AWS Launches AI Supercomputer to Rival Nvidia
Amazon Web Services (AWS) has unveiled a new AI supercomputer, "Ultracluster," powered by its custom Trainium2 chips, aiming to challenge industry leader Nvidia. This initiative includes the "Ultraserver," designed to connect 64 Trainium2 chips, offering enhanced performance and cost efficiency for AI model training.
AWS's partnership with AI startup Anthropic will utilize this supercomputer, positioning it among the world's largest AI training systems upon its expected operational status in 2025. Additionally, AWS announced that Apple will adopt its Trainium2 chips, marking a significant endorsement of Amazon's AI hardware capabilities.
Looking ahead, AWS plans to release the next generation of its AI chips, Trainium3, in 2025, aiming to further enhance performance and maintain competitiveness in the rapidly evolving AI chip market.
These developments underscore AWS's commitment to advancing AI infrastructure, offering alternatives to Nvidia's GPUs, and meeting the growing demand for efficient AI model training solutions.
