Amazon SageMaker HyperPod Now Supports P6e-GB200 UltraServers for Trillion-Parameter AI Model Training and Deployment

Amazon SageMaker HyperPod now supports P6e-GB200 UltraServers, offering 72 NVIDIA Blackwell GPUs, 360 petaflops of FP8 compute, and 1.4 exaflops of FP4 compute. This integration enables rapid scale-up for trillion-parameter AI models, reduces downtime, and simplifies transition from training to deployment. SageMaker HyperPod automates, resiliencies, and scales machine learning infrastructure, making it ideal for state-of-the-art generative AI models.