Key information
At the OCP Global Summit 2026, ASUS introduced the XA P2N-E2, a 2U server utilizing NVIDIA MGX architecture. This system is powered by NVIDIA Vera CPUs and is specifically engineered to handle agentic AI and reinforcement learning tasks, including data orchestration, sandboxing, and tool usage.
The XA P2N-E2 platform offers hardware flexibility by accommodating as many as two dual-slot NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. ASUS designed the server to leverage high-bandwidth memory and strong single-thread CPU performance to ensure responsiveness during complex evaluation and execution processes.
Beyond this individual server, ASUS showcased a wide range of infrastructure solutions, including the XA VR721-E3 AI POD. This rack-scale system integrates 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs within a fully liquid-cooled environment, intended for large-scale inference and training.
ASUS servers are currently available globally, though the company notes that specific product configurations and regional availability may differ. Interested parties can view the full portfolio at Booth #B9 during the OCP Global Summit 2026.
