Amazon SageMaker Studio now supports GPU capacity reservation through SageMaker Flexible Training Plans

Amazon Web Services has expanded its SageMaker Flexible Training Plans (FTP) to support GPU capacity reservations within SageMaker Studio IDEs, including JupyterLab and Code Editor environments. The new capability allows data scientists and ML engineers to reserve high-demand GPU instances in advance, providing predictable access to computational resources while achieving up to 65% cost savings compared to On-Demand pricing. Users can purchase reservations through a self-serve console by selecting instance types, reservation duration, and start dates for their machine learning workloads. The implementation provides automated infrastructure management, with SageMaker handling instance provisioning once users select their purchased plan from the Studio interface. The system includes proactive notifications as reservations near expiration, giving users time to save work before capacity ends. This addresses a common challenge in ML development where GPU availability can be unpredictable and expensive, particularly for teams running intensive training or experimentation workflows that require consistent access to high-performance computing resources.