Nvidia to sell 1 million chips to Amazon by end of 2027 in cloud deal
Massive cloud deal secures AWS's AI infrastructure with Nvidia's latest chips, starting deliveries this year.
Nvidia has secured a landmark agreement to supply 1 million of its cutting-edge graphics processing units (GPUs) to Amazon Web Services, with deliveries scheduled to commence this year and continue through 2027. The deal, confirmed by Nvidia's vice president of hyperscale computing, Ian Buck, represents one of the largest single procurements of AI accelerator chips to date. It solidifies AWS's commitment to building out massive, Nvidia-powered infrastructure for cloud-based artificial intelligence, ensuring it can compete in providing the raw computational horsepower required for training frontier AI models.
This strategic supply agreement goes beyond just hardware. It encompasses "a host of the AI giant's other offerings," likely including Nvidia's full AI enterprise software stack, networking solutions like InfiniBand, and its DGX Cloud supercomputing platform. For AWS customers, this translates to imminent, scalable access to the latest Nvidia architectures (such as the Blackwell platform) for developing and deploying large language models, generative AI applications, and advanced simulations. The scale of the order underscores the immense and growing demand for AI compute, positioning both Nvidia as the indispensable enabler and AWS as a primary conduit for this critical resource in the coming years.
- Nvidia will supply 1 million GPUs to Amazon Web Services in a deal extending through 2027.
- Deliveries under the agreement are scheduled to begin before the end of this year.
- The procurement includes Nvidia's GPUs plus associated AI software and platform offerings.
Why It Matters
Secures critical AI compute supply for AWS, enabling massive scale for enterprise AI training and inference in the cloud.