Skip to main content

NVIDIA Announces DGX Spark and DGX Station Personal AI Computers

NVIDIA has introduced the DGX Spark and DGX Station, personal Artificial Intelligence (AI) supercomputers powered by the Grace Blackwell architecture. These systems are designed to empower developers, researchers, and data scientists to prototype and fine-tune large AI models directly from their desktops.

The DGX Spark, which follows Project DIGITS, and the DGX Station offer users the ability to run AI models locally or deploy them in cloud environments. This marks a shift as the capabilities of the Grace Blackwell architecture expand from data center to personal computing environments.

Jensen Huang, founder and CEO of NVIDIA, stated, “AI has transformed every layer of the computing stack. It stands to reason a new class of computers would emerge — designed for AI-native developers and to run AI-native applications.”

The DGX Spark is highlighted as the world’s smallest AI supercomputer, incorporating the NVIDIA GB10 Grace Blackwell Superchip. This superchip is equipped with a powerful Graphics Processing Unit (GPU) and optimized for performance with workloads relevant to AI development.

A key feature of the GB10 is its NVLink-C2C interconnect technology, which enhances data access and throughput between the Central Processing Unit (CPU) and GPU significantly. This design aims to support memory-intensive AI tasks efficiently.

On the other hand, the DGX Station provides data-center-level performance with the GB300 Grace Blackwell Ultra Superchip. This platform is equipped with extensive memory and advanced communication features, making it suitable for demanding AI development tasks.

Both the DGX Spark and DGX Station not only enable powerful AI computing but also streamline the transition of projects between desktop operations and cloud systems, minimizing the need for code alterations.

Reservations for the DGX Spark are open, while the DGX Station is expected to be available from various manufacturing partners within the year.

For further insights, users can access the NVIDIA GTC keynote sessions taking place through March 21.