Improve cost visibility of Machine Learning workloads on Amazon EKS with AWS Split Cost Allocation Data

AWS has introduced split cost allocation support for accelerated workloads in Amazon EKS, enabling more granular cost tracking for machine learning and AI applications.

• Provides cost visibility for Trainium, Inferentia, NVIDIA and AMD GPU workloads
• Enables pod-level cost tracking across multi-tenant Kubernetes clusters
• Automatically calculates costs using GPU, CPU, and memory usage
• Supports cost allocation tags for easier tracking and chargebacks
• Available across AWS commercial regions at no additional cost

The feature helps organizations optimize resource usage, drive accountability, and make informed decisions about their containerized machine learning infrastructure.