Amazon and Nvidia have expanded their partnership with a plan to add another two million Nvidia GPUs to Amazon Web Services data centers. The chips include Blackwell Ultra, Rubin and Rubin Ultra models and are scheduled for deployment during 2027 and 2028.
The commitment follows an agreement announced five months earlier to deploy more than one million Nvidia GPUs across AWS infrastructure beginning this year. Nvidia said demand from startups, enterprises, AI labs and governments had exceeded the companies’ earlier expectations. Neither company disclosed financial terms, although the hardware commitment is likely worth tens of billions of dollars at current accelerator prices.
The partnership also covers Nvidia networking, CPUs, open models, data-processing software and robotics technology across AWS. It is growing even as Amazon develops Trainium chips as an alternative for AI workloads. The additional order secures more near-term capacity for AWS, but delivery schedules, data-center power and actual customer use will determine how quickly that hardware becomes productive cloud capacity.