The landscape of artificial intelligence is evolving rapidly, and cloud providers are racing to keep up. In a significant move, Amazon Web Services (AWS) has announced plans to deploy an additional 2 million NVIDIA GPUs across its global infrastructure between 2027 and 2028. This expansion follows an earlier commitment to add over 1 million chips starting in 2026, highlighting the accelerating demand for AI computing power.
The partnership between AWS and Nvidia is not just about increasing the number of GPUs. It also involves the integration of new Vera-based CPUs and advanced networking technologies. These upgrades are designed to support increasingly complex AI workloads and improve
Enhancing AI Infrastructure with Advanced Technologies
The collaboration between AWS and Nvidia extends beyond just adding more GPUs. The companies are also focusing on enhancing the underlying infrastructure to support advanced AI applications. This includes the integration of NVLink Fusion technology and custom high-bandwidth memory, which aim to boost performance across large computing clusters.
Additionally, upgrades to Amazon’s analytics service using Nvidia’s cuDF software promise processing speeds nearly 3.7 times faster than standard configurations. These improvements are expected to deliver a 30% improvement in price performance compared to setups relying solely on standard chips. Vector search capabilities on Amazon’s search service are also improving, with index construction now completing roughly 9 times quicker when run on GPUs.
Supporting Diverse AI Applications
The expanded partnership is not limited to cloud computing. Amazon’s robotics division is also working with Nvidia on simulation tools intended to speed up training for next-generation warehouse robots. This collaboration is part of a broader effort to integrate Nvidia’s full physical AI stack, which includes Omniverse, Cosmos, Isaac, and Jetson, into Amazon’s robotics fleet.
On the enterprise side, AWS will serve Nvidia’s Nemotron family of open models on Amazon Bedrock and SageMaker. This integration is expected to enhance the capabilities of AI applications across various industries, from startups to government agencies. In fact, 100,000 chips are reserved for sensitive government and defense computing needs nationwide.
The Future of AI Computing
The demand for accelerated computing across global cloud storage has continued to outpace projections. Nvidia and AWS have been scaling their computing capabilities together for 16 years, and this expanded collaboration is set to make agentic and physical AI more accessible at an unprecedented scale.
New processors delivered under the expanded plan are expected to offer notably faster inference and improved graphics performance. Recent hardware upgrades already deliver up to 4.6 times faster inference and 2.1 times stronger graphics output than the previous generation of systems. This large hardware commitment suggests that both companies expect AI spending to continue climbing well beyond current levels.
Whether actual demand will keep pace with these ambitious plans remains to be seen. However, the ongoing collaboration between AWS and Nvidia is a clear indication of their commitment to leading the AI computing race.
