Advanced Micro Devices (AMD) unveiled its next-generation artificial intelligence server, “Helios,” as part of an ambitious push to challenge Nvidia’s dominance in the AI hardware market. CEO Lisa Su introduced the MI350 and MI400 series chips during the “Advancing AI” developer conference in San Jose, California, highlighting their role in the 2026 server rollout.
The Helios server, powered by AMD’s MI400 chips, is designed to rival Nvidia’s Blackwell-based systems and marks a shift from individual chip sales to fully integrated server systems. These high-performance servers will include networking components and offer open standards — a contrast to Nvidia’s proprietary NVLink technology.
OpenAI CEO Sam Altman joined Su onstage and confirmed the adoption of AMD’s MI300X and MI450 chips. Other tech giants, including Meta, Oracle, and Elon Musk’s xAI, also endorsed AMD’s processors, signaling broader industry support.
Crusoe, a specialized AI cloud provider, announced plans to invest $400 million in AMD chips. Meanwhile, AMD continues to expand its capabilities with the acquisition of server builder ZT Systems and over 25 AI-related firms in the past year, including key hires from Untether AI and Lamini.
Despite its ROCm software stack trailing Nvidia’s CUDA in adoption, AMD aims to deliver competitive hardware and complete AI systems, bolstering its position in a booming market. Su reiterated expectations of strong double-digit growth in AI chip revenue, even amid U.S. export restrictions to China.
Swifteradio.com
