Nvidia, which leads AI computing technology, recently unveiled a large number of products and specifications, partner collaboration support, and references targeting the AI market at Computex 2023. Riding the wave of AI market expansion created by generative AI, Nvidia is making bold strides. <br/> <br/> <b>■ DGX GH200 AI Supercomputer Announcement</b> <br/> <br/> Nvidia announced the DGX GH200, a large-capacity memory AI supercomputer. The DGX supercomputer is equipped with Nvidia's GH200 Grace Hopper Superchip and Nvidia NVLink Switch System, supporting the development of next-generation large models to be used in generative AI language applications, recommendation systems, and data analytics workloads. <br/> <br/> The shared memory space of DGX GH200 combines 256 GH200 superchips using NVLink interconnect technology and the NVLink Switch System to function as a single GPU. Through this, it delivers 1 exaflop of performance and 144 terabytes of shared memory, which is nearly 500 times the memory capacity of a single Nvidia DGX A100 system. <br/> <br/> Google Cloud, Meta, and Microsoft have reportedly been among the first in the industry to adopt DGX GH200 and are exploring generative AI workload capabilities. Nvidia also plans to support cloud service providers and other hyperscalers in customizing the DGX GH200 design as a blueprint for their own infrastructure. <br/> <br/> <b>■ Grace Hopper Superchip Supplied to Japan Distributed AI Data Centers</b> <br/> <br/> Nvidia announced a collaboration with SoftBank to build platforms for generative AI and 5G/6G applications. SoftBank plans to apply Nvidia's GH200 Grace Hopper Superchip based on Arm Neoverse to the platform and deploy it to new distributed AI data centers throughout Japan...(more) <br/> <br/> #AI반도체 #챗GPT #AI서버 #AI컴퓨팅 #인공지능가속기 #반도체수요 #AI수요