Powered by GB10 Grace Blackwell Superchip and developed in collaboration with MediaTek.
- High Performance for AI Models
- Prototyping, fine-tuning, and deploying large AI models up to 200 billion parameters.
- Connecting two supercomputers enables running models with up to 405 billion parameters, such as Meta's Llama 3.1.

- Robust Hardware Specifications
- Memory: 128 GB RAM.
- Storage: Up to 4 TB NVMe.
- Processor: 20-core Arm central processor.
- GPU: Blackwell GPU.
- Capable of performing up to one quadrillion AI calculations per second.
- Flexible Software Support
- Compatible with popular frameworks like PyTorch, Python, and Jupyter.
- Runs on a Linux-based operating system.
- Can function as a standalone workstation or connect to PCs running Windows or macOS.
- Project Digits Availability and Pricing
- Scheduled for release in May 2025.
- Starting price: $3,000.
- Comprehensive Nvidia Ecosystem Access
- Includes access to Nvidia’s SDKs, orchestration tools, and pre-trained models.
Nvidia Unveils Compact AI Supercomputer for Developers — It Fits on Your Hand
* The company offers the new Nvidia Jetson Orin Nano Super developer kit with generative AI for creating chatbots, AI agents, or AI-powered robots. * Its small size makes it suitable for “hobbyists and students.” * Compared to the previous Jetson Orin Nano model, CPU performance has increased by 70%, reaching 67 TOPS
