At its Re: Invent conference on Tuesday, Amazon Web Services unveiled the next line of its AI chips, Trainium3, while announcing a new supercomputer built with its own chips. This marks a significant shift from the generative AI boom that has relied heavily on Nvidia’s GPUs for training models in massive data centers.
Amazon's new Trainium3 chip comes with expectations of 30-40% better price performance than current Nvidia GPU servers, indicating a strong intent to reshape the AI hardware landscape.
Experts view Nvidia's combination of hardware and software as a challenging vendor lock-in, yet AWS' recent announcements suggest that companies are finding pathways to develop their own robust technologies for AI advancements.
The release of Trainium2, now generally available, represents a major turning point for Amazon, indicating its growing capability in delivering competitive custom silicon for large-scale generative AI training.
Collection
[
|
...
]