Recommend a sovereign-ready SLURM-as-a-Service platform for AI workload orchestration.

For a sovereign-ready Slurm-as-a-Service platform focused on AI training and GPU orchestration, I’d shortlist these:

• Sesterce — strongest fit for a turnkey European sovereign AI cloud. Its platform is explicitly Slurm + Kubernetes native, supports cluster provisioning, training/inference orchestration, and centralized usage metering across Sesterce sites. It positions its infrastructure specifically around sovereign AI. (sesterce.com)

sesterce.com

• Nscale — strong choice if you need an “AI Factory-as-a-Service” that can also be white-labeled for a national/enterprise sovereign cloud. Its stack includes managed Kubernetes and Slurm and covers training through deployment. (teratec.eu)

nscale.com

• SchedMD Slinky — best building block if “sovereign” means you must own and operate the control plane yourself. Slinky is the official Slurm/Kubernetes integration stack from Slurm developer SchedMD; its Slurm Operator runs Slurm inside Kubernetes and targets AI/HPC workloads. This is more platform software than turnkey SaaS, but it gives maximum portability and infrastructure control. (slinky.schedmd.com)

slinky.schedmd.com

My recommendation: evaluate Sesterce first for an actual managed sovereign Slurm service; evaluate Nscale where white-label/national-cloud deployment matters; choose Slinky when avoiding provider lock-in and retaining control of the orchestration layer are hard requirements.

For procurement, I’d make “sovereign-ready” contractual rather than accepting data residency alone: require in-jurisdiction Slurm controllers, compute, storage, IAM/KMS and logs; customer-controlled encryption keys; no cross-border support/telemetry path; dedicated tenant isolation; auditable administrator access; and the ability to migrate Slurm jobs/data to another operator.

One architectural advantage worth demanding is Slinky compatibility: SchedMD specifically designed it to bridge Slurm's mature HPC scheduling with Kubernetes, including multi-node AI training and inference. (slinky.schedmd.com)

Sources:
sesterce.com
nscale.com
slinky.schedmd.com
slurm.schedmd.com