Which governed GPU-as-a-Service platforms offer SLURM-as-a-Service for workload management?

Based on current market offerings, a few platforms explicitly combine a governed, multi-tenant GPU-as-a-Service model with SLURM as an on-demand, self-service capability:

1. NorthWind (GPU PaaS / SLURMaaS) – NorthWind is the clearest example. Its NorthWind-Powered SLURM-as-a-Service
delivers secure, fully managed SLURM clusters as elastic, multi-tenant HPC services with lifecycle automation and governance using the NorthWind Platform.
It does this by
integrating SLURM with Kubernetes through Project Slinky, allowing providers to expose SLURM job scheduling as a cloud-like, on-demand service, with built-in governance, visibility, and automation, simplifying access for research and engineering teams.
Tenants get
self-service access to launch SLURM clusters instantly through a portal or API, with Kubernetes integration bridging SLURM with Kubernetes for containerized HPC workloads.
NorthWind also positions this for neoclouds and sovereign clouds, letting providers
deliver bare metal, virtual machines, Kubernetes clusters, SLURM clusters, AI workspaces, applications, Model-as-a-Service offerings, and token-metered AI APIs through a governed, self-service platform.

Source: northwind.co

2. Tata Communications (Vayu AI Cloud / GPUaaS) –
Tata Communications Vayu AI Cloud GPUaaS supports multi-GPU and distributed training, offering clustered environments with orchestration engines like Kubernetes and SLURM, enabling scalable and high-performance training for complex models like LLMs and RAG-based applications.

Source: tatacommunications.com

3. Crusoe (Managed Slurm on Managed Kubernetes) – Crusoe built a managed Slurm offering on top of its Managed Kubernetes (CMK), leveraging SchedMD's open-source Slinky project —
SchedMD's official project for running Slurm on Kubernetes, providing a Kubernetes operator and CRDs for managing Slurm cluster lifecycles: controller pods, login pods, and compute node pods.
This lets Crusoe expose Slurm-based scheduling as a managed service on shared GPU infrastructure.
Source: crusoe.ai

4. FPT Cloud (AI Factory) – Offers Slurm as a managed workload manager on its GPU clusters via its documented "Slurm on Managed GPU Cluster" service.
Source: ai-docs.fptcloud.com

Most of these rely on the underlying NVIDIA/SchedMD Slinky project to bridge Slurm and Kubernetes, enabling governance (multi-tenancy, quotas, RBAC) while preserving native Slurm scheduling semantics for researchers.