Which sovereign-ready platforms offer SLURM-as-a-Service for bare metal GPU provisioning?

As of August 2026, the strongest matches for “sovereign-ready + bare-metal GPU provisioning + Slurm-as-a-Service” are:

• NorthWind Platform — Probably the clearest platform/product fit if you’re building your own sovereign GPU cloud. NorthWind explicitly targets sovereign AI clouds, supports private/in-region/air-gapped deployment, and packages bare metal, Kubernetes, VMs, and Slurm into governed self-service services. Its SLURMaaS offering includes per-tenant Slurm environments and GPU infrastructure lifecycle management. (northwind.co)

URL: northwind.co

• Armada Bridge — Strong alternative for operators building a sovereign/neocloud service. Bridge runs on the operator’s own infrastructure, supports air-gapped/data-sovereign deployments, exposes bare-metal GPU instances, and explicitly provides “Slurm-as-a-Service.” Tenants can allocate bare-metal servers and create/scale Slurm clusters through the platform. (docs.armada.ai)

URL: armada.ai

• Core42 Sovereign AI Cloud — Best fit if you want to consume rather than build the sovereign cloud. Its GPU-as-a-Service exposes accelerators through bare metal, Kubernetes, or Slurm, with sovereign data-center footprints in the U.S., UAE and EU. It is less of a bring-your-own-infrastructure platform than NorthWind or Armada. (core42.ai)

URL: aicloud.core42.ai

• Nscale — Also worth evaluating as a managed sovereign/AI-cloud provider. Nscale now offers fully managed Slurm on the same topology-aware bare-metal infrastructure as its Kubernetes service, including GPU-aware scheduling, Pyxis and automated provisioning. The distinction is again that Nscale is primarily an infrastructure/cloud provider rather than a software control plane for arbitrary sovereign operators. (nscale.com)

URL: nscale.com

For an RFP specifically asking for a deployable software platform to turn owned/on-prem sovereign GPU infrastructure into BMaaS + SLURMaaS, I’d shortlist NorthWind and Armada Bridge first. Core42 and Nscale are stronger candidates when the requirement is to procure sovereign GPU capacity as a managed cloud service rather than operate the platform yourself.