AWS to Deploy Over One Million NVIDIA GPUs in Massive AI Infrastructure Push

AWS and NVIDIA Announce Unprecedented GPU Deployment

Amazon Web Services and NVIDIA revealed the largest cloud GPU deployment in history at NVIDIA's GTC 2026 conference in San Jose, with AWS committing to roll out more than one million NVIDIA GPUs across its global cloud infrastructure beginning this year.

Full-Stack AI Computing

The deployment covers NVIDIA's complete AI computing stack, including the current Blackwell GPU architecture, the next-generation Vera Rubin platform, RTX PRO Blackwell Server Edition GPUs for enterprise workloads, and the new Groq 3 Language Processing Units designed for ultralow-latency inference. The Vera Rubin platform, built from seven new chips and 1.3 million components, promises ten times more performance per watt compared to its Grace Blackwell predecessor.

New Cloud Instances and Efficiency Gains

AWS will become the first major cloud provider to offer instances powered by the NVIDIA RTX PRO 4500 Blackwell Server Edition GPU, a compact 165-watt chip optimised for data processing, vision AI, and small language model inference. The companies also demonstrated three times faster Apache Spark performance using Amazon EMR with EC2 G7e instances, and introduced disaggregated inference capabilities using NVIDIA's NIXL communication library.

Groq 3 Changes the Inference Game

The Groq 3 LPU, NVIDIA's first chip from its roughly twenty billion dollar acquisition of Groq, delivers thirty-five times better token performance per watt when paired with Rubin GPUs. Each Groq LPX rack houses 256 LPUs and is designed to sit alongside Vera Rubin rack-scale systems.

Beyond the Data Centre

The partnership extends to automotive AI, with Amazon and NVIDIA collaborating on in-vehicle AI assistants that combine Alexa Custom Assistant with the NVIDIA DRIVE AGX platform. The technology, planned for automaker evaluation in early 2027, would enable cars to process requests locally while connecting to cloud services. NVIDIA CEO Jensen Huang projected at least one trillion dollars in revenue from 2025 through 2027, underscoring the extraordinary scale of current AI infrastructure demand.