Nvidia (NVDA) CEO Jensen Huang opened CES 2025 with a keynote outlining the company’s AI ambitions for consumer and enterprise users over the coming year. Huang debuted a number of new software technologies designed to help train humanoid robots and power various AI applications.
The company also announced new gaming hardware, including a pricey $1,999 GeForce RTX 5090 graphics card for desktop PCs.
Mr. Huang covered many things in his speech, but these are the biggest takeaways from the event.
Nvidia has a new AI superchip. No, this is not a successor to the company’s latest Blackwell chip, which debuted during the GTC 2024 conference last March. This is a smaller version of the high-power GB200 platform.
The chip, called GB10, is based on the GB200, and customers including Amazon, Google, Meta, Microsoft, and Tesla are picking it up en masse. However, the GB200 combines two Blackwell graphics processing units (GPUs) with one of Nvidia’s Grace central processing units (CPUs), while the GB10 combines one Blackwell GPU with one Grace CPU.
“This little thing is in full swing,” Huang said in his keynote address. “This computer is scheduled to be released in May.”
Nvidia isn’t promoting GB10 as a data center platform either. Instead, the company says it will be available on a desktop called “Project DIGITS.” Project DIGITS combines GB10 with 128 GB of unified memory and 4 TB of storage to create an AI computing system that sits on your desk.
In addition to new chips and desktops, Nvidia showed off its new Cosmos platform for developing physical AI systems such as robots and self-driving cars. The platform uses the World Foundation Model (WFM), an AI model that simulates real-world situations.
The idea is that companies can use Cosmos to simulate different usage scenarios in a virtual setting without using expensive robots or driving cars in the real world. to be able to assist in the development of the software necessary to power the
Nvidia also provided information about its Isaac GROOT Blueprint software. Developers can use the software via Apple’s Vision Pro headset to teach humanoid robots how to move in specific scenarios and situations.
On the vehicle front, Nvidia said Toyota, Continental and Aurora are using its automotive technology suite to power their advanced and autonomous driving systems.
“The (self-driving car) revolution has arrived,” Huang said in his keynote speech. “I predict that this will probably be the first multi-trillion dollar robotics industry.”
story continues
