· NVIDIA GB10 Grace Blackwell Superchip: Delivers 1 petaflop of AI performance
· NVIDIA AI Software Stack: A full-stack solution that provides all the tools needed to get AI projects up and running quickly
· Scalable architecture: NVIDIA NVLink-C2C for ultrafast CPU-GPU memory communication and NVIDIA ConnectX-7 networking for dual GX10 clusters
· Precision-crafted for ultimate thermal efficiency: Delivers 1.6x more efficient thermal coverage compared to similar compact systems
ASUS announced that the new Ascent GX10 will be available on October 15, 2025. This compact and powerful desktop AI supercomputer is designed to make advanced AI development more accessible to developers, AI researchers and data scientists. Powered by the groundbreaking NVIDIA® GB10 Grace Blackwell Superchip and the NVIDIA AI software stack, the Ascent GX10 delivers full-stack AI performance in a minimal footprint.
“ASUS offers a comprehensive suite of personal AI supercomputers, from workstations to NVIDIA MGX servers accelerated by NVIDIA RTX PRO™ 6000 Blackwell Server Edition GPUs, and large-scale infrastructure featuring the NVIDIA HGX B300 and NVIDIA GB300 NVL72 platforms,” said Jackie Hsu, ASUS Senior Vice President and Co-Head of Open Platform Business Group & AIoT Business Group. “As we nurture accounts, we will support customers on their AI journey to expand their footprint across the NVIDIA ecosystem.”
Empowering AI innovation with unmatched performance
ASUS Ascent GX10 is engineered for AI development. Accelerated by the NVIDIA GB10 Superchip that features a 20-core NVIDIA Grace CPU and NVIDIA Blackwell GPU, Ascent GX10 delivers up to 1 petaflop of AI performance for inference and model fine-tuning. It also supports 128GB of unified memory to enable work on up to 200-billion-parameter models directly on a desktop. At a compact 150 x 150 x 51mm, Ascent GX10 is perfect for any workspace, offering a minimal footprint with exceptional performance.
Despite its small stature, Ascent GX10 is a true full-stack AI solution accelerated by the comprehensive NVIDIA AI software stack. This integrated platform provides all the tools needed for tasks like prototyping, fine-tuning and inference for development of applications in robotics, computer vision and Vision-Language Models (VLMs).
Scalable architecture allows for unparalleled flexibility. Users can connect two Ascent GX10 units via the high-speed ConnectX-7 network interfaces to instantly double AI performance to 2 petaflops, with up to 256GB of unified memory and 8GB of storage. The Ascent GX10 enables powerful local AI training at a lower price point. It also offers enterprise-level security, keeping sensitive data local.
Ascent GX10 offers flexible storage options to meet every AI development need. The 1TB M.2 2242 NVMe™ PCIe 4.0 x4 SSD option is well-suited for AI newcomers working with smaller datasets and entry-level test projects. The upcoming 2TB M.2 2242 NVMe PCIe 4.0 x4 SSD-equipped model is ideal for supporting larger datasets, multiple models, experiments running concurrently. For those tackling large-scale, extensive workflows, the forthcoming Ascent GX10 featuring a 4TB M.2 2242 NVMe PCIe 5.0 x4 SSD is suitable for users managing numerous datasets or projects simultaneously, ensuring all data is on hand.
Ascent GX10 is meticulously-engineered and features a thermal design that’s optimized to handle demanding workloads effortlessly and ensure sustained high performance. The advanced thermal system features 7-level fan control, ultrawide fins, five heat pipes, and twin 140 x 80mm fans to draw air through discreet bottom vents to ensure powerful and efficient airflow.
Contact details:
