AWS and NVIDIA plan to deploy two million additional GPUs by 2028

AWS and NVIDIA plan to deploy two million additional NVIDIA GPUs across AWS infrastructure during 2027 and 2028, the companies announced on August 26.

The companies also announced plans for U.S. government AI factories with 100,000 GPUs on secure AWS infrastructure for federal and national-security workloads classified at Impact Level 6 and above. They did not say whether those GPUs are included in the two-million total or would be additional. The government focus provides infrastructure context for DataPhoenix’s coverage of NVIDIA Nemotron deployments in air-gapped U.S. government systems.

Both figures describe capacity the companies intend to add, not hardware already delivered or installed. AWS and NVIDIA disclosed no financial terms, year-by-year or regional allocation, delivery milestone, or completed deployment count. They also did not identify which U.S. agencies or programs would use the government infrastructure or where it would be located.

AWS said the two-million-GPU deployment will include NVIDIA Blackwell Ultra, Rubin, and Rubin Ultra chips. NVIDIA’s release gives the same figures and timeframe.

AWS attributed the expansion to customer demand running ahead of its expectations but published no audited demand measure with the announcement.

“Customers want the freedom to choose the best tools for their AI workloads, and they want confidence that everything works seamlessly together,” said Matt Garman, CEO of AWS.

The expanded collaboration also covers Vera CPU-based infrastructure, NVLink Fusion with NVIDIA custom high-bandwidth memory, integration with the AWS Nitro system and Elastic Fabric Adapter networking, Nemotron open models, GPU-accelerated data processing and vector indexing, and robotics technology.

“For 16 years, we have scaled NVIDIA computing in the cloud together. Now we are expanding our partnership across the full stack—GPUs, CPUs, networking, open models and software,” said Jensen Huang, founder and CEO of NVIDIA.

Amazon said the accelerated data stack delivers up to 3.7 times faster Amazon EMR processing, 30% better price-performance, and up to nine times faster vector indexing. Those are company-reported comparisons; neither company released independent benchmark results with the announcement.

AWS said in March 2026 that it would add more than one million NVIDIA GPUs across its global regions starting in 2026. That earlier figure was also announced capacity, and the companies have not reported how much of it has been delivered.