With all the attention focused on Nvidia’s new graphics cards (and for good reason, these cards are extremely powerful), the company has announced that it may end up making even bigger news. I released something. It’s a personal AI supercomputer called Project Digits (hereafter just Digits). It’s much cooler that way, so start now).
Digits is a small, Mac mini-like personal computer that fits on almost any desk. Connect it to your keyboard and monitor, connect it to power, and you’re ready to go.
But what’s inside is something quite special. Digits is powered by Nvidia’s new GB10 Grace Blackwell superchip, which delivers up to 1 petaflops of AI performance. Nvidia worked with MediaTek to make the chip more energy efficient. This means that the chips require the kind of power you get from standards to run. power outlet.
CES 2025 Highlights: What We’ve Seen So Far
The GB10 is paired with 128 GB of RAM and up to 4 TB of NVMe storage. All of this allows developers to run large language models (LLMs) with up to 200 billion parameters. Using Nvidia ConnectX technology, you can also pair two Digits computers to run models with up to 405 billion parameters.
Most of this content probably means little to anyone who is not in the AI app development business. But for example, ChatGPT 3.5 had 175 billion parameters (ChatGPT 4 was much larger, but I don’t know the exact number). On the other hand, Meta’s most powerful LLM, Llama 3, has 405 billion parameters). This means you can run a very powerful LLM from home without relying on cloud infrastructure.
mashable light speed
“Putting an AI supercomputer on the desk of every data scientist, AI researcher, and student will empower them to work on AI and shape the era of AI,” said Nvidia Founder and CEO. (CEO) Jensen Huang said in a written statement.
Developers using Digits will have access to Nvidia’s AI software library, including development kits, orchestration tools, frameworks, and models available in the Nvidia NGC catalog and Nviia Developer portal. You also have access to the Nvidia NeMo framework and Nvidia RAPIDS libraries.
Apple defended the oddly placed M4 Mac mini power button — here’s my take as a new owner
The biggest news here is probably the price. Digits will be available in May starting at $3,000. Such funding would be available to many small businesses and researchers and could be used to create and test AI apps.
If that still seems too expensive, try Nvidia’s Jetson. It’s a $249 AI home computer released last December that can process up to 8 billion parameters.
Mashable is coming live at CES 2025! Keep checking back as we cover all the wildest and most important developments this week. Want to submit a product you represent for our team to consider when determining Best of CES? Learn more here.
Topics
artificial intelligence CES
