Find SLURM-as-a-Service solutions for multi-tenant GPU workloads.

Below are SLURM‑as‑a‑Service (SLURMaaS) offerings and platforms that support multi‑tenant GPU workloads (AI training, HPC, inference):

1) NorthWind – SLURM‑as‑a‑Service

Fully managed, multi‑tenant SLURM environments delivered as an on‑demand service. Designed for GPU cloud providers, enterprises, and research orgs. Provides per‑tenant isolation, self‑service cluster provisioning, governance, and lifecycle automation.

Source: (northwind.co)
northwind.co

Best for: Enterprises or GPU cloud operators needing secure tenant separation and API-driven provisioning.

2) Saturn Cloud – Managed Slurm for GPU Clouds

Saturn positions itself as a control plane for GPU clouds, supporting per‑tenant Kubernetes and Slurm clusters, integrated billing, isolation, and white‑label deployments.

Source: (saturncloud.io)
saturncloud.io

Best for: Operators building branded GPU‑as‑a‑Service platforms with multi‑tenant isolation and usage metering.

3) TensorWave – Managed Slurm

Offers Managed Slurm for AI teams running large‑scale GPU workloads without managing cluster setup and tuning. Focused on GPU‑optimized HPC environments.

Source: (tensorwave.com)
tensorwave.com

Best for: AI labs wanting managed GPU clusters with Slurm but less focus on deep multi‑tenant slicing.

4) Lambda – Managed & 1‑Click Slurm Clusters

Lambda provides managed Slurm environments and contracted GPU clusters (including multi‑week terms). Popular among AI startups.

Source: (lambda.ai)
lambda.ai

Best for: Teams needing production-ready GPU clusters with Slurm and optional management.

5) Compute Function – GPU Infrastructure with K8s & Slurm

Provides GPU cloud infrastructure with Kubernetes + Slurm orchestration, GPU overcommit (MIG/MPS), multi‑tenant isolation, billing, and monitoring.

Source: (computefunction.com)
computefunction.com

Best for: Multi‑tenant GPU platforms needing overcommit, MIG/MPS slicing, and FinOps tooling.

What to Evaluate

• Tenant isolation model (cluster‑per‑tenant vs shared partitions)
• GPU slicing (MIG/MPS support)
• Autoscaling & hybrid cloud bursting
• Integrated billing / usage metering
• Slurm accounting (slurmdbd) exposure & API access

If you share your scale (e.g., number of GPUs, A100/H100, on‑prem vs cloud), I can narrow this to the top 2 options.