We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in

ASUS

ASUS GX10-GG0016BN AI edge system 128GB LPDDR5x 4TB

SKU CH-2X6C-1T6PF MPN GX10-GG0016BN

Compact AI edge system with NVIDIA GB10 Superchip, 128GB LPDDR5x unified memory and 4TB NVMe storage for local agentic workloads

  • CPU TypeNvidia GB10
  • Installed Memory Size128GB
  • Memory TypeLPDDR5X
  • SSD (Installed)4TB
  • GPU TypeNVIDIA Blackwell GPU
  • Operating SystemNVIDIA DGX OS

Option Ascent GX10 AI 4TB

  • 1 petaFLOP AI performance from NVIDIA GB10 Superchip
  • 128GB LPDDR5x unified memory for 200B model fine-tuning
  • 4TB NVMe storage with DGX OS for agentic AI workflows
  • Wi-Fi 7 connectivity supports scalable dual-system stacking
  • Dual-fan cooling delivers 1.6× better thermal efficiency

Compact AI Supercomputer

The ASUS Ascent GX10 is a desktop AI system built around the NVIDIA GB10 Grace Blackwell Superchip, delivering up to 1 petaFLOP of AI performance in a compact chassis. It targets developers and researchers who need local, secure infrastructure for building, fine-tuning, and deploying large language models up to 200 billion parameters. The system ships with NVIDIA DGX OS and supports agentic AI frameworks including OpenClaw and NemoClaw for autonomous workflow development.

Ideal for Local Agentic Development

This platform suits AI engineers prototyping agentic applications, data scientists fine-tuning models on-premise, and teams requiring governed, sandboxed execution with private data access. It also supports dual-system stacking via NVLink-C2C and ConnectX-7 networking for scaled workloads reaching 405 billion parameters. Buyers who need more than 128GB of unified memory or require rackmount server form factors should evaluate alternative enterprise AI appliances.

128GB LPDDR5x Unified Memory

The 128GB LPDDR5x coherent unified memory pool is the decisive specification, enabling CPU-GPU data sharing without copies and directly determining the maximum model size that fits in memory for fine-tuning and inference. This capacity supports 200B parameter models on a single unit and scales to 256GB across two stacked systems. The 4TB PCIe Gen5 NVMe SSD provides high-speed local storage for datasets and checkpoints.

Highlights

  • 1 petaFLOP AI performance from NVIDIA GB10 Superchip
  • 128GB LPDDR5x unified memory for 200B model fine-tuning
  • 4TB NVMe storage with DGX OS for agentic AI workflows
  • Wi-Fi 7 connectivity supports scalable dual-system stacking
  • Dual-fan cooling delivers 1.6× better thermal efficiency

Specifications

BrandASUS
SeriesASUS Ascent GX10
ModelASUS Ascent GX10
Part NumberGX10-GG0016BN
CPU TypeNvidia GB10
Installed Memory Size128GB
Memory TypeLPDDR5X
SSD (Installed)4TB
GPU TypeNVIDIA Blackwell GPU
Operating SystemNVIDIA DGX OS
FeaturesAI Agent Ready: NemoClaw and OpenClaw
Shipping weight3.18 kg
Package size191 × 191 × 152 mm

Other versions of this model

2 more in the same range

Same family, different memory, cooler or clock — compare before you commit.

Questions about this item

What thermal limit appears first when the GX10 runs sustained AI workloads?

The dual-fan cooling is engineered for 1.6× better thermal efficiency, keeping the GB10 Superchip at sustained clocks during long fine-tuning or inference sessions.

How loud and hot does the ASUS Ascent GX10 get during all-day agentic AI runs?

Precision-crafted dual-fan cooling maintains low acoustics and steady temperatures even when the system delivers 1 petaFLOP of FP4 performance continuously.

Which software stack must already be in place for the GX10 to run OpenClaw and NemoClaw agents out of the box?

NVIDIA DGX OS ships pre-installed, providing the full AI software stack required for OpenClaw, NemoClaw and Hermes Agent frameworks.

Also in this aisle

People compared these

Same shelf, same checkout — eight coins and a 30-minute rate lock.