. Republished with attribution.
