Nvidia has announced Project DIGITS, a groundbreaking desktop AI supercomputer priced at $3,000, set to launch in May. This compact powerhouse represents a major step forward in bringing data center-grade AI capabilities to personal computing.
At the heart of Project DIGITS is the new GB10 Grace Blackwell Superchip, which combines an ARM-based Grace CPU featuring 20 cores with Nvidia's latest Blackwell GPU technology. The system delivers an impressive 1 petaflop of AI computing performance while running on standard household power.
The desktop unit comes equipped with 128GB of unified memory and up to 4TB of NVMe storage. It can handle AI models with up to 200 billion parameters, and when two units are connected, they can process models reaching 405 billion parameters.
Running on Linux-based NVIDIA DGX OS, Project DIGITS supports popular development frameworks including PyTorch, Python, and Jupyter notebooks. Users have access to Nvidia's extensive AI software library, development tools, and pre-trained models through the NGC catalog.
"AI will be mainstream in every application for every industry," said Jensen Huang, Nvidia's CEO. "With Project DIGITS, the Grace Blackwell Superchip comes to millions of developers."
The system allows developers to prototype and test AI models locally before deploying them to cloud or data center infrastructure. This seamless workflow maintains compatibility across different computing environments through Nvidia's AI Enterprise software platform.
MediaTek, known for its expertise in ARM-based chip design, collaborated with Nvidia to optimize the GB10's power efficiency and performance. The resulting hardware delivers unprecedented AI computing power in a form factor similar to a Mac Mini.
This release marks a notable shift in making advanced AI computing accessible to individual developers, researchers, and students, potentially transforming how AI development happens at the desktop level.