Asus has introduced the Ascent GX10, a compact AI supercomputer designed to bring high-performance computing capabilities to developers, researchers, and data scientists. This device, powered by Nvidia’s Grace Blackwell GB10 Superchip, aims to facilitate local AI workload processing, enabling users to prototype, fine-tune, and run large AI models without exclusive reliance on cloud or data center resources.
The Ascent GX10 features 128GB of unified memory and a Blackwell GPU equipped with fifth-generation Tensor Cores and FP4 precision support, delivering up to 1000 TOPS of AI processing power. It also incorporates a 20-core Grace Arm CPU, which enhances data processing and orchestration for AI inferencing and model tuning. Asus asserts that the GX10 can handle AI models with up to 200 billion parameters.
KuoWei Chao, General Manager of Asus IoT and NUC Business Group, emphasized the device’s role in democratizing AI development, stating, “The Asus Ascent GX10 is designed to bring this transformative power to every developer’s fingertips.”
The GX10 utilizes NVLink-C2C, providing increased bandwidth for efficient CPU and GPU memory sharing, which improves performance across AI workloads. The system also includes an integrated ConnectX network interface, allowing for the linking of two GX10 units to manage larger AI models.
Asus plans to make the Ascent GX10 available for pre-order in Q2 2025. Nvidia has indicated a price point of $2999, which includes 1TB of storage. In comparison, Nvidia’s DGX Spark is priced at $3999 and offers 4TB of storage.
The Ascent GX10’s compact design and performance capabilities position it as a potential tool for AI development, offering an alternative to cloud-based solutions.
