Amazon has agreed to purchase two million Nvidia GPU chips to expand its cloud computing data centers over the next two years. The agreement includes shipments of Nvidia Blackwell Ultra, Rubin, and Rubin Ultra processors scheduled for deployment across Amazon Web Services infrastructure in 2027 and 2028, according to TechCrunch.
The expansion follows an earlier agreement reached five months ago, in which Amazon committed to deploying more than one million units. Nvidia stated during its quarterly earnings report that rising demand from enterprises, startups, and research institutions required a substantial increase in hardware allocations.
Surging Compute Demand Drives Chip Orders
Rising interest in advanced artificial intelligence workloads has increased the necessity for specialized processing hardware. As organizations build larger models, data centers require high-density compute systems to handle training and inference workloads efficiently.
“AI is now doing productive and useful work. If we had more compute, we could generate more profitable tokens, which results in more profit for all of the services.”
Jensen Huang, CEO of Nvidia
Hardware Delivery of Nvidia GPU chips
The deployment of Nvidia GPU chips will be paired with Nvidia Vera central processing units across Amazon data centers. Nvidia Chief Financial Officer Colette Kress confirmed that some Vera units will be integrated directly with Rubin processors, while others will operate as standalone hardware.
In addition to compute processors, the partnership encompasses networking hardware that connects thousands of processors into unified clusters. Amazon will also host Nvidia Nemotron open foundation models on Amazon Bedrock and SageMaker for commercial cloud customers.
Integration With Warehouse Robotics Systems
Beyond traditional server operations, Amazon plans to integrate Nvidia robotics platforms into its logistics infrastructure. This deployment includes the Omniverse simulation system, the Cosmos physical world model platform, the Isaac robotics development framework, and Jetson computing modules for edge automation.
These tools will help manage logistics and autonomous systems across commercial facilities. Meanwhile, Amazon continues to invest in its proprietary custom silicon, including Trainium processors and Arm-based Graviton server chips, which recently reached a $25 billion annualized revenue run rate.
Financial Results and Future Hardware Commitments
Nvidia reported total second-quarter revenue of $96.2 billion, with data center operations contributing $89 billion, representing a 117% increase year over year. The chipmaker committed $279 billion to secure manufacturing capacity, ensuring reliable delivery of enterprise computing infrastructure, including future allocations of Nvidia GPU chips for hyperscale operators worldwide.
