According to Reuters, engineers at Amazon's chip lab in Austin, Texas, recently tested the secret new servers. According to the Economic Times, Rami Sino, engineering director at Amazon's AWS subsidiary Annapurna Research Center, revealed that these new servers are equipped with Amazon's AI chips that can compete with Nvidia's chips.
Amazon is reportedly developing processors that will reduce reliance on expensive NVIDIA chips for some of its AWS AI cloud services.
Amazon hopes that its chips will enable customers to perform complex calculations and crunch huge amounts of data at low cost, as rivals Microsoft and Alphabet do.
But while Amazon is a latecomer to the AI chip space, it is an industry leader in non-AI processing chips: its flagship non-AI processing chip, Graviton, has been in development for nearly a decade and is now in its fourth generation, while its other two AI chips, Trainium and Inferentia, are relatively new designs.
David Brown, vice president of computing and networking at AWS, said that in some cases the performance of these chips could be 40% to 50% higher than Nvidia's, and the cost should be about half that of the same model of Nvidia's chip.
AWS accounts for about 20% of Amazon's total revenue. The company's January-March revenue rose 17% year over year to $25 billion. AWS controls about a third of the cloud computing market, followed by Microsoft's Azure with about 25%.
Amazon announced that it has deployed 250,000 Graviton chips and 80,000 custom AI chips to handle the surge in platform activity during the recent Prime Day.
read more
(Photo courtesy of Amazon)
This article cites the following sources: Economic Daily News and WeChat account DRAM replacement.
Next Article
[News] TSMC's N3 process is stimulating demand, and supply chain is expected to remain active through next year
