ASUS Launches ExpertCenter Pro ET900N G3 Built on NVIDIA DGX Station Architecture

2026-06-23 · Source: The AI Journal · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Cloud Computing & IT Infrastructure · Depth: Expert, short

Summary

ASUS announced the global availability of its ExpertCenter Pro ET900N G3 on June 23, 2026, a deskside AI supercomputer designed to bring data-center-class AI performance to enterprises, developers, and researchers. This system is powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip and built on NVIDIA DGX Station GB300 architecture. It offers multi-petaflop-scale AI computing, delivering up to 20 PFLOPS of AI performance. The ET900N G3 features 748GB of coherent CPU-GPU memory, enabling the training of AI models up to 1 trillion parameters and accelerating local AI workflows. It supports demanding AI workloads like LLM fine-tuning, generative AI, autonomous AI agents, and deep learning. Stress testing with vLLM demonstrated approximately 864 tokens per second output throughput for the Qwen open-source AI model, with combined input/output processing reaching around 1,600 tokens per second. This system aims to provide local AI infrastructure with greater control, lower latency, predictable costs, and enhanced data privacy, with future support for Windows-based AI development.

Key takeaway

For AI Engineers and researchers developing large-scale AI models or autonomous agents, the ASUS ExpertCenter Pro ET900N G3 offers a compelling on-premises solution. You can achieve data-center-class performance with 748GB coherent memory, enabling local LLM fine-tuning and reducing cloud dependency. Consider this system to enhance data privacy, lower latency, and gain predictable operational costs for your advanced AI workflows. Explore its capabilities for secure, local AI execution and agentic environments.

Key insights

The ASUS ET900N G3 delivers data-center AI performance and massive memory to the deskside, enabling local, secure, and efficient advanced AI development.

Principles

• Local AI infrastructure enhances control and privacy.
• Coherent memory boosts large model training.
• Deskside supercomputers reduce cloud dependency.

Method

The system supports NVIDIA AI software stack and NemoClaw workflows for building and deploying always-on AI assistants and autonomous agents in secure local environments.

In practice

• Fine-tune LLMs locally with 748GB coherent memory.
• Deploy autonomous AI agents on-premises.
• Accelerate generative AI and deep learning.

Topics

• ASUS ExpertCenter Pro ET900N G3
• NVIDIA GB300 Superchip
• Deskside AI Supercomputer
• Large Language Models
• Autonomous AI Agents
• On-premises AI

Best for: CTO, VP of Engineering/Data, AI Architect, AI Engineer, AI Scientist, Director of AI/ML

Related on AIssential

• The $4,000 Backpack Supercomputer That Changed My Mind About Local AI — Unified memory architectures like NVIDIA's Grace Blackwell overcome VRAM limitations for local large language model deployment.
• With Nvidia Groq 3, the Era of AI Inference Is (Probably) Here — Specialized hardware for AI inference is critical due to its distinct low-latency computational requirements.
• Now Generally Available, NVIDIA RTX PRO 5000 72GB Blackwell GPU Expands Memory Options for Desktop Agentic AI — The NVIDIA RTX PRO 5000 72GB Blackwell GPU expands desktop AI capabilities with increased memory for larger models.
• Saturate your Tensor Cores: Intel at NVIDIA GTC 2026 — Balanced CPU+GPU system design is crucial for maximizing GPU utilization and scaling AI workloads effectively.
• Nvidia's DGX Station is a desktop supercomputer that runs trillion-parameter AI models without the cloud — Nvidia's DGX Station brings data center-scale AI compute to the desktop, enabling local execution of trillion-parameter models.
• Nvidia showed up at my house — Local AI is advancing rapidly, enabling powerful models and autonomous agents on personal systems.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The AI Journal.