AWS and NVIDIA announced on August 26, 2026 that they plan to deploy two million additional NVIDIA GPUs across AWS infrastructure in 2027–2028. The announcement expands their collaboration as customers move more AI workloads into production.
The planned deployment includes Blackwell Ultra, Rubin and Rubin Ultra GPUs. It follows an earlier commitment to add more than one million NVIDIA GPUs starting in 2026. The companies also outlined work on Vera CPU infrastructure, networking, open models and robotics. These are forward-looking deployment plans, rather than confirmation that the additional capacity is already available.
Meridian’s working view
For infrastructure suppliers, the headline number is a signal to plan beyond individual accelerator shipments. Compute systems arrive within a wider delivery program involving racks, network equipment, power distribution and cooling. Each component must reach a site in a sequence the installation team can use.
Project schedules should connect supplier release dates with site access, receiving capacity and commissioning milestones. High-value equipment also calls for clear handling instructions and documented handoffs between origin, transport, storage and final delivery.
The practical question is how deployment forecasts translate into confirmed shipment windows. Procurement and logistics teams can use those windows to identify where consolidation, secure staging or specialist delivery resources may be needed.