# Amazon EKS and EKS Distro add Kubernetes 1.37 support > AWS announces Kubernetes 1.37 support for EKS and EKS Distro, with generally available Metrics API and DRA device controls and beta HPA scale-to-zero. Canonical URL: https://www.devobs.io/news/news-event-6bedca652f6a779e011c6a0f/ By: Finn Larson Published: 2026-10-05T04:53:24.080Z Updated: 2026-10-05T04:53:24.080Z Event date: 2026-10-02 Section: Cloud ## Availability across EKS regions AWS announced on October 2, 2026, that Amazon EKS and Amazon EKS Distro support Kubernetes 1.37. EKS support is available in all AWS Regions where the service operates, including AWS GovCloud (US). EKS Distro builds are available through ECR Public Gallery and GitHub. According to the announcement, teams can create new EKS clusters or upgrade existing clusters to 1.37 using the EKS console, the `eksctl` command-line interface, or infrastructure-as-code tooling. [AWS announcement](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-eks-distro-kubernetes-version-1-37) ## Generally available features versus beta autoscaling AWS identifies two features reaching general availability: - **Metrics API:** Exposed as `metrics.k8s.io/v1`, it supplies Pod and node CPU and memory usage for the Horizontal Pod Autoscaler (HPA) and `kubectl top`. - **Dynamic Resource Allocation device taints and tolerations:** These let DRA drivers and administrators mark devices such as GPUs so the scheduler avoids them unless the workload tolerates the taint. HPA scale-to-zero has a different maturity label: **beta, enabled by default**. AWS says autoscalers configured with `minReplicas: 0` and object or external metrics can scale workloads down to zero Pods when idle and back up when demand returns. The capability being enabled by default does not mean every workload automatically scales to zero; the stated autoscaler configuration still matters. [AWS feature details](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-eks-distro-kubernetes-version-1-37) ## Separate upgrade readiness from feature adoption The important engineering distinction is the metrics dependency: the scale-to-zero behavior AWS describes requires object or external metrics, not merely the CPU and memory readings highlighted in the Metrics API update. Before relying on the beta feature, teams should validate that metric path and test both idle shutdown and demand-driven recovery. AWS also points operators to EKS cluster insights to check for issues that could affect Kubernetes upgrades. [AWS upgrade guidance](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-eks-distro-kubernetes-version-1-37) Treat cluster compatibility and autoscaling policy as separate rollout decisions. An upgrade to Kubernetes 1.37 need not be coupled with configuring production workloads to run at zero replicas. ## Source references -