ASUS Unveils ExpertCenter Pro ET900N G3: A Desktop Powerhouse with NVIDIA's GB300 Blackwell Ultra Superchip

ASUS Introduces Groundbreaking AI Workstation

ASUS has quietly launched the ExpertCenter Pro ET900N G3, a desktop system that marks a significant milestone in the democratization of high-end AI hardware. This workstation is one of the first to feature NVIDIA's GB300 Grace Blackwell Ultra "desktop superchip," signaling a shift in NVIDIA's strategy of reserving its most powerful AI components for its own DGX Station lineup 1

1

.

Unprecedented Performance in a Desktop Form Factor

The ExpertCenter Pro ET900N G3 combines NVIDIA's Grace ARM-based CPU with the new Blackwell Ultra GPU, mirroring the configuration found in the DGX Station announced at GTC 2025. This powerhouse offers:

• Up to 784GB of unified LPDDR5X and HBM3E memory
• 20 PFLOPS of AI performance
• ConnectX-8 SuperNIC networking (800 Gb/s) for high-speed scaling
• Support for NVIDIA DGX OS 2

  2

These specifications make the system ideal for AI training, inference, and large-scale model development, bringing server-grade performance to a desktop form factor.

Expandability and Future-Proofing

Despite its unassuming appearance, the ExpertCenter Pro ET900N G3 offers serious expandability options:

• Three PCIe x16 slots for additional GPUs or accelerators
• Three M.2 slots for SSD storage
• Power delivery via standard ATX + EPS12V connectors
• Three 12V-2×6 connectors capable of delivering up to 1,800W to the GPU 1

  1

This design allows users to potentially stack additional RTX Blackwell cards, further increasing compute power and future-proofing their investment.

NVIDIA's Broader AI Hardware Strategy

The introduction of the ExpertCenter Pro ET900N G3 aligns with NVIDIA's evolving strategy in the AI hardware market. By partnering with OEMs like ASUS, Dell, and Lambda, NVIDIA is expanding beyond its exclusive DGX lineup, making high-performance AI systems available to a broader range of professionals 3

3

.

This move also showcases NVIDIA's growing ambitions beyond GPUs. The inclusion of the ARM-based Grace processor, designed specifically for AI and HPC workloads, signals NVIDIA's entry into territory traditionally dominated by AMD and Intel. While not a direct competitor to general-purpose CPUs, Grace complements NVIDIA's GPU dominance by enabling unified, high-bandwidth memory and compute performance that traditional setups struggle to match 1

1

.

Industry Impact and Future Outlook

The ExpertCenter Pro ET900N G3 arrives as NVIDIA's GB300 Blackwell Ultra is scaling across the AI industry. Recent deployments include Dell's GB300 NVL72 clusters at CoreWeave, featuring 72 Blackwell Ultra GPUs and 36 Grace CPUs per rack, with 1.1 exaFLOPS of FP4 inference performance 1

1

.

While pricing details for the ExpertCenter Pro ET900N G3 have not been revealed, industry experts expect it to be in the five-digit range, given that SXM-based compute GPUs alone cost tens of thousands of dollars 2

2

. Despite the high cost, this launch represents a significant step towards making enterprise-grade AI hardware more accessible to a wider range of businesses and research institutions.