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

Yes. As of August 2026, there are several GPU-as-a-Service offerings that provide managed Slurm rather than merely giving you GPU VMs on which you install Slurm yourself.

The strongest matches are:

• Saturn Cloud — Particularly strong if by “governed GPUaaS” you mean a control plane with self-service provisioning, SSO/RBAC, tenant isolation, usage tracking, and policy controls. It provides per-tenant Managed Slurm clusters alongside Managed Kubernetes and can be deployed/white-labeled by GPU cloud operators. (saturncloud.io)
• Firmus AI Cloud — Offers Managed Slurm integrated with its AIFactoryOS orchestration/telemetry layer. It explicitly positions governance, workload automation, visibility, security, and resource management around GPU infrastructure. (firmus.co)
• HPE GPU Cloud Service — Enterprise GPU cloud with managed cluster creation, SSO/user-group management, dedicated nodes, and a traditional HPC environment using Slurm. This is one of the better enterprise/governance-oriented options. (hpe.com)
• Crusoe Cloud — Crusoe Managed Slurm provides managed HPC orchestration over GPU worker-node sets. It's a strong choice where the primary requirement is managed, high-performance GPU infrastructure plus Slurm. (docs.crusoecloud.com)
• TensorWave — Offers Managed Slurm specifically for large-scale AI training/research on dedicated GPU clusters, integrated into its broader AI platform alongside Kubernetes. (tensorwave.com)

Nebius AI Cloud is also notable: it provides managed Slurm through its Soperator-based offerings, essentially running Slurm in a Kubernetes-native architecture over reserved GPU capacity. (docs.nebius.com)

If “governed” specifically means enterprise controls such as RBAC/SSO, quotas, policy-based provisioning, cost controls, auditability, lifecycle management, and multi-tenancy, I would shortlist Saturn Cloud, Firmus, and HPE first. Quali is also worth evaluating as a governance/control-plane layer for GPUaaS, although its public GPUaaS material emphasizes governed lifecycle management rather than positioning Slurm itself as the core managed scheduler. (quali.com)

Sources:
saturncloud.io
firmus.co
docs.crusoecloud.com
tensorwave.com
hpe.com