NVIDIA Project DIGITS: The Smallest AI Supercomputer — Grace Blackwell-powered, $3,000, petaflop-class AI performance. Available May 2025.
At CES 2025, NVIDIA introduced Project DIGITS, a groundbreaking personal AI supercomputer powered by the new NVIDIA GB10 Grace Blackwell Superchip. Designed to simplify the AI development workflow, Project DIGITS delivers the performance of a petaflop-class AI system in a compact, desktop-friendly form factor. This innovation is set to revolutionize how AI researchers, data scientists, and students prototype, fine-tune, and deploy large AI models.
Project DIGITS is built on NVIDIA’s Grace Blackwell architecture, combining the latest GPU and CPU technology advancements. The system can run 200 billion-parameter models locally, making it the world’s smallest AI supercomputer. For developers, this means the ability to experiment with large-scale AI models directly on their desktops, with seamless scalability to cloud or data center environments.
The Power of the GB10 Superchip
At the heart of Project DIGITS is the NVIDIA GB10 Superchip, a system-on-a-chip (SoC) that delivers up to 1 petaflop of AI performance at FP4 precision. The GB10 combines an NVIDIA Blackwell GPU with next-generation CUDA® cores, Tensor Cores, and a high-performance NVIDIA Grace CPU featuring 20 power-efficient Arm-based cores. The two components are connected via NVLink-C2C, ensuring high-speed communication and efficiency.
Each system includes 128GB of unified memory and up to 4TB of NVMe storage, enabling developers to tackle many demanding AI workloads. For even larger models, two Project DIGITS systems can be linked via NVIDIA ConnectX networking, allowing users to run models with up to 405 billion parameters.
Seamless AI Development and Deployment Project
Project DIGITS is more than just a hardware solution—it’s a complete AI development platform. Running on a Linux-based NVIDIA DGX OS, the system integrates seamlessly with NVIDIA’s broader ecosystem, including NVIDIA DGX Cloud™ and data center infrastructure. This allows developers to prototype AI locally and scale their work effortlessly to enterprise-grade environments.
The platform also provides access to NVIDIA’s extensive software library, including experimentation, prototyping, and deployment tools. Developers can leverage the NVIDIA NGC catalog, which offers pre-trained models, frameworks, and software development kits. Popular tools like PyTorch, Python, and Jupyter notebooks are fully supported, as are NVIDIA’s specialized frameworks, such as NeMo™ for fine-tuning large language models and RAPIDS™ for data science acceleration.
For those building advanced AI applications, NVIDIA offers NVIDIA Blueprints and NIM™ microservices, enabling the creation of agentic AI systems. When ready for production, developers can transition to enterprise-grade environments with the NVIDIA AI Enterprise license, which provides security, support, and regular updates.
Empowering the Next Generation of AI Innovation
With Project DIGITS, NVIDIA makes AI supercomputing accessible to a global audience. By placing the power of the Grace Blackwell platform on every desk, the company is empowering developers, researchers, and students to push the boundaries of AI innovation.
Whether it’s prototyping the next breakthrough in generative AI, fine-tuning large language models, or developing agentic AI applications, Project DIGITS provides the tools and performance needed to shape the future of AI.
Coming in at $3,000 and available this May through top board partners, Project DIGITS is set to create a paradigm shift in developer workflows.
Engage with StorageReview
Newsletter | YouTube | Podcast iTunes/Spotify | Instagram | Twitter | TikTok | RSS Feed
