▲ Amazon Data Center
Nvidia announced on the 26th local time that it will supply an additional 2 million Nvidia graphics processing units (GPUs) to Amazon Web Services (AWS) global infrastructure between 2027 and 2028.
In a press release issued that day, Nvidia stated, "We are significantly expanding our strategic cooperation to respond to the soaring global demand for artificial intelligence (AI) infrastructure."
This volume is in addition to Amazon's previous announcement that it would install 1 million Nvidia chips in AWS data centers starting this year.
The chips included in this contract are Nvidia's Blackwell Ultra, Rubin, and Rubin Ultra GPUs.
While Amazon and Nvidia did not disclose the specific financial terms of the contract, the total amount is estimated to reach at least tens of billions of dollars, given that the list price per unit for these products is at least tens of thousands of dollars.
Nvidia GPUs are serving as a key component of the AI boom, supporting the building and operation of the vast majority of large language models (LLMs).
At the same time, customer companies such as Amazon, Microsoft, and Google are also developing their own chips to compete with Nvidia.
These so-called hyperscalers, or large-scale data center operators, believe that by building hardware optimized for their own data centers, they can pass efficiency gains back to their customers.
Amazon has invested heavily in recent years in Annapurna Labs, its in-house semiconductor development division.
This division produces central processing units (CPUs) as well as AI accelerators designed to rival Nvidia's GPUs.
Amazon has previously announced plans to utilize Nvidia's networking technology for the next version of its proprietary AI chip, Trainium.
(Photo: Getty Images)
※
Copying, redistribution, and unauthorized use in AI training are strictly prohibited.