Amazon and Nvidia expand AI collaboration with new chip orders and tech integration

Amazon and Nvidia have announced a significant expansion of their partnership, with Amazon committing to add another 2 million Nvidia GPU chips to its data centers. These GPUs, including the Blackwell Ultra, Rubin, and Rubin Ultra models, will be deployed in AWS data centers starting in 2027 and 2028. The deal, announced during Nvidia’s quarterly earnings call, follows a previous agreement to deploy over 1 million Nvidia GPUs across AWS infrastructure, which Nvidia stated had exceeded initial demand expectations. The partnership extends beyond chip procurement, with Nvidia’s technology, including networking hardware, open models, CPUs, data processing software, and robotics platforms, being integrated across AWS. This move is attributed to surging demand from startups, enterprises, AI labs, and governments. Amazon, meanwhile, continues to invest in its own AI chip development, particularly with CPUs, as it seeks to reduce reliance on Nvidia and compete in the market. Nvidia’s CEO, Jensen Huang, highlighted the growing importance of AI in generating profitable tokens, emphasizing the need for increased compute power. The company reported $96.2 billion in Q2 sales, with data center revenue reaching $89 billion, up 117% year-over-year. Nvidia expects third-quarter revenue to hit $108 billion, partly driven by its next-gen Rubin GPUs, which are already in production shipments. Nvidia has also committed $279 billion to secure supply and manufacturing capacity for data-center projects, up from $119 billion in the previous quarter. The company plans to deploy its Vera CPUs, some integrated with Rubin GPUs and others standalone, to major hyperscalers, neoclouds, AI labs, and system OEMs, including Oracle and SpaceXAI. The partnership also includes Amazon’s adoption of Nvidia’s full physical AI stack for warehouse robots, encompassing platforms like Omniverse, Cosmos, Isaac, and Jetson. On the enterprise side, AWS will serve Nvidia’s Nemotron family of open models on Amazon Bedrock and SageMaker. Nvidia’s CFO, Colette Kress, noted that the partnership reflects the industry’s shift toward leveraging AI for productive and useful work, with investors closely watching whether increased compute translates into additional profits as AI infrastructure investments grow.