AMD has unveiled a new rack-scale system called Helios, designed to meet the computing demands of large AI labs, during its Advancing AI conference in San Francisco. AMD Chair and CEO Dr. Lisa Su announced that Helios will begin shipping later this year and has already attracted a clientele that includes Microsoft, OpenAI, Meta, Oracle, and Anthropic.
Helios is aimed at competing with Nvidia’s Vera Rubin and Grace Blackwell systems. Dr. Su described Helios as the tech industry’s “highest-performance AI rack,” built to handle demanding AI models at massive scale. The system will be deployed by leading AI companies, utilizing gigawatts of power, according to AMD.
Microsoft CEO Satya Nadella confirmed that the company plans to expand its Azure infrastructure with the Helios system. Furthermore, AMD announced a strategic partnership with Anthropic to deploy up to two gigawatts of GPUs using Helios.
In addition to Helios, AMD introduced its new CPU, Venice-X, designed for data centers with high-computing workloads, with an expected launch in 2027. Dr. Su projected that by 2030, chips powering AI will play a significant role in the overall computing market due to an increase in computing demand attributed to advancements in AI.
Dr. Su stated that the AI accelerator market is projected to reach approximately $1.4 trillion by 2030. She emphasized that GPUs are expected to dominate this market as AI algorithms and workloads continue to evolve, favoring programmability in silicon. “When you ask the agent to do something, it actually has dozens of steps,” she noted, explaining the complexity of AI tasks that necessitate the use of multiple GPUs.
