ASUS
Ascent GX10
Ultra-small AI supercomputer powered by the NVIDIA® GB10 Grace Blackwell Superchip and NVIDIA DGX™ Spark.
Bring high-performance local AI development, model fine-tuning and inference into your workspace with a compact, scalable system.
Petaflop-scale AI computing for your desktop.
The ASUS Ascent GX10 combines the NVIDIA Blackwell GPU, a 20-core Arm CPU and 128GB of coherent unified memory in a compact AI development platform.
Blackwell GPU
Fifth-generation Tensor Cores and FP4 support for demanding AI workloads.
20-Core Arm CPU
Accelerates data preprocessing, orchestration and real-time inference workflows.
NVLink™-C2C
A cohesive CPU and GPU memory model with up to five times the bandwidth of PCIe 5.0.
Compact footprint
150 × 150 × 51mm form factor designed for desks, labs and office environments.
Built for the next generation of local AI development.
From local experimentation to larger AI models.
Integrated NVIDIA® ConnectX®-7 networking lets two ASUS Ascent GX10 systems work together for higher model capacity and distributed AI workloads.
High-speed AI data throughput
Designed for rapid data transfer and low-latency communication between nodes.
Secure, intelligent networking
Hardware acceleration for TLS, IPsec and MACsec supports encrypted data transmission with reduced CPU overhead.
Precision timing
IEEE 1588v2 PTP support helps enable microsecond-level synchronisation for time-sensitive AI and edge computing workloads.
Secure and scalable local agent workflows.
ASUS highlights support for NVIDIA NemoClaw, an open-source stack that adds policy-based privacy and security controls through NVIDIA OpenShell. Local execution can improve data control and cost predictability for suitable open-model workflows.
Third-party services, proprietary models, web plugins and API usage may still carry separate costs. Compatibility and usage should be verified for each deployment.
Local development. Scalable deployment.
A compact platform for teams that need to prototype and evaluate AI workloads locally before moving suitable models to cloud, data centre or other accelerated infrastructure.
AI Prototyping
Develop and test AI applications, assistants and workflows in a dedicated local environment.
Fine-Tuning & Inference
Fine-tune supported models and run local inference workloads with high-capacity unified memory.
Data Science
Support data processing, predictive analytics, research and simulation-oriented workflows.
Computer Vision
Develop object detection, tracking and other vision-based AI applications.
Education & Research
Enable hands-on AI experimentation, learning and model development in campus or lab environments.
Enterprise AI
Explore private on-premises AI initiatives for sectors such as manufacturing, retail, finance, healthcare and technology.
Compact design. Engineered thermal control.
The GX10 uses an advanced dual-fan design with seven-level control to support airflow management during demanding workloads.
Its small footprint is designed to bring high-density AI computing into workspace, lab and office environments.
I/O ports for AI development workflows.
ASUS Ascent GX10 specifications.
Specifications, bundled software and availability may vary by market and are subject to change. Please contact Dreamcore for current configuration and availability.
ASUS Ascent GX10 FAQ.
What workloads is the ASUS Ascent GX10 designed for?
It is designed for local AI development, prototyping, model fine-tuning, inference, data science, computer vision, simulation and research workflows. Suitability depends on the model, software configuration and workload requirements.
What AI software is available?
ASUS lists NVIDIA DGX OS and the NVIDIA AI software stack, with support for developer tools and frameworks including CUDA, PyTorch, TensorFlow, JAX, Jupyter, Ollama, TensorRT, NVIDIA NIM™ and Blueprints.
Can two GX10 systems be connected?
Yes. ASUS states that up to two GX10 systems can be linked through integrated NVIDIA ConnectX-7 networking to support larger AI models and higher-capacity workloads.
Can it run AI models locally?
Yes, compatible open-source models and AI frameworks can be configured to run locally. Model compatibility, performance and any third-party licensing requirements should be assessed for the intended deployment.
Is it suitable for enterprise use?
It is designed as a compact local AI computing platform for organisations exploring on-premises AI workloads. Dreamcore can advise on deployment requirements, networking and suitable enterprise infrastructure.
Build your local AI development environment.
Speak with Dreamcore about ASUS Ascent GX10 availability, deployment requirements and AI infrastructure solutions for your organisation.
ASUS Ascent GX10
Request a Custom Quote
Share your AI workload, model size, deployment environment, storage, networking, timeline and business requirements with Dreamcore.