Amazon and Nvidia have deepened their collaboration with a major new agreement that will bring an additional 2 million GPU chips to Amazon Web Services data centers. The chips, including Nvidia's Blackwell Ultra, Rubin, and Rubin Ultra generations, are scheduled to arrive in 2027 and 2028.
The expansion follows a previous plan announced just months ago to deploy more than 1 million Nvidia GPUs across AWS infrastructure. Nvidia said demand has climbed beyond early expectations, pushing the two companies to scale up the partnership faster than planned.
Beyond GPUs, the agreement also broadens the use of Nvidia's wider technology stack across AWS. That includes networking systems, CPUs, data processing tools, open models, and robotics software. On the enterprise side, AWS will host Nvidia's Nemotron open models through Bedrock and SageMaker.
The deal also reflects the growing importance of AI infrastructure for startups, enterprises, research labs, and public institutions. At the same time, Amazon continues to invest in its own chips, including Trainium and Graviton, signaling a dual strategy that combines in-house innovation with external partnerships.
Nvidia also highlighted strong financial momentum, reporting $96.2 billion in quarterly sales, with data center revenue accounting for the largest share. The company expects further growth as its next-generation Rubin platform begins shipping.
As AI workloads expand, partnerships like this are shaping the next era of cloud computing, robotics, and large-scale digital services.
