HyperPod Slurm Clusters Can Now Run on Your Own Custom AMIs!

Hi everyone, it's Shii-chan! If you're building AI/ML infrastructure, here's news that makes environment setup a bit easier.

AWS What's New aws.amazon.com

What was announced?

This comes from AWS What's New. Amazon SageMaker HyperPod now supports custom AMIs (Amazon Machine Images) for Slurm-orchestrated clusters, so you can deploy clusters with pre-configured, security-hardened environments that match your organization's requirements.

Why it matters

Teams running AI/ML workloads on HyperPod Slurm clusters need environments that meet strict security, compliance, and operational requirements, while still keeping cluster startup fast. Until now, that meant writing complex lifecycle configuration scripts, which slowed deployments and created inconsistencies across cluster nodes.

What changes

You can now build on HyperPod's performance-optimized base AMIs while baking in your own security agents, compliance tools, proprietary libraries, and specialized drivers directly into the image. That means faster startup, better reliability, and stronger security compliance.

Security teams can embed organizational policies directly into base images, so AI/ML teams get pre-approved environments that speed up time-to-training while still meeting enterprise security standards.

Dive Deep

You can specify a custom AMI through three APIs:

• CreateCluster API — when creating a new cluster
• UpdateCluster API — when adding instance groups
• UpdateClusterSoftware API — when patching an existing cluster

Custom AMIs must be built from HyperPod's public base AMIs, which keeps them compatible with distributed training libraries and cluster management features. This is available in every AWS Region where Amazon SageMaker HyperPod is supported.

Wrap-up

• HyperPod Slurm clusters now support custom AMIs
• Bake in your own security agents, compliance tools, libraries, and drivers
• Specify custom AMIs via CreateCluster, UpdateCluster, or UpdateClusterSoftware
• Must be built from HyperPod's public base AMIs

If your team needs to meet strict security standards while keeping AI/ML cluster environments consistent, this update is for you.