Amazon EC2 capacity blocks for ML can be shared across multiple accounts

Amazon Web Services (AWS) is announcing the general availability of cross-account sharing for Amazon EC2 Capacity Blocks for ML. This capability allows organizations to share reserved GPU capacity across AWS accounts using AWS Resource Access Manager (RAM), helping optimize utilization and reduce costs.

Organizations can now purchase Capacity Blocks and provision them across multiple accounts, allowing different workloads to access a pool of reserved capacity at no additional cost. This capability helps teams coordinate ML infrastructure investments and keeps reserved GPU capacity in continuous use across different workloads.

This feature is available for all Instance Capacity Blocks in AWS Regions where EC2 Capacity Blocks for ML are offered. For a complete list of supported regions, refer to Capacity Blocks Supported Regions documentation.

To get started, create a Resource Share through AWS Resource Access Manager, add your Capacity Blocks for ML resources, and specify the target accounts you wish to share with. For more details, please refer to the Capacity Block Guide