SageMaker Inference supports MIG GPU partitioning
Understand the story
On November 7, 2026, AWS announced on the AWS Machine Learning Blog (firsthand source) that Amazon SageMaker Inference has launched a GPU partitioning feature. Based on NVIDIA's MIG (Multi-Instance GPU) technology, the feature can divide a single physical GPU into up to 7 hardware-isolated partitions, allowing multiple models to be deployed on one GPU. Users can also allocate partial GPU compute to models on demand instead of occupying the entire card. AWS said the move aims to improve GPU utilization for model endpoints. This is the first report of the event, and no follow-up developments have been disclosed yet.
AI overview of reporting · Updated 1 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- AWS Machine Learning BlogSageMaker Inference now supports Multi-Instance GPU to maximize GPU utilization for model endpoints
Amazon SageMaker Inference has launched a GPU partitioning feature based on NVIDIA MIG, splitting a physical GPU into up to 7 hardware-isolated partitions so multiple models can be deployed on a single GPU with on-demand selection of partial compute.
Story heat trend
Not enough continuous observations to show a trend yet.