Amazon just tripled its order of Nvidia chips over 'surging demand'
Amazon has expanded its partnership with Nvidia by ordering an additional 2 million GPUs for its Amazon Web Services data centers across 2027 and 2028. The deal follows an agreement five months prior to deploy over 1 million units, driven by heavy customer demand across commercial, startup, and governmental sectors. Beyond compute hardware, AWS will also integrate Nvidia networking tech, open models, robotics development platforms, and software solutions.
Key Takeaways
- Amazon Web Services announced a massive expansion of its hardware agreement with Nvidia, adding 2 million GPU chips scheduled for installation in 2027 and 2028.
The upcoming hardware allocation includes Nvidia Blackwell Ultra, Rubin, and Rubin Ultra chips.
- Amazon will also receive Nvidia Vera CPUs to complement these data center deployments.
The collaboration stretches far beyond standard server hardware procurement into robotics and software infrastructure.
- For developers, this broad integration underscores how hardware vendors and cloud service providers are building tightly coupled ecosystems that combine custom silicon, networking, and foundation models.
This expanded purchase occurs alongside Amazon's ongoing efforts to build its proprietary silicon and lower its reliance on third-party manufacturers.
- Amazon agreed to add 2 million Nvidia Blackwell Ultra, Rubin, and Rubin Ultra GPUs to AWS infrastructure in 2027 and 2028.
The expansion comes just five months after an initial order of over 1 million units failed to keep pace with surging market demand.
- Despite buying more Nvidia hardware, Amazon continues to expand its own internal chip line, including Trainium and Graviton processors.
Stats & Key Facts
- #Amazon has expanded its partnership with Nvidia by ordering an additional 2 million GPUs for its Amazon Web Services data centers across 2027 and 2028.
- #The deal follows an agreement five months prior to deploy over 1 million units, driven by heavy customer demand across commercial, startup, and governmental sectors.
- #Amazon Web Services announced a massive expansion of its hardware agreement with Nvidia, adding 2 million GPU chips scheduled for installation in 2027 and 2028.
- #This decision builds upon a deal struck five months earlier for over 1 million GPUs, as Nvidia noted that customer interest significantly outstripped original projections.
Amazon Web Services announced a massive expansion of its hardware agreement with Nvidia, adding 2 million GPU chips scheduled for installation in 2027 and 2028. The upcoming hardware allocation includes Nvidia Blackwell Ultra, Rubin, and Rubin Ultra chips. This decision builds upon a deal struck five months earlier for over 1 million GPUs, as Nvidia noted that customer interest significantly outstripped original projections.
Amazon will also receive Nvidia Vera CPUs to complement these data center deployments. The collaboration stretches far beyond standard server hardware procurement into robotics and software infrastructure. Amazon plans to integrate Nvidia's full physical AI stack, such as Omniverse, Cosmos, Isaac, and Jetson, to power its automated warehouse systems.
Additionally, AWS will offer Nvidia's Nemotron open models directly through its Bedrock and SageMaker platforms. For developers, this broad integration underscores how hardware vendors and cloud service providers are building tightly coupled ecosystems that combine custom silicon, networking, and foundation models. This expanded purchase occurs alongside Amazon's ongoing efforts to build its proprietary silicon and lower its reliance on third-party manufacturers.
For more details please read the original article at TechCrunch AI.
Continue Learning
Comments
Comments appear only after moderation. Your email identifies your submission to the moderator and is never displayed here.
No approved comments yet.