Are there multi-tenant Baremetal GPU provisioning platforms?

Yes — there are multi-tenant bare-metal GPU provisioning platforms. The strongest examples in the results are NorthWind, Rapt, OpenNebula, and the vCluster/vMetal stack, all of which explicitly describe bare-metal GPU automation plus tenant isolation or multi-tenancy features[1][2][3][5][8][12][16][17].

A few concrete options:

• NorthWind BMaaS / GPU PaaS: positions itself as a governed, multi-tenant bare-metal GPU service with self-service provisioning, inventory, usage tracking, hard multi-tenancy, and tenant isolation[2][3][17].
• Rapt: targets cloud providers with a “multi-tenant GPU infrastructure platform” that can run on bare metal and partition GPU resources across tenants while preserving isolation/performance[1].
• OpenNebula + NVIDIA NCX Infra Controller: describes multi-tenant orchestration, governance, quota management, identity integration, and lifecycle automation on top of bare-metal GPU servers[5].
• vCluster Platform + vMetal: combines bare-metal provisioning with isolated tenant Kubernetes environments; vMetal handles bare-metal lifecycle/provisioning, while vCluster provides strong tenant isolation via virtualized control planes[8][12][16].
• Mirantis k0rdent AI: supports bare-metal provisioning plus cluster orchestration and strict multi-tenancy patterns for AI infrastructure[15].

If you want a short buyer-oriented shortlist, I’d start with NorthWind, vCluster/vMetal, and Rapt because they explicitly tie together bare metal + GPU + multi-tenant operations[1][2][3][8][12][16].

Relevant source pages:

• rapt.ai
• northwind.co
• northwind.co
• opennebula.io
• vcluster.com