In a nearly 2-hour presentation opening CES 2025, Nvidia CEO Jensen Huang introduced an “AI supercomputer” called Project Digits, powerful enough to run AI models.

Project Digits features a GB10 chip, which is roughly on par with the Grace Blackwell, paired with a Grace CPU. Nvidia partnered with semiconductor company MediaTek to create the GB10.

nvidia project digits banner lg offset laptop.jpg
Nvidia's Project Digits AI supercomputer. Photo: Nvidia

Unlike traditional GPUs for regular users, Project Digits is aimed at machine learning researchers, small companies, and universities that want to develop advanced AI but don't have billions of dollars to build data centers or buy cloud credits.

Nvidia says the supercomputer will cost around $3,000 when it hits shelves in May. The Project Digits name could also change.

CNBC commented that Project Digits is a completely different type of product compared to the GPUs that have made Nvidia explode over the past two years.

OpenAI and other AI model developers, along with cloud providers, are racing to buy Nvidia GPUs to run complex models and computing jobs.

Data center sales accounted for about 88% of Nvidia’s $35 billion in revenue in the most recent quarter, and Wall Street has focused on the company’s ability to diversify its business to reduce its dependence on a few large customers.

It’s “a little scary” that Nvidia has come out with such a good product, said Ben Reitzes, an analyst at research firm Melius. He also said Nvidia has grabbed all the attention at CES 2025 with a series of announcements, including graphics cards for gamers, robotic chips and a deal with Toyota.

David Bader, director of the Data Science Institute at the New Jersey Institute of Technology, said Project Digits could significantly improve the capabilities of researchers and universities.

Having worked on research projects with Nvidia, he said, Project Digits appears to have enough data and information processing power to train the largest, most advanced models.

Its capacity is equivalent to the $100 million supercomputers of companies like Anthropic, Google, Amazon...

For the price of a high-end laptop, users get a compact product that can be used at home or in the office. According to Reitzes, this could be Nvidia's first move into the $50 billion PC and laptop chip market.

(According to CNBC)