10:05Machine-translated from Azerbaijani by AIRead in Azerbaijani
Amazon Triples Its Order for Nvidia Chips

Amazon and Nvidia are significantly expanding their partnership in artificial intelligence infrastructure. The companies have agreed to deploy an additional 2 million Nvidia GPUs at AWS data centers. The chips are scheduled to be delivered to Amazon Web Services (AWS) data centers in 2027 and 2028. The decision comes just five months after Amazon ordered more than 1 million Nvidia GPUs for its AWS infrastructure. Nvidia said in a statement that demand anticipated at the time subsequently exceeded forecasts. Although the financial details of the new agreement were not disclosed, given GPU prices, the deal is estimated to be worth tens of billions of dollars.
The new chips will include Nvidia’s Blackwell Ultra, Rubin, and Rubin Ultra GPUs. In addition, Nvidia’s networking equipment, CPUs, data processing software, open AI models, and robotics platforms will be integrated more extensively into the AWS ecosystem. The companies cited growing demand for computing resources from startups, large companies, artificial intelligence laboratories, and government agencies as the main reasons for expanding the partnership. Although Amazon is developing its own chips, it is not moving away from Nvidia. Interestingly, Amazon is also developing its own artificial intelligence chips to reduce its dependence on Nvidia. The company’s Trainium chips are designed for deep-learning workloads as an alternative to Nvidia GPUs such as the H100 and Blackwell. Amazon is even discussing the possibility of selling Trainium chips to other companies.
