Amazon Triples Nvidia GPU Commitment to 3 Million Chips for AWS Amid Surging AI Demand

Amazon and Nvidia announced an expanded partnership on August 26, 2026, with Amazon committing to an additional 2 million Nvidia GPUs for AWS data centers, tripling its previous order to a total of 3 million chips. This significant deal, involving Blackwell Ultra, Rubin, and Rubin Ultra GPUs, underscores the surging demand for AI infrastructure, with deliveries scheduled for 2027 and 2028.

Expanded GPU Deployment for AWS

The core of the deepened collaboration centers on the substantial increase in Nvidia GPU deployment within AWS. The additional 2 million GPUs are slated to enhance AWS's capacity to support advanced AI workloads, reflecting the rapid growth in AI development and adoption across various industries. This move positions AWS to offer more robust computing resources for its cloud customers, who rely on high-performance GPUs for training and deploying complex AI models.

The integration of next-generation Nvidia GPUs, such as the Blackwell Ultra, Rubin, and Rubin Ultra, indicates Amazon's focus on leveraging the latest hardware innovations to maintain its competitive edge in cloud computing. The phased delivery through 2027 and 2028 suggests a long-term strategy to continuously upgrade and expand AWS's AI capabilities.

Nvidia's Financial Performance and Future Outlook

The announcement coincided with Nvidia reporting strong financial results. The company achieved a record Q2 revenue of $96.2 billion, with data center revenue specifically reaching $89 billion. Nvidia's net income for the quarter was $59.7 billion. Looking ahead, Nvidia provided Q3 guidance projecting $108 billion in revenue, signaling continued growth in the AI hardware market.

The production shipments of Rubin GPUs, which commenced in Q2 FY2027, underscore Nvidia's ability to bring its latest technologies to market, meeting the high demand from major cloud providers like Amazon. These financial figures and product milestones highlight Nvidia's central role in the global AI infrastructure build-out.

Integrating Nvidia's AI Stack Across Amazon Operations

Beyond GPU hardware, the partnership extends to Amazon's broader AI strategy. Amazon plans to adopt Nvidia's comprehensive "physical AI" stack for its warehouse robots. This integration includes Nvidia Omniverse, Cosmos, Isaac, and Jetson platforms, designed to enhance the intelligence and autonomy of Amazon's robotic systems in logistics and fulfillment centers.

Furthermore, AWS will deploy Nvidia's Vera CPUs and Nemotron open models across its cloud infrastructure. This move indicates a deeper integration of Nvidia's software and processing units, providing AWS customers with access to a broader range of Nvidia's AI ecosystem. The use of Nemotron open models suggests a commitment to flexible and accessible AI development within the AWS environment.

Amazon's Custom Silicon and AI Strategy

While expanding its partnership with Nvidia, Amazon continues to develop its custom-chip business. The company's custom silicon, including Trainium and Graviton processors, has achieved an annualized run rate of $25 billion. This demonstrates Amazon's dual strategy of both leveraging external partnerships for cutting-edge AI hardware and investing in its proprietary chip development to optimize performance and cost for specific workloads.

The growth of Amazon's custom silicon business alongside its increased commitment to Nvidia GPUs illustrates a comprehensive approach to AI infrastructure, aiming to provide diverse and powerful options for its AWS customers and internal operations. For more information on various AI tools and platforms, readers can explore AI platforms available today.

Key Takeaways

• Amazon committed to an additional 2 million Nvidia GPUs for AWS, tripling its previous order.
• Deliveries of Blackwell Ultra, Rubin, and Rubin Ultra GPUs are scheduled for 2027 and 2028.
• Nvidia reported Q2 revenue of $96.2 billion and projected Q3 revenue of $108 billion.
• Amazon will integrate Nvidia's physical AI stack (Omniverse, Cosmos, Isaac, Jetson) for its warehouse robots.
• AWS will deploy Nvidia's Vera CPUs and Nemotron open models.
• Amazon's custom chip business (Trainium/Graviton) reached a $25 billion annualized run rate.

Conclusion

The expanded partnership between Amazon and Nvidia, marked by Amazon's commitment to 3 million Nvidia GPUs for AWS, underscores the intense demand for advanced AI computing power. This collaboration not only strengthens AWS's position as a leading cloud provider for AI workloads but also highlights Nvidia's continued dominance in the AI hardware market. As both companies push the boundaries of AI development and deployment, the industry can anticipate further innovations and increased capabilities in cloud-based AI services and robotic automation.

Sources

• AWS and NVIDIA deepen strategic collaboration to accelerate AI from pilot to production | Artificial Intelligence
• Amazon just tripled its order of Nvidia chips over 'surging demand' | TechCrunch
• Nvidia is about to be a hundred-billion-dollar-a-quarter company | The Verge
• Amazon’s adding another 2 million of Nvidia’s GPUs to its AWS data centers. | The Verge