NVIDIA has unveiled its cutting-edge Blackwell AI systems, introducing the DGX Station and DGX Spark, each powered by the innovative GB300 and GB10 technologies. These moves mark a significant expansion in NVIDIA’s collaboration with leading manufacturers like Acer, GIGABYTE, and MSI, aiming to equip developers, data scientists, and researchers worldwide with unparalleled AI performance and efficiency.
Today’s enterprises, software providers, government agencies, startups, and research institutions demand high-performance systems that merge the capabilities of an AI server with the convenience of a desktop. These systems deliver robust performance without sacrificing data size, privacy of proprietary models, or scalability speed.
The DGX Spark is designed to fuel innovation with its NVIDIA GB10 Grace Blackwell Superchip and advanced Tensor Cores, offering up to 1 petaflop of AI compute and 128GB of unified memory. It seamlessly integrates with NVIDIA DGX Cloud or any accelerated infrastructure for exporting models.
For those tackling the most demanding AI workloads, the DGX Station stands out with the GB300 Grace Blackwell Ultra Desktop Superchip. It boasts up to 20 petaflops of AI power and 784GB of unified system memory. Enhanced with the NVIDIA ConnectX-8 SuperNIC, it supports network speeds up to 800Gb/s, providing high-speed connectivity and enabling multi-station scalability.
Whether serving as a standalone desktop for complex AI models or as a centralized compute hub for multiple users, the DGX Station supports NVIDIA’s Multi-Instance GPU technology. This flexibility allows the system to partition into up to seven instances, each with its own high-performance resources, essentially functioning as a personal AI cloud for development teams.
Availability for DGX Spark is planned through major brands like Acer, ASUS, Dell Technologies, GIGABYTE, HP, Lenovo, and MSI, along with global partners, starting in July. Reservations are now open. DGX Station is set to be released later this year through established vendors such as ASUS, Dell Technologies, GIGABYTE, HP, and MSI.
