bartek@aws: ~/news
$ whoami
$ AWS Architect · DevOps · Cloud
Thursday, November 19, 2026

Squeeze More ML Power: Running Multiple Pods on One GPU with NVIDIA Run:ai

Want to pack more Kubernetes workloads onto your GPUs without breaking the bank? This guide shows you how to split a single NVIDIA A10G GPU between two memory-bounded Pods using NVIDIA Run:ai GPU Fractioning on Amazon EKS, complete with memory enforcement and monitoring.

source: [aws/machine-learning-blog]