Durga Puja: no dispatch from Kolkata on 18, 19, 20 and 21 Oct.

nationalPC

ASUS ExpertCenter Pro ET900N G3 AI Supercomputer

1 / 5

ASUS ExpertCenter Pro ET900N G3 AI Supercomputer is a powerful deskside AI system powered by the NVIDIA GB300 Grace Blackwell Superchip, delivering 748GB coherent memory and up to 20 petaFLOPS of AI performance for advanced model training, inference and AI-agent development.

Price on request

Prices change daily - ask for today's price.

Key features

  • NVIDIA GB300 Grace Blackwell Superchip
  • 72-core NVIDIA Grace CPU with Arm Neoverse V2 architecture
  • NVIDIA Blackwell Ultra GPU
  • 748GB coherent memory
  • 496GB LPDDR5X CPU memory | 252GB HBM3e GPU memory
  • Up to 20 petaFLOPS FP4 AI performance
  • Ubuntu operating system with NVIDIA AI Developer Tools
  • 2 x M.2 PCIe 5.0 x4 slots with pre-configured OS RAID 1
  • 2 x M.2 PCIe 6.0 x4 slots for training-data storage
  • 2 x NVIDIA ConnectX-8 SuperNIC QSFP112 ports
  • 400G QSFP networking support | 10G LAN | Dedicated 1G BMC management LAN
  • ASPEED AST2600 BMC with IPMI 2.0 and DMTF Redfish API
  • 1600W Titanium ATX power supply
  • Professional tower design for local AI training, inference and agentic workloads
MODEL: EXPERTCENTER PRO ET900N G3 | DESKSIDE AI SUPERCOMPUTER

ASUS ExpertCenter Pro ET900N G3 AI Supercomputer

ASUS ExpertCenter Pro ET900N G3 brings data-center-class AI capability to the deskside. Built on NVIDIA DGX Station architecture, it combines the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip, 748GB coherent memory, high-speed networking and a full AI software environment for advanced local AI development and deployment.

ASUS ExpertCenter Pro ET900N G3 AI Supercomputer
20 PFLOPSUp to AI performance
748GBLarge coherent memory
72-CoreNVIDIA Grace Neoverse V2 CPU
400GQSFP high-speed networking

Enterprise-Class AI Compute at the Deskside

ExpertCenter Pro ET900N G3 is designed for teams that need to develop, run and manage complex AI workloads close to their data. The NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip links the Grace CPU and Blackwell Ultra GPU through NVLink-C2C, enabling a large coherent memory pool for large models, AI agents, real-time inference and memory-intensive research workloads.

Large local models: 748GB coherent memory supports large language models and demanding AI datasets locally.
Secure AI workflows: NVIDIA software environments help enterprises deploy AI while retaining control over sensitive data.
Scalable platform: ConnectX-8 networking and NVLink-C2C enable high-bandwidth scale-up and AI infrastructure workflows.

Designed for Sustained AI Deployment

A data-center-grade thermal architecture with closed-loop AIO liquid cooling is designed to keep the ET900N G3 operating reliably under continuous, compute-heavy AI workloads. Its standard tower form factor brings advanced on-premise compute to teams without requiring a traditional rack-scale deployment.

NVIDIA AI software stack: Ubuntu with NVIDIA AI Developer Tools provides a preconfigured foundation for AI development and inference.
AI management: ASUS Zenni AI Hub supports model deployment, APIs, knowledge workflows and multi-agent management.
Flexible compute: NVIDIA Multi-Instance GPU can partition GPU resources into up to seven isolated instances.
ASUS ExpertCenter Pro ET900N G3 AI supercomputer tower

ASUS ExpertCenter Pro ET900N G3 Specifications

ModelASUS ExpertCenter Pro ET900N G3
ArchitectureNVIDIA DGX Station architecture
Desktop SuperchipNVIDIA GB300 Grace Blackwell Ultra
CPU / GraphicsNVIDIA Grace 72-core Neoverse V2 CPU with NVIDIA Blackwell Ultra GPU
AI PerformanceUp to 20 PFLOPS
Memory748GB coherent memory
StorageUp to 4 x M.2 2280 NVMe PCIe 5.0 SSDs; up to 8TB total storage
High-Speed NetworkingNVIDIA ConnectX-8 SuperNIC; dual QSFP 400G ports
Additional Networking10G LAN and dedicated 1G LAN for BMC
Operating EnvironmentUbuntu with NVIDIA AI Developer Tools and NVIDIA AI software stack
CoolingClosed-loop AIO liquid cooling with data-center-grade thermal design
Dimensions584 x 232 x 565 mm

Best Use Cases

Large Language Models

Develop, tune and run large AI models close to private data and internal teams.

AI Agents and Knowledge

Build private agentic AI, enterprise document intelligence and API-connected AI applications.

Research and Simulation

Accelerate computer vision, scientific computing, engineering simulation and generative AI.

Frequently Asked Questions

What is the ExpertCenter Pro ET900N G3?

It is a deskside AI supercomputer based on NVIDIA DGX Station architecture, designed for AI development, inference and enterprise deployment.

What type of workloads is it built for?

It is suited to large language models, AI agents, computer vision, generative AI, scientific computing and other memory-intensive workloads.

Does it offer high-speed networking?

Yes. The system includes NVIDIA ConnectX-8 networking with dual QSFP 400G ports, plus 10G LAN and dedicated BMC networking.

Bring Data-Center-Class AI to Your Team

Choose ASUS ExpertCenter Pro ET900N G3 for high-capacity local AI development, enterprise inference and scalable deskside compute.

Buy ASUS ExpertCenter Pro ET900N G3 from NationalPC
Price on request