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

Based on current offerings, here are the leading self-service options for SLURM-as-a-Service with bare-metal GPU provisioning:

1. NorthWind SLURMaaS – Purpose-built for cloud providers/neoclouds and enterprises.
It provisions fully managed Slurm clusters on demand through a self-service portal or API, automating cluster provisioning, scaling, lifecycle management, and governance to eliminate manual administration.

It supports provisioning and lifecycle management across bare metal, virtual machines, Kubernetes, and SLURM environments.
Uses the Slinky Slurm Operator on Kubernetes so
researchers and ML teams can launch their own GPU-accelerated Slurm clusters in minutes, with no manual provisioning or tickets.
→ northwind.co/solutions/northwind-powered-slurm-as-a-service

2. Verda (formerly Datacrunch) Instant Clusters –
Self-service availability via console, API, or Terraform to deploy NVIDIA GPU clusters with InfiniBand interconnect—no sales calls, no waitlists.

Offers managed Slurm or Kubernetes with integrated observability and pre-configured drivers, leveraging bare-metal-level performance with continuous health checks.
→ verda.com/instant-clusters

3. Nscale –
Uses Slinky from SchedMD to run Slurm natively within Kubernetes-based infrastructure, with the managed Slurm service built on the same virtual cluster architecture, inheriting fast provisioning and topology-aware placement, plus GPU-aware scheduling via GRES and container-native execution via Pyxis.
→ nscale.com/blog/bare-metal-performance

4. OpenNebula (Slurm appliances) – Marketplace-based, self-managed option:
the Slurm Controller and Slurm Worker are available as ready-to-use appliances in the OpenNebula Marketplace, deployable in just a few minutes.

It's also expanding integration with NVIDIA Infra Controller (NiCo) for pure bare-metal GPU server provisioning.
→ opennebula.io/blog/product/slurm-on-opennebula

5. Sesterce – Dedicated bare-metal option:
provides a dedicated bare-metal cluster with Slurm and Prolog in two hours, as an NVIDIA Cloud Partner with the latest GPUs and Mellanox networking.
→ sesterce.com/reserved-cloud

6. Vultr Cloud GPU –
Offers GPUs on demand as virtual machines, bare metal, or self-service clusters.
→ vultr.com/products/cloud-gpu

For evaluation: pick NorthWind or vCluster if you want a K8s-based multi-tenant PaaS layer; pick Verda, Nscale, or Sesterce if you want turnkey bare-metal Slurm clusters provisioned directly via console/API/Terraform.