ASUS has made a splash at the GTC 2025, the premier event for artificial intelligence and accelerated computing organized by NVIDIA, as a Diamond sponsor. The company seized the opportunity to introduce its latest innovation: the ASUS AI POD, based on the NVIDIA GB300 NVL72 platform, and confirmed it has already received significant orders, marking a milestone in the tech industry.
A Quantum Leap in AI Infrastructure
The ASUS AI POD, powered by the NVIDIA Blackwell Ultra architecture, integrates 72 NVIDIA Blackwell Ultra GPUs and 36 Grace CPUs in a large-scale rack design. This system offers unprecedented computational power, with up to 40 TB of high-speed memory per rack, and is optimized for training and inference of language models (LLMs) with trillions of parameters.
Additionally, it incorporates advanced technologies such as NVIDIA Quantum-X800 InfiniBand and Spectrum-X Ethernet, as well as a 100% liquid cooling system, ensuring energy efficiency and stable performance even under massive workloads.
Servers for Generative AI and Data Centers
ASUS has also unveiled a new range of NVIDIA-certified servers designed to accelerate AI adoption in businesses. Highlights include:
- ASUS XA NB3I-E12: Equipped with NVIDIA HGX B300 NVL16, it offers 2.3 TB of HBM3e memory and is optimized for video inference and AI reasoning.
- ASUS ESC NB8-E11: Featuring 8 NVIDIA Blackwell HGX B200 GPUs, it is designed for intensive generative AI workloads.
- ASUS ESC N8-E11V: A dual-socket server with 8 NVIDIA H200 GPUs, compatible with air or liquid cooling, ideal for demanding data centers.
The Desktop Supercomputer: ASUS Ascent GX10
One of the most surprising announcements was the launch of the ASUS Ascent GX10, a compact supercomputer that brings AI power to development environments. Equipped with the NVIDIA GB10 Grace Blackwell Superchip, it delivers 1,000 TOPS of AI performance and supports models with up to 200 billion parameters.
Edge AI Solutions for Industrial Environments
ASUS IoT has introduced two key devices for edge AI:
- PE2100N: Based on NVIDIA Jetson AGX Orin, it achieves 275 TOPS and is designed for robotics, smart cities, and autonomous vehicles.
- PE8000G: A rugged device with 2 NVIDIA RTX PCIe GPUs, capable of operating in extreme conditions (-20°C to 60°C), ideal for machine vision and real-time video analysis.
Commitment to Efficiency and Sustainability
ASUS has strengthened its commitment to sustainable infrastructure, with systems that reduce Power Usage Effectiveness (PUE) and total cost of ownership (TCO). Furthermore, its ASUS Infrastructure Deployment Center (AIDC) facilitates large-scale deployments, while the ASUS Control Center (ACC) enables remote management of systems.
Kaustubh Sanghani, Vice President of GPU Products at NVIDIA, highlighted: “The collaboration with ASUS is driving the next wave of innovation in data centers, allowing companies to explore new possibilities in AI”.
With 30 years of experience in servers, ASUS is establishing itself as a key player in the AI revolution, providing scalable and efficient solutions for a future dominated by artificial intelligence.
Note: This article is based on official information provided by ASUS and NVIDIA during the GTC 2025. All technical data and statements correspond to verifiable public announcements.
Source: Asus