NVIDIA has introduced DGX Spark and DGX Station, personal AI supercomputers powered by the NVIDIA Grace Blackwell platform. DGX Spark, formerly Project DIGITS, and DGX Station enable AI developers, researchers, data scientists, and students to prototype, fine-tune, and inference large models on desktops. These systems allow users to run models locally or deploy them on NVIDIA DGX Cloud or other cloud infrastructures.
DGX Spark and DGX Station bring the Grace Blackwell architecture, previously exclusive to data centers, to desktops. Global system builders, including ASUS, Dell, HP Inc., and Lenovo, are developing these systems. Jensen Huang, NVIDIA’s founder and CEO, emphasized the emergence of a new class of computers designed for AI-native developers and applications.
DGX Spark is the world’s smallest AI supercomputer, featuring the NVIDIA GB10 Grace Blackwell Superchip optimized for desktop use. The GB10 includes a powerful NVIDIA Blackwell GPU with fifth-generation Tensor Cores and FP4 support, delivering up to 1,000 trillion operations per second for AI compute. It uses NVIDIA NVLink-C2C interconnect technology for enhanced performance in memory-intensive workloads.
NVIDIA’s full-stack AI platform allows DGX Spark users to seamlessly transition their models from desktops to DGX Cloud or other infrastructures. DGX Station, built with the GB300 Grace Blackwell Ultra Desktop Superchip, features 784GB of coherent memory space for large-scale training and inferencing. It includes the NVIDIA ConnectX-8 SuperNIC for high-speed networking and data transfers.
Reservations for DGX Spark systems are open at nvidia.com, with DGX Station expected to be available later this year from manufacturing partners. Users gain access to NVIDIA NIM microservices with the NVIDIA AI Enterprise software platform.
— news from NVIDIA Blog
