Nvidia Deepens Grip on Cloud AI With Major AWS Chip Deal
Amazon Web Services (AWS) will deploy about 1 million Nvidia GPUs through 2027 to expand its AI infrastructure, responding to rising demand for AI services. The rollout will begin this year across AWS’s global cloud regions, with Nvidia also working with AWS on networking and infrastructure for advanced AI systems. AWS continues to develop its own AI chips but is relying heavily on Nvidia hardware, especially for inference—the process of running trained AI models in real time—which now constitutes about two-thirds of AI computing demand. The inference chip market is projected to exceed $50 billion by 2026. The collaboration signals Nvidia’s move from being just a chip supplier to becoming a fundamental infrastructure partner for cloud providers. While AWS’s infrastructure gains flexibility by integrating both Nvidia and its own chips, Nvidia is embedding its technology deeper into AWS’s hardware stack. Despite developing proprietary AI chips, AWS remains dependent on Nvidia for key components, creating high switching costs. This partnership reflects broader trends in cloud-AI markets, where cooperation and competition coexist amid evolving U.S. controls on advanced chip exports.
