Nvidia founder and chief executive Jensen Huang kicked off CES 2025 on Monday with a 90-minute keynote in which he announced a fleet of new tools to boost gaming, autonomous vehicles, and robotics.
At the major annual technology conference in Las Vegas, Huang announced the introduction of new products that use artificial intelligence (AI) to train robots and cars, gaming chips for neural rendering capabilities, and a new supercomputer for data scientists and students.
“It started with perception AI, understanding images, words, and sounds. Then generative AI, creating text, images and sound,” said Huang at the event. “Now, we’re entering the era of physical AI, AI that can proceed, reason, plan and act.”
The chief executive also introduced Cosmos, a new foundation model designed to accelerate physical AI development with new models and video data processing technology for robots, autonomous vehicles and vision AI.
The chipmaker said physical AI models are costly to develop and require vast amounts of real-world data and testing.
To provide a solution, Cosmos generates massive amounts of photoreal, physics- based synthetic data to train and evaluate their existing models, with developers able to build custom models.
Cosmos models will be available under an open model licence, with robotics and automotive companies including Agile Robots, Agility, Figure AI, Foretellix and Uber, already adopting the technology.
Huang said that the models aim to democratise physical AI and put general robotics in reach of every developer.
“The ChatGPT moment for robotics is coming,” he added. “Like large language models, world foundation models are fundamental to advancing robot and AV development, yet not all developers have the expertise and resources to train their own.”
Huang also announced the launch of RTX 50 series gaming chips, powered by Nvidia’s Blackwell AI technology, aimed at delivering AI and neural rendering capabilities to gamers and creators to power hyper realistic transformative AI experiences, with features including the ability to generate flawless human faces, enabling creators to complete workflows in less time.
The price of the chips will range from $549 to $1,999, with products being launched officially at the end of the month.
The Nvidia chief executive also unveiled AI foundation models designed to run on Nvidia RTX AI PCs that will feature services for developers and enthusiasts to build AI agents and creative workflow and productivity on PC.
The models include a wave of low-code and no-code tools, enabling enthusiasts to use AI models in complex workflows via simple graphical user interfaces.
“AI is advancing at light speed, from perception AI to generative AI and now agentic AI,” said Huang.
The list of new developments also featured NVIDIA Project DIGITS, a personal AI supercomputer that aims to help AI researchers, data scientists and students worldwide offering petaflop of AI computing performance for prototyping, fine-tuning and running large AI models.
With Project DIGITS, users can develop and run inference on models using their own desktop system, then deploy the models on accelerated cloud or data centre infrastructure.
The chipmaker also announced a partnership with Toyota, Aurora and Continental for safe next-gen autonomous vehicle development, with the chipmaker providing accelerating computing and AI through the implementation of NVIDIA DriveOS operating system to provide advanced driving assistance capabilities.
“The autonomous vehicle revolution has arrived, and automotive will be one of the largest AI and robotics industries,” said Huang.
/cdn.vox-cdn.com/uploads/chorus_asset/file/23962441/acastro_STK067__02.jpg "T-Mobile is once again being sued over its 2021 data breach")
