The ASUS ExpertCenter Pro ET900N G3 is built for teams that are serious about AI development and deployment. It’s aimed at:

• AI agents and training teams who need to iterate on models and run intensive training workloads.
• AI researchers, data scientists, and ML engineers working on large, complex models and experiments.
• AI developers and startup founders building and productizing AI services.

Typical use cases include:

• Scalable AI training and deployment — handling large AI development and inferencing workloads with up to 20 petaFLOPS of AI performance and 748GB of coherent memory.
• Accelerated data analytics and autonomous systems — for scenarios that need fast model response and high-throughput data processing.
• Industry-focused AI workloads — especially in healthcare and life sciences, smart manufacturing, fintech, smart city and infrastructure, and advanced research.

In short, if your team needs a dedicated AI “supercomputer” in a standard tower form factor for training, inference, and research, the ET900N G3 is designed with those needs in mind.

The ET900N G3 is engineered to reimagine what you can do with a tower system for AI. Compared to a typical workstation, it offers:

• Extreme AI performance: Powered by the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, it delivers up to 20 petaFLOPS of AI performance, giving you headroom for large model training and high-performance inference.
• High memory capacity: With 748GB of coherent memory, you can work with large models and datasets that often don’t fit on conventional workstations.
• Scalable architecture:
  • NVIDIA NVLink-C2C enables ultra-fast CPU–GPU memory communication.
  • NVIDIA ConnectX-8 networking allows you to stack two ET900N G3 systems, increasing total memory capacity and enabling larger model deployments.
• Developer-optimized platform: It ships with Ubuntu and NVIDIA AI Developer Tools, providing a full-stack AI solution from development through deployment. This shortens setup time and helps streamline workflows for AI researchers and developers.
• Enterprise-ready tower form factor: You get advanced compute power in a standard tower chassis, making it easier to integrate into existing office or lab environments without rethinking your entire infrastructure.

The combination of AI-optimized hardware, high-bandwidth interconnects, and a pre-integrated software stack is what sets the ET900N G3 apart from traditional workstations.

The ET900N G3 is built as an AI-focused tower with hardware and I/O tailored to demanding workloads:

Core platform

• Processor: NVIDIA Grace 72-core Neoverse V2 CPU paired with the NVIDIA Blackwell Ultra Superchip.
• Memory: 748GB of coherent memory, helping support large AI models and multi-tenant workloads (with MIG technology for partitioning).
• Storage: Up to 4 x M.2 2280 NVMe PCIe 5.0 SSDs, total capacity up to 8TB, for fast dataset and model access.
• Power: 1600W TITANIUM ATX power supply designed for high, sustained compute loads.
• OS & software: Ubuntu preloaded with NVIDIA AI Developer Tools for a ready-to-use AI development environment.

Networking and I/O

• Front I/O:
  • 2 x USB 10Gbps Type-C ports
  • 2 x USB 10Gbps Type-A ports
  • 1 x USB 2.0 port
  • Headphone and microphone ports
• Rear I/O:
  • 4 x USB 10Gbps Type-A ports
  • 1 x USB Micro-B (for BMC)
  • 1 x 10G LAN (Marvel)
  • 1 x 1G LAN dedicated to BMC management (Realtek)
  • 2 x NVIDIA ConnectX-8 (CX-8) NIC ports and QSFP 400G ports for high-speed networking and system stacking
  • 1 x mini DisplayPort (for BMC)
  • 3 x audio jacks

Physical characteristics & compliance

• Dimensions (W x D x H): 232 x 584 x 565 mm (9.1 x 23 x 22.3 inches).
• Certifications: CE, FCC (Class A), BSMI, UL, KCC.

Note: Specifications, content, and product availability may change by region. Actual performance will vary based on application, usage, and environment. For the latest details, ASUS directs users to check the product page on www.asus.com.