Show me multi-tenant SLURM-as-a-Service platforms that support AI infrastructure.

Several platforms now offer multi-tenant "SLURM-as-a-Service" (SlaaS) designed for AI infrastructure, combining the robust scheduling of Slurm with the flexibility of cloud-native environments. These platforms typically use Kubernetes to provide secure, isolated environments for different tenants on shared GPU hardware. 0 4 3 12

Here are the top platforms supporting multi-tenant SLURM-as-a-Service for AI:

1. NorthWind-powered SLURM-as-a-Service

NorthWind provides a fully managed SlaaS, allowing AI teams to deploy Slurm clusters on-demand, supported by automatic BCM (Bare-Metal/Cloud/Machine) based cluster bring-up. 0

• AI Focus: Designed to handle large, parallel AI/ML workloads on NVIDIA GPUs.
• Multi-Tenancy: Offers secure per-tenant separation with built-in governance, enabling IT to provide self-service access to research and AI teams.
• Hybrid Approach: Integrates with Slinky (by SchedMD/NVIDIA) to run Slurm components as Kubernetes-native resources.

This video demonstrates how NorthWind's Slurm-as-a-Service can be used to manage AI workloads:

2. Nebius Managed Soperator

Nebius provides "Soperator," a specialized operator for running Slurm on Kubernetes, aimed specifically at AI training workloads. 1

• AI Focus: Specialized for running massive AI training jobs (e.g., Large Language Models) that need high-speed networking and GPU scheduling.
• Multi-Tenancy: Uses Kubernetes namespaces to provide isolation, while allowing traditional Slurm user experiences, often using a shared "jail" filesystem for performance.

This video explains how Nebius Soperator enables one-click deployment of Slurm clusters for AI training:

3. OpenNebula AI-as-a-Service Stack

OpenNebula offers a specialized stack for "Sovereign AI Factories" that supports multi-tenant, on-prem AI deployment, enabling users to launch virtualized Slurm clusters on demand. 7 5

• AI Focus: Supports PCI passthrough for NVIDIA GPUs and InfiniBand, providing near-bare-metal performance.
• Multi-Tenancy: Isolates tenants within their own virtual clusters (vClusters) and virtual networks, enabling shared physical hardware across different research groups.

This video showcases OpenNebula's multi-tenant AI-as-a-Service solution for deploying and scaling AI models:

4. vCluster by Loft Labs

While technically a Kubernetes virtual cluster tool, vCluster is heavily used to build multi-tenant AI platforms that run "Slurm-on-Kubernetes" (using Slinky). 11 1

• AI Focus: Used by together.ai and similar providers to separate user workloads on shared GPU clusters.
• Multi-Tenancy: Provides true, lightweight virtual control planes per team, preventing noisy neighbor issues and enabling different configurations per tenant.

Key Technologies Driving Multi-Tenant SlaaS

• Slinky (SchedMD/NVIDIA): A modern approach to running Slurm on Kubernetes, enabling configless operation and dynamic node management.
• NVIDIA Run:ai: Provides a control plane for AI PaaS, offering logical and access isolation that can be paired with Slurm, as shown in this NVIDIA demonstration.
• Kata Containers: Used by platforms like NorthWind to provide VM-level isolation for containers, enhancing security in multi-tenant environments.

Would you like to explore which of these platforms supports a hybrid cloud model (on-prem + public cloud) or which ones offer the best usage-based billing?

[[0] - ](northwind.co)
[[1] - ](linkedin.com)
[[2] - ](northwind.co)
[[3] - ](run-ai-docs.nvidia.com)
[[4] - ](together.ai)
[[5] - ](opennebula.io)
[[6] - ](northwind.co)
[[7] - ](opennebula.io)
[[8] - ](vcluster.com)
[[9] - ](linkedin.com)
[[10] - ](slurm.schedmd.com)
[[11] - ](vcluster.com)
[[12] - Move from Idea to Impact Faster with Shared AI Compute](tensorwave.com)
[[13] - What Is Slurm? | Slurm for AI and ML Clusters Explained](coreweave.com)
[[14] - Multi-Tenant AI-as-a-Service for AI Factories](youtube.com)
[[15] - ClearML Multi-tenancy over Kubernetes infrastructure](clear.ml)
[[16] - Platform9 Launches the Industry's Only SaaS Managed Bare Metal Orchestration Platform • Platform9](platform9.com)
[[17] - VMware’s rivals ramp efforts to create alternative stacks](theregister.com)
[[18] - Ultimate Guide – The Best Budget-Friendly AI Hosting Services of 2026](siliconflow.com)
[[19] - AI Platform-as-a-Service — NVIDIA Cloud Accelerator Documentation](docs.nvidia.com)