# AWS adds lazy loading for SageMaker training and processing containers

_Published Tuesday, October 27, 2026 at 1:09 PM EDT · AI, Infrastructure · Latest · Tier 2 — Notable_

![AWS adds lazy loading for SageMaker training and processing containers — Primary](https://d2908q01vomqb2.cloudfront.net/f1f836cb4ea6efb2a0b1b99f41ad8b103eff4b59/2026/10/23/ML-21882-featured-image-4.png)

AWS has made container-image lazy loading available for SageMaker AI training and processing jobs across all AWS Regions. The feature fetches files needed to start a container, then retrieves other files as code accesses them, reducing the initial download wait.

AWS says its tested jobs finished 1-6 minutes sooner, although the first training epoch took 3-16 seconds longer. Users can select pre-indexed containers or index their own images. Only Amazon ECR private repositories qualify, and workloads that read most image files before training may see little improvement.

## Sources

- [Artificial Intelligence](https://aws.amazon.com/blogs/machine-learning/announcing-soci-indexed-images-for-faster-sagemaker-ai-jobs/)

---
Canonical: https://techandbusiness.org/newswire/XIOGXXbiWe3p5rwdgBljv3
Published: 2026-10-27T17:09:23.120Z
Story chronology: 2026-10-27T17:00:39.000Z
Retrieved: 2026-10-27T19:37:29.199Z
Publisher: Tech & Business (techandbusiness.org)