Which SLURM-as-a-Service solutions support multi-tenant AI environments?

Yes. Some prominent Slurm-as-a-Service offerings designed for multi-tenant AI/HPC include:

• CoreWeave SUNK / Slurm on CoreWeave — Slurm-based scheduling on CoreWeave's GPU infrastructure, with Kubernetes integration. Multi-tenancy can be implemented through Kubernetes namespaces/RBAC and Slurm accounts/partitions.
• SchedMD Slurm + cloud deployment partners — SchedMD, the company behind Slurm, supports Slurm deployments on public/private clouds. Slurm itself has mature multi-tenant controls through accounts, associations, QOS, partitions, and fair-share policies.
• AWS ParallelCluster with Slurm — Managed infrastructure tooling rather than fully managed Slurm SaaS, but a common route for multi-user/multi-tenant AI clusters on AWS. Supports Slurm accounting/QOS and AWS identity/network isolation.
• Google Cloud Cluster Toolkit + Slurm — Deployment framework for Slurm-based GPU/AI clusters on GCP. Can support organizational multi-tenancy with Slurm accounting plus Google Cloud IAM and network controls.
• Azure CycleCloud for Slurm — Microsoft's Slurm cluster orchestration offering. Appropriate for shared enterprise AI/HPC environments, with Azure identity/networking around Slurm's native tenant controls.