AWS Plans 2 Million Additional NVIDIA GPUs by 2028: What It Means for Cloud Investors

Amazon Web Services (AWS) and NVIDIA announced on August 26, 2026 that they are expanding their strategic partnership: during 2027 and 2028, 2 million additional NVIDIA GPUs will be deployed across AWS global infrastructure. The announcement appeared on the official channels of both companies and was picked up by several industry publications between August 26 and September 4, 2026.

Scope and Timeline of GPU Deployment

The deployment includes GPUs from the Blackwell Ultra, Rubin, and Rubin Ultra platforms. This expansion follows an already-announced plan to deploy over 1 million NVIDIA GPUs starting in 2026. The doubling of capacity announced within a few months demonstrates the speed at which demand for AI computing power is growing.

A central component of the expansion is the establishment of AI-Factories for the U.S. Government. This infrastructure with 100,000 GPUs operates on secure AWS infrastructure and is specifically designed for federal and national security workloads. The announcement from August 26, 2026 names Frontier Labs, global enterprises, startups, and governments as target users for the infrastructure.

Technology Integration and New Components

Beyond the sheer number of GPUs, AWS is integrating several new NVIDIA technologies into its cloud platform. The NVIDIA Vera CPU-based infrastructure will be made available on AWS for the first time. NVIDIA NVLink Fusion will be expanded with customized NVIDIA High-Bandwidth Memory (NVHBM) and integrated into the AWS Nitro System and Elastic Fabric Adapter (EFA).

On the software side, NVIDIA Nemotron Open Models will continue to be supported on Amazon Bedrock and Amazon SageMaker. Customers thus gain expanded choices in open AI models. For data processing and vector indexing, AWS is accelerating Amazon EMR and Amazon OpenSearch services through NVIDIA cuDF and cuVS CUDA-X libraries.

Physical AI and Robotics

Amazon Robotics is adopting NVIDIA's physical AI platform to advance robotics workloads. The integration aims to accelerate innovation in warehouse automation and next-generation robots. This component of the partnership connects cloud computing with concrete industrial applications within the Amazon ecosystem.

Strategic Context: 16 Years of Partnership

AWS and NVIDIA have been working together for 16 years. Matt Garman, CEO of AWS, explained in the announcement: "Customers want the freedom to choose the best tools for their AI workloads, and they want the assurance that everything works together seamlessly. That's why we've invested heavily with NVIDIA to make AWS the best place for NVIDIA AI technologies."

The expansion responds to a shift in customer behavior: according to the official announcement from August 26, 2026, customers are increasingly in the transition phase from pilot projects to production. Workloads are scaling in areas such as agentic AI (AI systems that perform actions independently), scientific discovery, enterprise automation, and robotics. Customers need broader model selection, faster data pipelines, and new capabilities for physical AI applications—while maintaining security and reliability for business-critical workloads.

Assessment for Investors

The announcement offers several perspectives for investors in the cloud and semiconductor sector. For AWS, the expansion demonstrates a willingness to keep pace or pull ahead in the competition for AI workloads against Microsoft Azure and Google Cloud Platform. The concrete order for 2 million GPUs—in addition to the already-committed 1 million—signals confidence in commercial demand.

For NVIDIA, the agreement secures a portion of production volume for the upcoming Blackwell and Rubin generations. The integration of NVIDIA Vera CPUs in AWS expands the product portfolio beyond pure GPUs and positions NVIDIA more broadly in the data center market. Collaboration on software (Nemotron models) and specialized libraries (CUDA-X) deepens ecosystem lock-in between the two companies.

The establishment of secure infrastructure for U.S. government customers opens a market that has traditionally imposed high compliance and certification requirements. This segment can provide long-term, stable revenue streams but requires specific investments in security architectures.

Broader Market and Competition

The partnership expansion between AWS and NVIDIA fits into a broader trend: all major cloud providers are ramping up their AI computing capacity. Microsoft has announced its own procurement agreements with NVIDIA as well as proprietary AI chips (Azure Maia). Google Cloud is combining NVIDIA GPUs with its own Tensor Processing Units (TPUs). Competition for computing power and model diversity is intensifying alongside rising demand.

For retail investors in the DACH region, it is important to note that both Amazon (AMZN) and NVIDIA (NVDA) are listed on U.S. exchanges. Currency risks between U.S. dollars and euros or Swiss francs can impact returns. The valuations of both companies already reflect high expectations for AI-driven growth—future price movements will depend on whether actual demand meets or exceeds these expectations.

The announcement from August 26, 2026 does not include information on investment volume, pricing, or concrete revenue forecasts for the GPU deployment. Without these figures, the direct financial impact on both companies remains difficult to quantify. The strategic significance—AWS secures computing capacity, NVIDIA secures sales—is nonetheless evident.