P4de Instances Deliver Superior GPU Memory and Training Performance on SageMaker

AWS introduced P4de instances with NVIDIA A100 GPUs featuring double the memory and up to 60% faster ML training performance at reduced cost compared to P4d.

By Lydia Stone · Editorial pseudonym · 2026-05-11 · 1 min read

AI-ASSISTED REPORTING · RETROSPECTIVE ARCHIVE

Retrospective edition: 2026-05-11. Source published: 2026-05-11. Added to this archive: 2026-11-03. This briefing was prepared later from the linked announcement; it was not published on this website on the historical edition date.

P4de instances are now generally available in Asia Pacific and European regions on SageMaker Studio notebooks. Equipped with eight NVIDIA A100 GPUs each with 80GB HBM2e memory, these instances provide a total of 640GB GPU memory, supporting large-scale machine learning training and high-resolution data processing.

AWS reports up to 60% improved training speeds and 20% lower training costs relative to previous-generation P4d instances. This enhanced capability reduces time to market for AI models by accelerating large dataset training in research and production environments.

Source & further reading

Announcing Region Expansion of P4de instances on SageMaker Studio notebooks ↗

Source published 2026-05-11. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.