
    Amazon EKS now supports up to 100,000 worker nodes per cluster, enabling ultra-scale AI/ML training and inference workloads with unprecedented compute accelerator consolidation.


<div>
<p>
Amazon EKS has expanded its cluster support to 100,000 worker nodes, enabling massive AI/ML training and inference workloads in a single cluster.
</p>
<ul>
<li>Supports up to 1.6 million Trainium chips with Trn2 instances</li>
<li>Can accommodate 800,000 NVIDIA GPUs using P5 and P6 instances</li>
<li>Allows organizations to build and deploy more powerful AI models</li>
<li>Reduces costs by efficiently sharing compute accelerators</li>
<li>Enables seamless use of existing AI/ML tools and frameworks</li>
</ul>
<p>
This enhancement allows organizations to run ultra-scale AI workloads with consolidated access to massive compute resources within a single Kubernetes cluster.
</p>
</div>


Amazon EKS now supports up to 100,000 worker nodes per cluster

Related articles