NVIDIA (NVDA) looks to continue its momentum from 2024 to 2025 with a number of AI-focused announcements to CES in Las Vegas on Monday. CEO Jensen Huang took the stage during the company’s keynote, laying out his vision for everything from AI software that will power robots and self-driving cars to a new AI supercomputer that holds on your desktop.

Nvidia’s stock price rose as much as 4.7% ahead of Monday’s keynote, as Wall Street braced for the latest offerings from the AI ​​darling. The company’s shares are up 205% in the past 12 months thanks to its prescient investments in AI hardware and its CUDA software, which allows developers to use its chips to run AI programs.

The latest announcements have focused on how programmers can take advantage of Nvidia’s existing hardware, its Hopper and Blackwell platforms. The company could launch its next-generation chip during its GTC conference in March.

At Monday’s event, Huang demonstrated Nvidia’s latest Blackwell-based chip, the GB10 superchip. It is a miniature version of the GB200 superchip, which combines a Grace central processing unit (CPU) with two Blackwell graphics processing units (GPUs). The smaller GB10 combines a Grace CPU and a Blackwell GPU.

Nvidia says the chip will be available in a small desktop system called Project DIGITS and will come with 128GB of memory and 4TB of storage. The company says the setup is powerful enough for researchers interested in “prototyping, fine-tuning, and running large AI models.”

Project DIGITS will start at $3,000 and will be available in May from Nvidia and its OEM partners.

Beyond its new chip and desktop, Nvidia also released its openly licensed Cosmos platform for developing physical AI systems. The platform uses World Foundation Models, or WFMs, which are AI models that simulate real-world conditions. Physical AI systems include technologies such as humanoid robots and self-driving cars.

The idea is that companies use Cosmos to develop the software needed to power robots and self-driving cars, simulating various usage scenarios in a virtual environment without having to use expensive robots or put cars on the road. road in the real world.

“The ChatGPT moment for robotics is coming,” Huang said in a statement.

“Like great language models, global foundation models are fundamental to advancing robot and AV development, but not all developers have the expertise and resources to form their own,” he said. Huang explained. “We created Cosmos to democratize physical AI and put general robotics within the reach of every developer.”