Amazon Triples Nvidia GPU Orders to Add 2 Million Chips Across AWS Data Centers

Amazon Web Services (AWS) has dramatically expanded its procurement agreement with Nvidia, committing to deploy an additional 2 million high-performance Nvidia GPUs across its global cloud data center footprint over the next two years. The massive order comes in response to relentless enterprise demand for model training and high-throughput inference capacity.

Beyond hardware procurement, the expanded alliance includes joint co-engineering of liquid-cooled server architectures and custom networking fabrics designed to maximize multi-node GPU cluster utilization. AWS executives noted that cloud customer demand for foundation model hosting has grown exponentially year-over-year.

The multibillion-dollar expansion solidifies AWS's position as one of the world's largest commercial AI cloud providers while ensuring Nvidia's order backlog remains locked through 2027.