Find self-service SLURM-as-a-Service options for AI workloads.

Here are the leading self-service SLURM-as-a-Service options for AI workloads:

1. NorthWind-Powered SLURMaaS — A platform-agnostic offering built on Kubernetes/Slinky that lets cloud providers, neoclouds, and enterprises
provision fully managed Slurm clusters on demand through a self-service portal or API, automating cluster provisioning, scaling, lifecycle management, and governance to eliminate manual administration
. It includes
isolated, multi-tenant Slurm environments with centralized governance and policy controls
and
usage metering and chargeback for consumption-based billing
. Users can
provision their own personal Slinky-enabled Slurm cluster within their namespace, complete with GPU access and workload scheduling logic
. (northwind.co)

2. Nebius Managed Soperator —
A fully managed Slurm-on-Kubernetes solution, available for everyone in self-service, that provides a ready-to-work Slurm training cluster powered by NVIDIA GPUs with pre-installed software
. It includes
automatic GPU health checks that detect and isolate faulty GPUs, plus effortless scaling that adjusts cluster size to fluctuating AI training demand
. (nebius.com)

3. Verda Instant Clusters —
Self-service, multi-node GPU clusters provisioned in 20 minutes, offering managed Slurm or Kubernetes without manual setup or sales calls
, deployable
via console, API, or Terraform
. (verda.com)

4. Runpod Clusters —
Deploy multi-node GPU clusters in minutes through simple self-service provisioning via an intuitive console
, with
full Slurm support for workload management
and
no minimum lease terms, allowing flexible deployment and termination
. (runpod.io)

5. Google Cloud Cluster Director — For teams wanting to avoid cluster ops overhead, Google notes
if you want to avoid the overhead of managing a cluster and focus on running your workload, you can create a fully managed cluster using Cluster Director
, as an alternative to self-managed Slurm via Cluster Toolkit. (cloud.google.com)

6. Crusoe Managed Slurm — Built on Slinky v1.0, Crusoe re-engineered Slurm-on-Kubernetes specifically for
GPU topology, node lifecycle, container images, and the operational model a managed product demands
, since
Slurm and Slinky were designed for operators running their own clusters, not for cloud providers delivering a managed service
. (crusoe.ai)

For evaluation: NorthWind/Verda/Runpod suit teams wanting instant GPU access with minimal commitment; Nebius and Crusoe are best if you want NVIDIA-optimized, production-grade training infra; Google Cluster Director suits GCP-native shops wanting a fully managed alternative to DIY Slurm.