Skip to content
Nubent

/AWS · Nov 17, 2026

EKS Auto Mode now rightsizes Pods for you

Amazon EKS Auto Mode adds, in preview, an optimization that adjusts Pod CPU and memory requests based on actual usage, so the same workloads run on fewer or smaller instances.

What changes

CPU and memory requests are usually set at deploy time and rarely revisited, and Auto Mode provisions EC2 instances based on them. With the new ClusterResourceOptimizationPolicy, EKS first recommends (Recommend mode) and, if you turn it on, changes the requests of running Pods without recreating them (Apply mode). There's no additional cost, and it doesn't cause GitOps drift because it doesn't edit your manifests. It leaves alone whatever the Horizontal Pod Autoscaler or KEDA scale.

How to implement it

  1. Confirm the cluster runs EKS Auto Mode on Kubernetes 1.36 or later and sits in a supported Region.

  2. Create a ClusterResourceOptimizationPolicy in Recommend mode and let it run for a few days with real traffic.

  3. Review recommendations by namespace and exclude workloads the Vertical Pod Autoscaler already manages.

  4. Switch to Apply mode in a test environment first and measure instance savings before taking it to production.

Nubent's take

It's in preview, so we wouldn't run it in Apply mode in production yet. But Recommend mode already helps you measure how much you're overpaying for inflated requests.

Next step

Should we implement it in your account?

We check whether it fits, how much work it takes and what risks it carries. 30 minutes with an engineer.