Core Insights - AMD's CEO Lisa Su introduced the Helios rack-level platform at CES 2026, featuring the Venice CPU built on TSMC's 2nm process and the latest Instinct MI455X accelerator, set to launch later this year [1][2] - The next-generation Instinct MI500 accelerator is in development, also utilizing 2nm technology, expected to enhance AI performance by 1000 times compared to the MI300X launched in 2023 [1] - The global AI user base has surged from 1 million to over 1 billion since the launch of ChatGPT, with projections of 5 billion users in the next five years, necessitating a 100-fold increase in computing power to support advanced AI capabilities [1] Group 1 - The Helios platform is described as the best AI rack platform globally, developed in close collaboration with major clients, featuring liquid cooling and support from the Instinct MI455X accelerator, EPYC Venice CPU, and Pensando Vulcano NIC [2] - The MI455 series accelerators are revolutionary, achieving a tenfold performance increase over the MI355 launched six months prior, enabling developers to create larger models and more powerful applications [2] - AMD's data center product roadmap includes the Helios platform, the Instinct MI440X GPU designed for enterprise AI deployment, and the Instinct MI430X platform for precision-dependent autonomous AI and supercomputing solutions [2]
AMD最新AI CPU台积电助力苏姿丰秀Helios机架级平台