Amazon just tripled its order of Nvidia chips over ‘surging demand’

By Pradhyuman, August 27, 2026

Amazon and Nvidia announced an expanded partnership to add 2 million Nvidia GPU chips to Amazon Web Services data centers in 2027 and 2028. The order includes Nvidia Blackwell Ultra and Rubin GPUs. This deal follows an agreement five months ago where Amazon planned to deploy 1 million GPUs. Nvidia stated that customer demand exceeded those original expectations.

expanded integration and custom chips

The partnership extends to other areas. Amazon plans to use Nvidia's physical AI stack, including the Omniverse and Isaac platforms, to power its warehouse robots. AWS will also run Nvidia's Nemotron open models on its Bedrock and SageMaker platforms. This expansion occurs while Amazon develops its own Trainium and Graviton chips. Nvidia CFO Colette Kress stated that Nvidia also plans to send Vera CPUs to Amazon, with some integrated into the Rubin chips.

Maintained by Pradhyuman.

Filed under: Nvidia, Amazon, Chips & Infrastructure

AI infrastructure company Cornelis raises $205M to chip away at Nvidia’s dominance September 14, 2026
Cornelis announced on Monday that it raised $205 million in a funding round led by IAG Capital Partners. The company spun off from Intel in 2020 and creates networking technology for artificial...
Nvidia’s AI advantage is moving beyond the GPU August 30, 2026
Nvidia hardware developments for data centers TechCrunch reported that Nvidia faces GPU competition from companies like Amazon and Google. In response, Nvidia rolls out its Vera Rubin architecture, which pairs the Rubin...
Nvidia-backed SiFive hits $3.65 billion valuation for open AI chips April 12, 2026
Remember when it felt like Intel and ARM completely dominated the chip conversation? Well, Nvidia just made a move that could rewrite the entire playbook, and if you're not paying attention, you're...
Amazon releases its own Jev clone as decision models flood the web October 1, 2026
Amazon Web Services released Strands Decider 2B, an open-source decision model from its Strands Labs group. The model uses the Qwen3.5-2B architecture and runs locally on devices. AWS distinguished engineer Marc Brooker...

Amazon just tripled its order of Nvidia chips over ‘surging demand’

expanded integration and custom chips

Related articles