NVIDIA CEO Jensen Huang has announced the first PC to feature HBM memory, a personal AI supercomputer called Project DIGITS. The system targets AI developers and researchers who need high-bandwidth memory in a desktop form factor.
GB10 Grace Blackwell Superchip powers the system
Project DIGITS is built around the GB10 Grace Blackwell Superchip, which combines an ARM-based Grace CPU with a Blackwell GPU. The chip provides sufficient performance for training and running large AI models on a single system.
Specifications
- Chip: GB10 Grace Blackwell Superchip
- Memory: 128GB unified memory
- Storage: Up to 4TB NVMe
- Price: $3,000
The system includes 128GB of unified memory, a configuration that allows it to handle massive datasets without swapping. Storage is handled by up to 4TB of NVMe SSD, providing fast access to models and data.
Pricing for Project DIGITS starts at $3,000. The system is scheduled to ship later this year, but specific release dates have not been disclosed.
Project DIGITS complements NVIDIA's existing workstation and server lineup. It provides a more affordable entry point for AI development than the DGX series, while delivering more performance than a standard consumer PC.
NVIDIA has confirmed the specs and pricing for Project DIGITS. The system is positioned as a personal AI supercomputer for developers and researchers.
Discussion
0 comments
Log in to join the thread with a thoughtful take, question, or correction.