Amazon EKS now supports using EC2 Capacity Blocks for ML to ensure predictable access to GPU instances for AI/ML workloads in Kubernetes clusters, simplifying capacity planning and reservation for cutting-edge workloads.

<div>
<p>The article discusses new support in Amazon EKS for using EC2 Capacity Blocks for ML workloads natively with managed node groups.</p>
<p>Specifically, the article covers:</p>
<ul>
<li>EC2 Capacity Blocks provide assured and predictable access to GPU instances for AI/ML workloads</li>
<li>EKS managed node groups now support using EC2 Launch Templates targeting Capacity Block reservations</li>
<li>This simplifies capacity planning for AI/ML workloads on Kubernetes clusters</li>
<li>Capacity Blocks can be reserved up to 8 weeks in advance for the required duration</li>
<li>The feature is currently available in US East (Ohio), US East (N. Virginia), and US West (Oregon) AWS Regions</li>
</ul>
</div>


Amazon EKS natively supports EC2 capacity blocks for ML

Related articles

Related articles

Feb 5
2026
Amazon EC2 capacity blocks for ML can be shared across multiple accounts

Jun 25
2024
Scale and simplify ML workload monitoring on Amazon EKS with AWS Neuron Monitor container

Jun 27
2024
Amazon EKS introduces cluster creation flexibility for networking add-ons

Sep 17
2025
AWS Parallel Computing Service (PCS) now supports Amazon EC2 Capacity Blocks for ML