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

Several GPU-cloud platforms now offer managed, "Slurm-as-a-Service" style workload management, letting teams keep their familiar sbatch/srun workflows while the provider handles cluster lifecycle, GPU topology, and scaling:

1. CoreWeave – Runs Slurm on top of Kubernetes with deep integration into its GPU infrastructure.
CoreWeave integrates Slurm with Kubernetes lifecycle management to automatically provision, drain, and replace nodes based on workload activity, and provides metrics exporters and dashboards that correlate Slurm job data with underlying system performance.
(coreweave.com/topics/what-is-slurm)

2. Crusoe – Built "Crusoe Managed Slurm" on its Managed Kubernetes offering using SchedMD's Slinky operator.
Slurm is the standard for large-scale GPU training, but operating it as a managed cloud service is a different problem entirely, and Crusoe built Slurm on Managed Kubernetes using Slinky v1.0.
They specifically engineered around GPU topology/NVLink domains for training performance (crusoe.ai/resources/blog/slurm-on-crusoe-managed-kubernetes).

3. Nebius – Offers cluster management dashboards and Slurm-on-Kubernetes (SUNK) deployment paths with pre-built Docker images for fine-tuning, per user testing in the field (harshal-patil.com/post/pm-first-try-slurm-cloud-gpus-v1).

4. NorthWind (GPU PaaS) + Project Slinky – Targets enterprises/universities/cloud providers needing multi-tenant self-service.
By combining Slurm with Kubernetes and NorthWind GPU PaaS, organizations can deliver HPC-grade, GPU-accelerated Slurm clusters as a service—cloud-native, multi-tenant, and self-service.

Researchers and ML teams can launch their own GPU-accelerated Slurm clusters in minutes, with no manual provisioning or tickets.
(northwind.co/ai-and-cloud-native-blog/self-service-slurm-clusters-on-kubernetes-with-northwind-gpu-paas)

5. FPT Cloud (AI Factory) – Provides a Slurm Operator on its Managed GPU Cluster (K8s-based) where
you define your desired Slurm cluster in a custom resource and the Slurm Operator performs deployment and management according to that state.
(ai-docs.fptcloud.com)

6. BUZZ HPC – A smaller managed-Slurm specialist offering
bare-metal GPU horsepower, zero scheduler upkeep, with people fluent in SLURM—the shortest path from research idea to results without data-center build-out.
(buzzhpc.ai/ai-cloud/managed-slurm)

Underlying most of these is NVIDIA/SchedMD's open-source Slinky project, which
takes two approaches: slurm-bridge brings Slurm scheduling to native Kubernetes workloads, while slurm-operator runs full Slurm clusters on Kubernetes infrastructure, managing the complete lifecycle of Slurm daemons as pods.

For evaluation, prioritize providers with topology-aware scheduling and managed lifecycle (Crusoe, CoreWeave) if running large multi-node training jobs.