AMD Unveils Ryzen AI Max PRO 400 Series Processors at Beijing Media Salon
On September 10, 2026, AMD hosted a media salon in Beijing under the theme "Join Hands to Win in the Era of Agentic AI." The company unveiled the new AMD Ryzen AI Max PRO 400 series processors and showcased multiple form factors for the first time. Alongside ecosystem partners such as CSDN, Shoujie Technology, and Xingyi Technology, AMD shared applications and solutions built on the Ryzen AI Max platform, spanning developer enablement, education and research collaboration, and vertical industry deployments.
Ji Zhaohui, vice president of marketing at AMD Greater China, outlined four projects under the Ryzen AI Max agentic host initiative. The "Chunlei" program focuses on industry scenarios and has delivered more than 20 solutions; the "Chunyu" program has brought local AI computing into over 100 universities; the "Chunsun" program connects AI communities and developers; and the Ryzen AI Agent Innovation Application Competition provides a showcase for super apps and "super individuals."
Systems based on the Ryzen AI Max+ 395 processor support up to 128GB of unified memory, with up to 96GB allocatable as dedicated video memory, enabling local deployment of large models, multimodal applications, and long-context workloads. Local processing helps keep data within organizations, improving security and privacy while natively supporting the x86 Windows ecosystem.
During the showcase, Shoujie Technology presented cross-scenario applications for film production, vehicle management, and legal services; Xingyi Technology introduced an Agentic PC architecture for multi-agent collaboration; Shenghu Future demonstrated a protein design agent; Kunlunyuan presented an on-device financial agent solution; and Liulian Intelligent showcased an AI workstation and Mini AI PC based on the Ryzen AI Max+ PRO 495, with mass production expected in Q4 2026.
AMD said the new Ryzen AI Max PRO 400 series is expected to further elevate local AI computing power, enabling more developers and enterprises to access capable AI within controlled data boundaries.
