AWS and NVIDIA to Deliver 2 Million Additional GPUs and Next-Generation Infrastructure for Agentic and Physical AI

Amazon Web Services (AWS) and NVIDIA have announced a major expansion of their strategic partnership to address the accelerating demand for enterprise AI infrastructure. Under the expanded agreement, AWS will deploy 2 million additional NVIDIA GPUs—including Blackwell Ultra, Rubin, and Rubin Ultra architectures—across its global cloud infrastructure between 2027 and 2028. This follows a previous commitment to add over 1 million GPUs starting in 2026. The collaboration extends across the full hardware and software stack. Key initiatives include the integration of NVIDIA Vera CPUs on AWS, expanding NVIDIA NVLink Fusion with custom high-bandwidth memory (NVHBM) for Amazon's next-generation Trainium chips, and building dedicated AI factories delivering 100,000 GPUs for U.S. government national security workloads. Additionally, the partnership accelerates Amazon Robotics via NVIDIA's physical AI platforms. This deployment solidifies NVIDIA's dominance in cloud computing infrastructure while reinforcing AWS's enterprise AI capabilities against competing hyperscalers, signaling sustained long-term capital expenditure and hardware demand across the generative AI ecosystem.