The real cost of EKS extended support

Standard or extended: what the EKS price list says

EKS bills the control plane per cluster and per hour, regardless of the number of nodes. The rate depends only on where the cluster's Kubernetes version sits in its life cycle:

• standard support: the first 14 months after the version is released on EKS, at $0.10 per cluster-hour;
• extended support: the next 12 months, at $0.60 per cluster-hour, which is the standard rate plus $0.50;
• after that: no support. AWS upgrades the control plane itself.

Three details matter. First, extended support is on by default for new and existing clusters: the upgrade policy is set to EXTENDED unless someone changed it, so nobody has to opt in to pay six times more. Second, the EC2 nodes under the cluster are billed separately and don't change. Third, what you buy is real: AWS keeps patching the control plane and the core add-ons for those versions. It is useful insurance during a migration, not a place to live.

Worked examples: 1, 3 and 10 clusters

The math is linear: 8,760 hours a year, times the hourly rate, times the number of clusters. A startup with production, staging and tooling clusters, all created the same month and never upgraded, is on the second line.

The surcharge doesn't stop at the forced upgrade either. If the cluster lands on a version that is itself in extended support, billing at $0.60 continues.

The EKS calendar as of November 2026

From the official EKS release calendar, dates in UTC:

Clusters on 1.31 are in the tightest spot. After November 26, 2026, AWS upgrades the control plane to the oldest supported version, gradually, at any time after that date, and without advance notice. Only the control plane is upgraded: managed node groups and self-managed nodes stay where they are, and add-ons need a manual update. A cluster upgraded this way cannot use the 7-day rollback that EKS introduced in July 2026.

Landing on 1.32 doesn't fix the bill, because 1.32 is in extended support until March 2027. To get off the $0.60 rate you need 1.34 at least, and 1.35 if you don't want to do it again in December.

Check your clusters

Version and support policy of one cluster:

aws eks describe-cluster --name my-cluster \
  --query "cluster.{version:version,support:upgradePolicy.supportType}"

Every cluster in a region (repeat per region, since list-clusters is regional):

for c in $(aws eks list-clusters --query "clusters[]" --output text); do
  aws eks describe-cluster --name "$c" \
    --query "cluster.[name,version,upgradePolicy.supportType]" --output text
done

End-of-support dates straight from the API, rather than from memory:

aws eks describe-cluster-versions \
  --query "clusterVersions[].[clusterVersion,status,endOfStandardSupportDate,endOfExtendedSupportDate]" \
  --output table

On the bill, filter Cost Explorer on the EKS service and group by usage type: extended support shows up as its own line. Check the exact usage type name in your own Cost and Usage Report rather than relying on a label from a blog post.

Turning extended support off: supportType=STANDARD

AWS lets you opt out of extended support per cluster, by changing its upgrade policy:

aws eks update-cluster-config \
  --name my-cluster \
  --upgrade-policy supportType=STANDARD

Before you do, know what it means:

• It only works in standard support. A cluster already in extended support can't leave it this way. For 1.31, 1.32 and 1.33, the only way out is to upgrade.
• The upgrade becomes automatic at the end of standard support. Instead of paying, the cluster moves to the next version on a date you don't pick.
• Only the control plane moves. Nodes and add-ons fall behind until you update them.
• Rolling back costs money. A cluster auto-upgraded at the end of standard support can be rolled back within 7 days, but only after switching the policy back to EXTENDED, which restarts extended billing.

Our take: for a 1.34 cluster, switching to STANDARD is a sensible budget guardrail if the upgrade is already planned before December 2, 2026. It is not an upgrade plan. On a production cluster with no rehearsal, a forced upgrade costs more than a few weeks at $0.60.

The part that isn't on the invoice

The $0.50 per hour is the visible cost. The rest is harder to put a number on, so we won't invent one. It is the version skew that builds up when the control plane moves and the nodes don't, the add-ons that are suddenly two versions behind, and an upgrade that happens on AWS's schedule instead of in your maintenance window. For a cluster several versions behind, each hop is a separate upgrade with its own checks. Our EKS upgrade checklist covers them in order.

KubePrune is independent and not affiliated with Amazon Web Services. Prices are AWS public list prices checked on November 3, 2026; check the pricing pages for your region and date.

Frequently asked questions

Sources

Related