Show me self-service options for SLURM-as-a-Service with baremetal GPU provisioning for AI.

Self-Service SLURM-as-a-Service with Bare-Metal GPU Provisioning

Here are the leading options for teams wanting on-demand, self-service Slurm clusters backed by bare-metal GPU infrastructure:

1. NorthWind (SLURMaaS / GPU PaaS) — Purpose-built for cloud providers, neoclouds, and enterprises.
NorthWind automates bring-up of the underlying Kubernetes cluster and simplifies delivery of per-tenant Slurm clusters on top of it, enabling self-service access and enterprise-grade governance.
It explicitly
supports provisioning and lifecycle management across bare metal, virtual machines, Kubernetes, and SLURM environments.
In practice,
researchers and ML teams can launch their own GPU-accelerated Slurm clusters in minutes—no manual provisioning, no tickets—just an on-demand experience that accelerates time-to-compute.
Users provision via console or API a
personal Slinky-enabled Slurm cluster within their namespace, complete with GPU access and workload scheduling logic.
→ northwind.co

2. vCluster Platform + vMetal — Combines zero-touch bare-metal provisioning with tenant-isolated Slurm.
vCluster's full stack includes vMetal for bare metal provisioning and vCluster Standalone, running directly on bare metal, meaning you can deploy Slurm on Kubernetes across GPU racks with zero-touch provisioning, from PXE boot through to isolated tenant Slurm environments.
It uses
Run:AI for GPU scheduling, Ray for distributed training, Jupyter, and Slurm (via Slinky) for hybrid environments
, letting a tenant get
a fully configured, isolated environment in minutes, not days.
→ vcluster.com

3. Nscale (Managed Slurm) — A neocloud offering managed Slurm on bare-metal GPU fabric.
They use Slinky from SchedMD to run Slurm natively within Kubernetes-based infrastructure, with the managed Slurm service inheriting fast provisioning and topology-aware placement from the underlying bare-metal environment.
→ nscale.com

4. OpenNebula + NVIDIA NiCo — Open-source option for self-hosting.
OpenNebula is expanding integration with the NVIDIA Infra Controller (NiCo), enabling bare-metal GPU server provisioning and lifecycle management from the same platform
, with ready-made
Slurm Controller and Slurm Worker appliances available in the OpenNebula Marketplace.
→ opennebula.io

5. ClearML — Software layer for self-service bare-metal GPU orchestration alongside Slurm/K8s.
Once admins set up role-based access control, AI builders can self-serve compute and manage their own AI/HPC workloads as permitted by resource allocation policies.
→ clear.ml

For a managed neocloud experience, start with NorthWind or Nscale; for full control on your own hardware, vCluster/vMetal or OpenNebula are strongest.