Amazon Web Services (AWS), the cloud computing arm of Amazon.com, clarified on Tuesday that it has not halted any orders for Nvidia’s advanced chips. Instead, the company has opted to acquire Nvidia’s latest cutting-edge chips for a collaborative supercomputer project. This statement came in response to a report by the Financial Times, which suggested that AWS had completely shifted its orders from Nvidia’s Grace Hopper chips to the newer Blackwell graphics processing units (GPUs).
The Financial Times reported that AWS had “fully transitioned” its previous orders for Nvidia’s Grace Hopper chip to its newer Blackwell GPUs. The Blackwell chips, announced by Nvidia in March, represent a significant advancement in GPU technology. This shift initially led people to perceive that AWS was moving away from the Grace Hopper chips entirely, raising questions about the status of existing orders.
AWS’s Official Statement
In response to the report, an AWS spokesperson clarified the situation to Reuters. The spokesperson emphasized that the transition from Grace Hopper chips to Blackwell chips is specific to Project Ceiba, a supercomputer project that AWS and Nvidia are jointly developing. AWS continues to utilize and offer services based on Nvidia’s Hopper chips for other applications, particularly in the realm of artificial intelligence (AI) training.
“To be clear, AWS did not halt any orders from Nvidia. In our close collaboration with Nvidia, we jointly decided to move Project Ceiba from Hopper to Blackwell GPUs, which offer a leap forward in performance,” the AWS spokesperson said. This statement underscores the ongoing partnership between AWS and Nvidia and the strategic decision to enhance the performance of Project Ceiba with the latest technology.
Project Ceiba Announcement
The transition of Project Ceiba to Blackwell GPUs was publicly announced in March, at the same time Nvidia unveiled the new Blackwell chips. This move was made to take advantage of the superior capabilities of the Blackwell GPUs, which promise improved performance and efficiency for the supercomputer project.
The AWS spokesperson further clarified that AWS did not cancel any Nvidia chip orders. Instead, the company made a deliberate decision, in collaboration with Nvidia, to upgrade Project Ceiba from Hopper to Blackwell GPUs to harness the enhanced performance benefits. Following this clarification, the Financial Times accurately reflected in its amended story that Amazon had not placed any chip orders. The newspaper updated its article to correct the misunderstanding and directed readers to the revised story.
AWS’s clarification highlights the company’s ongoing commitment to leveraging the best available technology in its projects. By transitioning Project Ceiba to Nvidia’s Blackwell GPUs, AWS aims to ensure the highest performance standards for its supercomputer initiative while continuing to support other AI training services with Nvidia’s Hopper chips. This decision reflects AWS’s strategic approach to innovation and collaboration with leading technology partners like Nvidia.
