NVIDIA and AWS Expand Full-Stack Partnership, Providing the Secure, High-Performance Compute Platform Vital for Future Innovation

Source: NVIDIA AI Blog

Overview

At AWS re:Invent, NVIDIA and Amazon Web Services expanded their strategic collaboration with new technology integrations across interconnect technology, cloud infrastructure, open models, and physical AI.

As part of this expansion, AWS will support NVIDIA NVLink Fusion — a platform for custom AI infrastructure — for deploying its custom‑designed silicon, including next‑generation Trainium4 chips for inference and agentic AI model training, Graviton CPUs for a broad range of workloads, and the Nitro System virtualization infrastructure.

Using NVIDIA NVLink Fusion, AWS will combine NVIDIA NVLink scale‑up interconnect and the NVIDIA MGX rack architecture with AWS custom silicon to increase performance and accelerate time to market for its next‑generation cloud‑scale AI capabilities.

AWS is designing Trainium4 to integrate with NVLink and NVIDIA MGX, the first of a multigenerational collaboration between NVIDIA and AWS for NVLink Fusion. AWS has already deployed MGX racks at scale with NVIDIA GPUs. Integrating NVLink Fusion will further simplify deployment and systems management across its platforms and allow AWS to harness the NVLink Fusion supplier ecosystem, which provides all components required for full rack‑scale deployment—from rack and chassis to power‑delivery and cooling systems.

By supporting AWS’s Elastic Fabric Adapter and Nitro System, the NVIDIA Vera Rubin architecture on AWS will give customers robust networking choices while maintaining full compatibility with AWS’s cloud infrastructure and accelerating new AI service rollout.

“GPU compute demand is skyrocketing — more compute makes smarter AI, smarter AI drives broader use and broader use creates demand for even more compute. The virtuous cycle of AI has arrived,” said Jensen Huang, founder and CEO of NVIDIA. “With NVIDIA NVLink Fusion coming to AWS Trainium4, we’re unifying our scale‑up architecture with AWS’s custom silicon to build a new generation of accelerated platforms. Together, NVIDIA and AWS are creating the compute fabric for the AI industrial revolution — bringing advanced AI to every company, in every country, and accelerating the world’s path to intelligence.”

“AWS and NVIDIA have worked side by side for more than 15 years, and today marks a new milestone in that journey,” said Matt Garman, CEO of AWS. “With NVIDIA, we’re advancing our large‑scale AI infrastructure to deliver customers the highest performance, efficiency and scalability. The upcoming support of NVIDIA NVLink Fusion in AWS Trainium4, Graviton and the Nitro System will bring new capabilities to customers so they can innovate faster than ever before.”

Convergence of Scale and Sovereignty

AWS has expanded its accelerated computing portfolio with the NVIDIA Blackwell architecture, including NVIDIA HGX B300 and NVIDIA GB300 NVL72 GPUs, giving customers immediate access to the industry’s most advanced GPUs for training and inference. Availability of NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, designed for visual applications, on AWS is expected in the coming weeks.

These GPUs form part of the AWS infrastructure backbone powering AWS AI Factories, a new AI cloud offering that provides customers worldwide with dedicated infrastructure to harness advanced AI services in their own data centers, operated by AWS, while maintaining control of their data and complying with local regulations.

NVIDIA and AWS are committing to deploy sovereign AI clouds globally, delivering secure, sovereign AI infrastructure that meets increasingly rigorous sovereign AI requirements. For public‑sector organizations, AWS AI Factories will transform the federal supercomputing and AI landscape, enabling seamless integration of AWS’s reliable, secure, and scalable cloud services with NVIDIA Blackwell GPUs and the full‑stack NVIDIA accelerated computing platform, including NVIDIA Spectrum‑X Ethernet switches.

The unified architecture ensures customers can access advanced AI services, train and deploy massive models, and retain absolute control of proprietary data while remaining fully compliant with local regulatory frameworks.

NVIDIA Nemotron Integration With Amazon Bedrock Expands Software Optimizations

Beyond hardware, the partnership expands integration of NVIDIA’s software stack with the AWS AI ecosystem. NVIDIA Nemotron open models are now integrated with Amazon Bedrock, enabling customers to build generative AI applications and agents at production scale. Developers can access Nemotron Nano 2 and Nemotron Nano 2 VL to build specialized agentic AI applications that process text, code, images, and video with high efficiency and accuracy.

The integration makes high‑performance, open NVIDIA models instantly accessible via Amazon Bedrock’s serverless platform, providing proven scalability and zero infrastructure management. Industry leaders such as CrowdStrike and BridgeWise are the first to use the service to deploy specialized AI agents.

NVIDIA Software on AWS Simplifies Developer Experience

NVIDIA and AWS are co‑engineering at the software layer to accelerate the data backbone of every enterprise. Amazon OpenSearch Service now offers serverless GPU acceleration for vector index building, powered by NVIDIA cuVS, an open‑source library for GPU‑accelerated vector search and data clustering. Early adopters report up to 10× faster vector indexing at a quarter of the cost, reducing search latency, accelerating writes, and unlocking faster productivity for dynamic AI techniques like retrieval‑augmented generation.

Production‑ready AI agents require performance visibility, optimization, and scalable infrastructure. By combining:

• Strands Agents – for agent development and orchestration
• NVIDIA NeMo Agent Toolkit – for deep profiling and performance tuning
• Amazon Bedrock AgentCore – for secure, scalable agent infrastructure

organizations can empower developers with a complete, predictable path from prototype to production.

This expanded support builds on AWS’s existing integrations with NVIDIA technologies—including NVIDIA NIM microservices, NVIDIA Riva, and NVIDIA BioNeMo—as well as model development tools integrated with Amazon SageMaker and Amazon Bedrock.