The main event was the GeForce RTX 50-series, the company’s first flagship desktop GPU offering in two years. The graphics cards are powerful enough to run AI models and can boost gaming frame rates up to eight times with DLSS 4 neural rendering technology.
Additionally, Jensen HuangWearing his signature leather jacket, unveiled a new NVIDIA supercomputer for AI researchers, a method for humanoid robots that allows them to learn through imitation, the Global Foundation Model Platform, and Partnership with Toyota for autonomous vehicles.
GeForce RTX 50 series desktop GPUs
The latest in the new GeForce series is the RTX 5090, which has 92 billion transistors and delivers 3,352 trillion AI operations per second.
“The GeForce RTX 50 Series delivers breakthroughs in AI-powered rendering, including neural shaders, digital human technologies, geometry and lighting,” claims NVIDIA. Press release.
The RTX 5090 will start at $1,999, which is a $400 increase over its predecessor, the RTX 4090, but also Twice as fast. The RTX 5080 delivers 1,800 TOPS and will start at $999. Both will be available from January 30.
The RTX 5070 Ti delivers the same 1,400 TOPS as the RTX 4090, and will cost $749. Meanwhile, the RTX 5070 offers 1,000 TOPS for $549. Both will be available in February.
Acer, ASUS, Dell, HP, Lenovo, and Razer have designed laptops that include the new RTX 50-series. Devices with all the chips will be available in March except for the RTX 5070 laptop, which will launch in April.
“Fusing AI-powered neural rendering and ray tracing, Blackwell is the most important innovation in computer graphics since we introduced programmable shading 25 years ago,” Huang said in the speech.
Personal supercomputers for AI researchers
On top of the new chip series, NVIDIA introduced the world to its personal AI supercomputer; Project Digits. It is intended for use by AI researchers, data scientists, and students. “Supercharge AI innovation,” According to NVIDIA.
The DIGITS computer features the GB10 Grace Blackwell Superchip, which consists of a GPU and a 20-core CPU, providing up to a petaflop of computational power. It can run large language models of up to 200 billion parameters—the number of parameters is usually proportional to the performance capacity.
Complete with 128GB of memory, up to 4TB of flash storage, and NVIDIA’s DGX operating system, the computers will be available in May. It will start at $3,000.
Simulation capability for humanoid robots
In March 2024, NVIDIA unveiled GR00T – a general-purpose foundation AI model for humanoid robots. At CES 2025, Huang announced a new model, or The blueprintFor GR00T that allows him to learn through imitation.
In other words, individuals will be able to perform the movements they want the robot to perform, it will capture them as one. The digital twinand then execute them repeatedly. All this can be done using an Apple Vision Pro headset.
The idea is that the technology will make it easier to train large fleets of robots deployed in manufacturing settings. “Physical AI will revolutionize the $50 trillion manufacturing and logistics industries. Everything that moves — from cars and trucks to factories and warehouses — will be robotic and embodied by AI,” Huang said.
This completes the new blueprint. Blueprint for real-time CAE digital twins Launched in November, it facilitates the creation of interactive simulations for engineering applications. Both leverage NVIDIA’s Omniverse platform for real-time 3D collaboration.
In addition, Huang unveiled other blueprints that could. Build custom AI agents for enterprises With NVIDIA NIMs: microservices containing APIs, domain-specific code, optimized inference engines, and enterprise runtimes required to run generative AI. He also announced. Llama Nemotron family of open large tongue models To help create and deploy agents. TechRepublic predicted at the end of the year that The use of AI agents will increase in 2025..
A global foundation model for robotics
Huang announced a new platform for “world foundation models”. These AI models are inspired by how humans create their own mental models of the world, called Cosmos. The first Cosmos models released are designed specifically for robotics, focusing on “physics-based simulation and artificial data generation”.
“Cosmos WFMs are aimed at physical AI research and development, and can create physics-based videos from a combination of input, image and video, as well as input such as robot sensors or motion data.” NVIDIA.
See: NVIDIA: Blackwell architecture to accelerate AI products in late 2024
All of them will be trained on 9,000 trillion tokens, or chunks of video data, including 20 million hours of data from autonomous driving, robotics, human interactions, and other domains. Companies like Uber and Wayve have already committed to deploying Cosmos models, including self-driving vehicles.
Cosmos models will be publicly available via NVIDIA, GitHub, and Hugging Face.
Autonomous Driving Partnership with Toyota
Finally, Huang announced a new partnership between NVIDIA and Toyota to develop next-generation autonomous vehicles. Upcoming Toyota models will feature NVIDIA’s DRIVE AGX Orin supercomputer and DriveOS operating system.
The news comes shortly after the announcement of a partnership between autonomous vehicle companies Aurora and Continental. Expand the deployment of driverless trucks using NVIDIA DRIVE.. Huang predicts NVIDIA’s automotive revenue to reach $5 billion. Its key note.
“The autonomous vehicle revolution has arrived, and automotive will be one of the biggest industries for AI and robotics,” he said in a statement. Press release.