At the Nvidia keynote at CES 2025, CEO Jensen Huang didn’t waste anytime showing off the new GeForce RTX 50 Series. Huang walked onstage carrying the graphics card to a round of applause.
This was the most anticipated moment of the Nvidia event, but not the only big announcement. The AI computing company integral to the rise of generative AI had many more cards to play at the Las Vegas tech conference. Nvidia is now building its own AI models, fueling robotics and autonomous vehicle development, and bringing some of the most powerful computing tools to the masses. Here’s everything that was announced at the Nvidia keynote.
GeForce RTX 50 series
The big news of course was Nvidia’s new GPUs, the GeForce RTX 50 Series. The graphics cards are underpinned by Nvidia’s new RTX Blackwell architecture and consist of the flagship GeForce RTX 5090 as well as the GeForce RTX 5080, 5070 Ti, and 5070. The RTX 50 series is powered by 92 billion transistors, which gives it 3,352 trillion AI operations per second (TOPS) and boasts 1.8TB/s of memory bandwidth. Mashable’s Chance Townsend and Alex Perry have the full details on specs, availability, and pricing, but rest assured, it’s “just a beast,” as Huang put it.
Cosmos World Foundation Models
The graphics card giant is getting into the world model game with the introduction of Nvidia Cosmos. World models are the underlying technology for robotics training. And Nvidia has made its Cosmos World Foundation Models (Cosmos WFM) available as an open license platform available on Github, granting broader access to robotics developers that previously lacked these resources or expertise. “The ChatGPT moment for general robotics is just around the corner,” said Huang.
AI foundation models for RTX AI PCs, Llama Nemotron
Nvidia also introduced AI foundation models for LLM development. AI foundation models for RTX PCs are “offered as Nvidia NIM microservices” and use the GeForce RTX 50 Series GPUs. Additionally, Huang shared the top manufacturers are launching PCs that support NIM with its new graphics cards, adding “AI PCs are coming to a home near you.”
Another NIM microservice announcement introduced Llama Nemotron family of LLMs. Llama Nemotron uses Meta’s open-source Llama models are primed for agentic capabilities and “excel at instruction following, chat, function calling, coding and math, while being size-optimized to run on a broad range of NVIDIA accelerated computing resources,” according to the announcement. Llama 3.1 Nemotron 70B is now available in Nvidia’s API catalog.
Project Digits
In keeping with the theme of empowering developers with access to powerful computing tools, Nvidia unveiled Project Digits. The device is a supercomputer about the size of a Mac mini that easily sits on a desk and plugs into a keyboard and monitor. With its GB10 Grace Blackwell Superchip, Digits can run up to 200-billion-parameters LLMs without the need for cloud infrastructure. And it’s $3,000 a pop, which in the grand scheme of things, is a pretty accessible price point for small businesses and solo developers. Project Digits is expected this coming May.
DRIVE Hyperion AV platform
Nvidia has also been working hard in the autonomous vehicle department, introducing the DRIVE Hyperion AV platform, powered by the AGX Thor system-on-a-chip (SoC). DRIVE Hyperion is an “end-to-end autonomous driving platform,” that includes the SoC, sensors, safety systems, and a DriveOS operating system that car manufacturers can use to build their autonomous vehicles. Nvidia also shared that Toyota joins its growing list of partners that includes Mercedes-Benz, Jaguar Land Rover, and Volvo using its AV platform.
Mashable is on the ground live at CES 2025! We’re covering all the wildest and most important developments this week, so please keep checking back in with us. Want to submit a product you represent for our teams’ consideration as we identify the Best of CES? Here’s more info on how to do it.
