Amazon and Nvidia deepen AI collaboration with expanded GPU orders and tech integration

Amazon and NVIDIA have significantly expanded their strategic partnership, with NVIDIA's GPU orders tripling following the deployment of over 1 million GPUs in Amazon's data centers. The companies plan to add another 2 million NVIDIA GPU chips, including the latest Blackwell Ultra, Rubin, and Rubin Ultra architectures, which are expected to be deployed into AWS data centers between 2027 and 2028. The financial value of the deal is estimated to be in the hundreds of billions of dollars, driven by the overwhelming demand for computing power from startups, large enterprises, top AI labs, and governments. NVIDIA highlighted that the rapid finalization of the deal was due to this unprecedented demand. Beyond chip procurement, NVIDIA is integrating its full technology stack into AWS, including network hardware for super clusters, open-source models, CPUs, data processing software, and robot platforms. The NVIDIA Vera CPU will be delivered to AWS, with some units integrated with the Rubin architecture and others used as standalone processors. Amazon also plans to adopt NVIDIA's physical AI technology stack in its warehouse robots and enterprise operations, including the Omniverse digital twin platform, Cosmos world model platform, Isaac robot development platform, and Jetson edge computing hardware. AWS's Amazon Bedrock and SageMaker platforms will also introduce NVIDIA's Nemotron series of open-source large models. Meanwhile, Amazon is advancing its own chip strategy to reduce reliance on NVIDIA. Its Arm-based Graviton series CPU and the Trainium chip, designed for deep learning workloads, are seen as direct competitors to NVIDIA's H100 and Blackwell chips. Amazon executives noted that the total commitment of 22.5 billion USD from AI labs like Anthropic and OpenAI has already made the annual recurring revenue run rate of its self-developed chip business exceed 25 billion USD. NVIDIA's financial performance remains robust, with recent quarterly revenue reaching 9.62 billion USD, surpassing analyst expectations. The data center business contributed 8.9 billion USD in sales, a 117% increase compared to the same period last year. For the third quarter, NVIDIA expects revenue to rise to 10.8 billion USD, including initial shipments of the next-generation Rubin GPU. To meet growing AI hardware demand, NVIDIA has increased its supply chain and manufacturing commitments to 27.9 billion USD, up from 11.9 billion USD in the previous quarter. This includes 9.2 billion USD in planned expenditures for the remainder of the fiscal year and 8.7 billion USD for the 2028 fiscal year. NVIDIA CEO Huang Renxun emphasized during the earnings call that the industry's core logic has shifted, with AI now generating actual productivity and creating substantial profits through 'tokens.' He stated that more computing power translates to more high-value profits, which is why tech giants are heavily investing in AI infrastructure development.