At CES 2025, NVIDIA unveiled Project DIGITS, a groundbreaking new system designed to bring the power of an AI supercomputer directly to developers' desks. Powered by NVIDIA’s GB10 Grace Blackwell Superchip, Project DIGITS represents a significant leap forward in making high-performance AI accessible outside the cloud.

What Is Project DIGITS?

NVIDIA’s Project DIGITS
NVIDIA’s $3,000 Project DIGITS. Source: NVIDIA

Project DIGITS is a Linux-based desktop system purpose-built for AI workloads. It features the GB10 Grace Blackwell Superchip, which integrates:

  • Grace CPU (Arm-powered) with 10 Cortex-X925 and 10 Cortex-A725 cores for cutting-edge performance.
  • Blackwell GPU, equipped with NVIDIA’s latest CUDA and RT cores, delivering up to 1 petaflop (1,000 TFLOPs) of AI computing power at FP4 precision.

With 128 GB of unified memory, the system is designed to handle the largest AI models — up to 200 billion parameters — on the edge. This allows developers to work on complex models without needing to offload tasks to a data center or cloud platform.

Whom Is It For?

Project DIGITS is targeted at:

  • AI Developers and Researchers who need the power to train, fine-tune, and deploy state-of-the-art AI models directly from their workspace.
  • Enterprises and universities working on advanced AI solutions who want in-house capabilities without relying on external resources.
  • Tech innovators looking to prototype and test AI applications at the cutting edge of innovation.

NVIDIA’s goal is to make high-performance AI development more accessible to individual users and teams, leveling the playing field for innovation.

Why Is It Important?

AI development traditionally relies on cloud-based infrastructure or expensive data centers. Project DIGITS changes that by bringing this power directly to desktops. The key benefits include:

  • Massive Compute Power: With up to 1 petaflop of AI compute at FP4 precision, DIGITS handles workloads that were previously out of reach for desktop systems.
  • Unified Memory: Its 128 GB of shared memory allows seamless handling of large datasets and AI models.
  • Heterogeneous Computing: The tight integration of Grace CPUs and Blackwell GPUs via NVIDIA NVLink enables optimized performance for a wide range of tasks.
  • Extensive Software Stack: Project DIGITS includes full access to NVIDIA’s libraries, frameworks, and AI models via the NVIDIA Developer portal, creating a complete AI development environment out of the box.

How Does It Compare to Consumer Systems?

As someone who regularly works on AI projects using an RTX 4070-powered setup with an i7 processor and 64GB of RAM, I find the contrast striking. While my system is capable of handling small to mid-sized AI tasks, such as fine-tuning pre-trained models or experimenting with neural networks, its 29 TFLOPS quickly hits its limits when dealing with larger models or datasets.

Compare Project DIGITS’s 1,000 TFLOPS — over 30 times the compute power of my RTX 4070, with the ability to handle workloads of up to 200 billion parameters. The inclusion of unified memory also means developers won’t face the bottlenecks of limited VRAM or CPU-to-GPU data transfer. Essentially, Project DIGITS is a purpose-built AI workstation that outclasses even high-end consumer setups for AI.

Future, Availability, and Pricing

NVIDIA’s Project DIGITS is a major step forward in making high-performance AI computing more accessible. It provides a desktop solution for developers and researchers to tackle advanced workloads without relying on cloud infrastructure, opening up access to powerful AI tools.

Project DIGITS will be available in May from NVIDIA and top partners, starting at $3,000.

Subscribe
Tag alert: Subscribe to the tag Embedded & AI and you will receive an e-mail as soon as a new item about it is published on our website!


You can see Arm’s press release here. (We told you about the potential acquisition of Arm by NVIDIA back in 2020. This never happened due to regulatory hurdles, but NVIDIA now has a 0.2% stake in Arm.)