AWS SageMaker AI inference endpoints use GPU capacity

Amazon Web Services (AWS) has introduced a new method for deploying AI inference endpoints on its SageMaker platform. This approach allows users to reserve specific GPU capacity, ensuring dedicated resources for model evaluation and deployment. The process involves searching for available GPU instances, creating a training plan reservation, and then launching the SageMaker endpoint on that reserved capacity. This feature aims to provide data scientists with more control and predictability over their inference workloads, managing endpoint lifecycles effectively within the reserved resource framework.