Amazon and Nvidia deepened their partnership Wednesday, agreeing to add 2 million Nvidia GPUs—including Blackwell Ultra, Rubin, and Rubin Ultra—to Amazon Web Services’ data centers by 2027 and 2028. The move follows Amazon’s earlier commitment to deploy over 1 million Nvidia GPUs this year, a deal Nvidia now says has already exceeded expectations due to “surging demand” from startups, enterprises, AI labs, and governments.
The expanded collaboration goes beyond chip orders. Nvidia’s networking hardware, open models, CPUs, data processing software, and robotics platforms will integrate across AWS, while Amazon’s own AI chip efforts—including Trainium and Graviton—continue to grow. Amazon reported its custom chip business reached a $25 billion annualized revenue run rate, driven by commitments from AI labs like Anthropic and OpenAI.
Nvidia also revealed plans to ship Vera CPUs to AWS and other hyperscalers, with shipments already underway to partners like Oracle and SpaceXAI. The companies highlighted Amazon’s adoption of Nvidia’s full AI stack for warehouse robots and enterprise tools, including Omniverse, Cosmos, Isaac, and Jetson. Nvidia posted $96.2 billion in sales for Q2, with data center revenue hitting $89 billion, up 117% year-over-year.
