"I'm not aware of anyone using AWS chips in any sort of large volumes," Rasgon told Business Insider, referring to Amazon's AI chips.
This time, the idea is to avoid paying for expensive Nvidia GPUs, while still providing cloud customers with powerful AI services.
An obvious response to this would be to have cloud customers use Amazon's AI chips instead.
However, some of the largest AWS customers have not been willing to use these homegrown alternatives, the documents said.
For example, AWS's AI chips still have "compatibility gaps" in certain open-source frameworks, making Nvidia GPUs a more popular option.
Persons:
—, Stacy Rasgon, Bernstein, I'm, Rasgon, Adam Selipsky, Jensen Huang, Andy Jassy, Inferentia, Trainium, Omdia, Snowflake's, Sridhar Ramaswamy, Eóin Noonan, CUDA, Ramaswamy, James Hamilton, Jassy, Gartner, Huang
Organizations:
Service, Microsoft, Google, Business, Services, Nvidia, Intel, Amazon, NVIDIA CUDA, Netflix, Neuron, AWS, NVIDIA, Amazon VP, James Hamilton Amazon, BI
Locations:
CUDA, Toronto