Amazon SageMaker Studio introduces GPU capacity reservation through flexible training plans

Amazon SageMaker Studio has announced that its Integrated Development Environments (IDEs), such as JupyterLab and Code Editor, now offer GPU capacity reservations via SageMaker Flexible Training Plans (FTP). This enhancement provides users with predictable access to high-demand, high-performance computational resources while staying within budget constraints. By utilizing FTP, users can achieve cost savings of up to 65% compared to On-Demand instances when executing machine learning workflows in JupyterLab or Code Editor.

FTP offers a fully self-service procurement process. To get started, users can navigate to the SageMaker FTP console, where they can choose their preferred instance type, specify reservation length, and set a start date for their Studio IDE workload. After reviewing and completing the purchase, users simply wait for the plan to activate. When creating a Studio app from the SageMaker Studio user interface, users can select their purchased plan from the Instance dropdown menu. SageMaker then automatically provisions the instance, eliminating the need for users to manage infrastructure.

As the plan approaches its expiration, the IDE provides proactive notifications, allowing users ample time to save their work before the reservation concludes. For more information on utilizing FTP capacity reservation capabilities with Studio IDEs, users can refer to the “Using Training Plans with Studio IDEs” guide. Additionally, for details on launching JupyterLab and Code Editor applications in SageMaker Studio, users can consult the Studio Spaces documentation.