ASUS Ascent GX10 Compact AI Supercomputer - 4TB

ASUS Ascent GX10 Compact Desktop AI Supercomputer - 4TB (GX10-GG0007BN) / (90MS0371-M000Z0) - 1 Year Onsite Warranty + Limited-Time Offer: Free Upgrade to 3 Years Warranty.

ASUS Ascent GX10 – Compact AI Supercomputer for Advanced Workflows

The ASUS Ascent GX10 Compact Desktop AI Supercomputer delivers extreme performance for creators, researchers, and engineers who need workstation-class capability in a space-efficient design. Built to execute high-speed data analysis, AI model training, 3D rendering, and complex simulation workloads, it brings supercomputer-grade AI performance directly to the desktop without requiring large-scale infrastructure.

AI-Optimized Performance for Demanding Computing

Powered by ASUS’s next-generation AI-accelerated architecture, the GX10 is designed for machine learning, deep learning, and generative AI tasks. It handles large language models, simulation engines, and production-grade 3D pipelines with smooth, stable output. Whether you’re running inference, training custom AI models, or pushing heavy visualization workloads, the system ensures consistently optimized performance.

4TB High-Speed Storage for Heavy AI Workloads

The Ascent GX10 features 4TB of ultra-fast storage, giving you ample room for massive datasets, training checkpoints, research files, and complex production assets. Its high read/write throughput reduces latency during large data transfers and accelerates workflow transitions. Ideal for AI labs, design studios, and engineering teams, it ensures seamless handling of data-intensive tasks.

Compact Design with Advanced Thermal Architecture

Despite its small footprint, the GX10 integrates a high-efficiency thermal design that keeps performance stable during prolonged high-load computing. ASUS engineered the system for low noise and optimal heat dissipation, making it suitable for offices, research environments, and studio setups that require power without sacrificing comfort or reliability.

Technical Specification: 

Processor (CPU + GPU Architecture)

• Powered by the NVIDIA GB10 Grace-Blackwell Superchip, combining a Grace CPU and Blackwell-class GPU into a single SoC

• Optimized for AI inference, model training, and high-performance computing workloads

• Delivers up to 1 petaFLOP FP4 AI performance, engineered for advanced generative AI pipelines

• Unified CPU-GPU memory architecture enables extremely high data throughput for LLMs and complex neural models

Graphics / AI Acceleration

• Integrated NVIDIA Blackwell GPU platform with next-generation tensor cores

• Designed for large model training, multimodal workloads, and real-time AI inference

• Supports NVIDIA AI Enterprise, CUDA, TensorRT, and full-stack GPU acceleration frameworks

Memory

• 128GB LPDDR5X Unified System Memory

• High bandwidth, low latency memory pool shared across CPU and GPU

• Optimized for handling large datasets, embeddings, and transformer-based models

Storage

• 4TB NVMe PCIe SSD

• Ultra-fast sequential read/write speeds for rapid dataset loading, checkpoint saving, and model deployment

• Suitable for large model storage (LLMs, vision models, vector databases, multimodal assets)

Connectivity & I/O

• Network:

  • 10GbE LAN for high-speed data transfer

  • NVIDIA ConnectX-7 SmartNIC for advanced networking, clustering, and multi-node AI scaling

• Ports:

  • 3 × USB-C (USB 3.2 Gen 2×2, 20Gbps, DisplayPort Alt-Mode)

  • 1 × USB-C Power-In

  • 1 × HDMI 2.1 for high-resolution display output

• Wireless:

  • Wi-Fi 6E + Bluetooth (where applicable, depending on configuration)

Cooling System

• Advanced quiet-cooling thermal design optimized for compact AI compute loads

• Intelligent airflow system ensuring stable performance under continuous high utilization

• Low-noise operation suitable for office, lab, or desktop environments

Operating System Compatibility

• Supports NVIDIA DGX OS (Ubuntu-based AI-optimized stack)

• Compatible with major Linux distributions for AI frameworks

• Windows 11 Pro supported for non-AI general workloads (not recommended for heavy ML training)

Form Factor

• Ultra-compact chassis: 150 × 150 × 51 mm

• Desktop-friendly footprint suitable for small workspaces, research labs, and edge deployments

• Lightweight, portable design at approximately 1.48 kg

Warranty

• Includes 1-Year Onsite Warranty

• Optional extended care and service packages available