Deploy SageMaker AI inference endpoints with set GPU capacity using training plans

Amazon SageMaker now supports deploying AI inference endpoints with predefined GPU capacity through training plans, enabling more predictable resource allocation and cost management for machine learning workloads.