ASUS presented the XA P2N-E2 server at OCP Global Summit 2026 in San Jose, United States: a 2U system based on the NVIDIA MGX architecture and powered by NVIDIA Vera CPUs, aimed at agentic AI and reinforcement learning workloads. The original announcement dates from October 12, 2026, and it was highlighted again in Inven Global coverage on November 2, 2026.
What the XA P2N-E2 does
According to ASUS, the XA P2N-E2 targets the execution layer of AI applications, not only the model itself: code execution, tool use, sandboxing, data processing and orchestration. The server supports up to two dual-slot NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, combining CPU computing with GPU acceleration in a compact 2U chassis.
The manufacturer's bet is that high single-thread CPU performance and high-bandwidth, energy-efficient memory help keep AI agents responsive during demanding execution and evaluation tasks. These are ASUS claims about its own product, not results from independent tests cited in the sources consulted.
From server to full rack
At the same event, ASUS showed the rack-scale XA VR721-E3 AI POD, which integrates 36 NVIDIA Vera CPUs and 72 NVIDIA Rubin GPUs on the NVIDIA Vera Rubin NVL72 platform, with a fully liquid-cooled architecture for large-scale training, post-training and inference. ASUS also highlighted the XR R1I12B-M1BHFS system, with NVIDIA HGX Rubin NVL8 and direct-to-chip liquid cooling for the GPU baseboard, plus networking, context-memory and inference technologies such as Spectrum-6, CMX and Groq 3 LPX.
In short, the XA P2N-E2 puts the CPU at the centre of agentic AI execution, while GPUs remain responsible for acceleration. ASUS says the portfolio will be available in markets around the world, without detailing pricing in the release consulted.
Image credit: image released by ASUS and reproduced by Inven Global.
Sources: ASUS official press release, Inven Global e Unite.AI.
Transparency: this article was produced with the assistance of artificial intelligence, based on cross-checking public posts and internet sources cited above. Please consult the original sources to confirm details before making any decision.
Muse
