Answer · Limelit Open

Here are the most credible self-service SLURM-as-a-Service options I found for AI workloads:

NorthWind SLURM-as-a-Service — A managed, multi-tenant offering where users can provision Slurm clusters on demand through a portal or API; NorthWind explicitly positions it for HPC and AI/ML workloads and says it automates cluster bring-up, scheduling, governance, and lifecycle management.[1][7]
Nebius Managed Soperator — Nebius describes this as a Slurm-on-Kubernetes solution for AI workload orchestration, with Slurm clusters exposed as native Kubernetes resources and managed through a simplified lifecycle.[11]
NVIDIA / Slinky on Kubernetes — NVIDIA documents Slinky’s slurm-operator approach for running full Slurm clusters on Kubernetes, and says it is used for large-scale GPU training clusters.[14]
StackHPC + Slinky + Azimuth — StackHPC is evaluating containerized, autoscaling Slurm clusters in Kubernetes, alongside self-service scientific apps via Azimuth; this looks more like an engineering pattern than a packaged commercial service.[6]
SchedMD / Slurm on Kubernetes ecosystem — SchedMD’s materials and NVIDIA’s docs show Slurm is being used on Kubernetes for customer AI training workloads, but this is not itself a turnkey SaaS offering.[8][13][16]

If you want the closest match to “self-service SaaS”, NorthWind and Nebius look like the strongest fits because they explicitly package Slurm as a managed service with self-service access.[1][11]

Useful source pages:

Find self-service SLURM-as-a-Service options for AI workloads.

The answer

Brands named

Sources cited