AWS To Deploy Two Million More Nvidia GPUs Globally In 2027-28
businessworld.in Aug 27, 2026

AWS To Deploy Two Million More Nvidia GPUs Globally In 2027-28

AI-summarised brief · reviewed before publication

Amazon Web Services and Nvidia announced plans to deploy two million additional Nvidia GPUs across AWS’s global cloud infrastructure during 2027 and 2028. This expansion, featuring Blackwell Ultra, Rubin, and Rubin Ultra chips, responds to AI computing demand that has outpaced earlier forecasts. AWS CEO Matt Garman stated customers are shifting from experimentation to large-scale production workloads. The partnership also includes building AI factories for the US government, with infrastructure equipped with 100,000 GPUs for federal and national-security tasks. This move extends a 16-year collaboration, adding Nvidia’s Vera CPUs and NVLink Fusion technology. The capacity will support agentic AI, scientific research, and enterprise automation. AWS previously planned to add one million GPUs starting in 2026, but increased demand necessitated this larger scale. The integration aims to enhance security, reliability, and performance through AWS’s Nitro System and Elastic Fabric Adapter.

💡 Why It Matters

  • · The dedicated 100,000-GPU infrastructure for federal workloads marks a critical pivot in securing national AI capabilities against geopolitical rivals.
  • · This specific government focus transforms cloud computing from a commercial utility into a strategic national security asset.