ASUS Ascent GX10

The ASUS ExpertCenter GX10 is purpose-built for teams that want serious AI performance without the overhead of traditional data-center infrastructure. Powered by the NVIDIA® GB10 Grace Blackwell Superchip and aligned with NVIDIA DGX Spark, GX10 delivers up to 1 petaFLOP of AI compute with 128GB of unified memory, enabling local fine-tuning and inference of large AI models up to 200 billion parameters. This tightly integrated CPU–GPU architecture, connected through NVLink™-C2C, removes memory bottlenecks that typically slow down training and inference, resulting in faster iteration cycles and more predictable performance. For developing generative AI, computer vision pipelines, or large language models, GX10 provides a familiar, developer-friendly environment fully compatible with the NVIDIA AI software stack, CUDA libraries, and modern frameworks—making it easy to move from experimentation to deployment without re-architecting your workflows.

Designed for real-world enterprise and research environments, the ASUS ExpertCenter GX10 combines extreme compute density with operational simplicity. Its compact footprint makes it ideal for AI labs, offices, and edge locations where space, power, and cooling are limited, while its advanced thermal design ensures consistent performance under sustained workloads. Built-in NVIDIA ConnectX-7 networking and NVLink-based interconnects allow dual-system stacking, giving teams a clear path to scale performance as workloads grow. More than just hardware, GX10 is a complete AI platform—bringing together compute, memory, networking, and software into a single, efficient system that reduces reliance on cloud resources and shortens development timelines. Delivered and supported by MBUZZ, the ASUS ExpertCenter GX10 empowers organizations to build, test, and deploy AI faster, closer to their data, and with greater control.