Amazon Web Services (AWS) announced plans to build a powerful new supercomputer called “Trainium” at its annual re:Invent conference. The supercomputer, expected to be operational by 2024, will be powered by Nvidia’s latest “Transformer Engine” chips and AWS’s custom-designed “Graviton” processors. It aims to provide advanced computing capabilities for AI workloads like large language models and generative AI. Trainium will offer over 1 exaflop of computing power, making it one of the fastest supercomputers globally. AWS also unveiled “Nova,” a new AI instance powered by Nvidia’s H100 GPUs, designed for training large language models and other AI workloads. The announcements highlight AWS’s focus on expanding its AI and high-performance computing offerings to meet the growing demand from customers across industries.