• NVIDIA GB10 Grace Blackwell Superchip — Combined ARM CPU + Blackwell GPU on a single superchip delivering up to 1,000 TOPS of AI compute for local inference and training
• ARM Cortex-X925 CPU Architecture — High-efficiency ARM CPU core architecture optimized for AI workload orchestration alongside the Blackwell GPU
• 128GB LPDDR5X Unified Memory — Ultra-high-bandwidth unified memory pool shared between CPU and GPU for large language models and complex AI pipelines
• 1TB NVMe SSD — Fast local storage for AI model repositories, datasets, and development environments
• NVIDIA DGX OS Pre-Installed — NVIDIA’s AI-optimized operating system with containerized workload support, CUDA, and full MLOps stack ready out of the box
• Compact Desktop Form Factor — Full AI supercomputer capability in a workstation-sized chassis for lab, office, and edge deployment
• Enterprise AI Acceleration — Designed for developers and researchers building production-grade AI applications locally

The ASUS Ascent GX10 is a compact AI supercomputer powered by the NVIDIA GB10 Grace Blackwell Superchip — the same GPU architecture used in NVIDIA’s flagship data center AI infrastructure, now in a desktop form factor accessible to individual developers and research teams. The GB10 combines an ARM Cortex-X925 CPU with a Blackwell GPU in a unified architecture, delivering up to 1,000 TOPS of AI compute with 128GB of shared LPDDR5X memory for seamless large model inference, fine-tuning, and multi-modal AI workloads without cloud dependency.

Pre-installed with NVIDIA DGX OS, the system arrives ready for immediate AI development — supporting PyTorch, TensorFlow, TensorRT, and CUDA out of the box with full container orchestration via Docker and Kubernetes. The 1TB NVMe SSD provides fast local storage for model checkpoints, training datasets, and development pipelines. For organizations building AI-powered products, autonomous systems, or running local LLM inference at scale, the Ascent GX10 delivers data center-class AI performance at the desk.

Ideal For

• Local LLM Inference & Fine-Tuning — Run and fine-tune large language models locally without cloud API dependency
• AI Research & Development — Full CUDA/TensorRT ecosystem for model development, testing, and iteration
• Edge AI Deployment — Deploy GPU-accelerated AI inference at the edge in a compact, power-efficient chassis
• Computer Vision & Robotics — Real-time inference for video analytics, object detection, and autonomous systems
• Enterprise AI Prototyping — Rapidly prototype and deploy AI solutions with DGX OS and NVIDIA software stack

Specifications

• Model — ASUS Ascent GX10-GG0010BN
• AI Platform — NVIDIA GB10 Grace Blackwell Superchip
• CPU — ARM Cortex-X925
• GPU — NVIDIA Blackwell
• Memory — 128GB LPDDR5X (unified CPU+GPU)
• Storage — 1TB NVMe SSD
• OS — NVIDIA DGX OS
• AI Performance — Up to 1,000 TOPS
• SKU — CDGX10GG0010BN
• MPN — GX10-GG0010BN