Nvidia's CEO, Jensen Huang, took the stage at CES 2025 in Las Vegas to unveil new advancements that could transform computing across multiple industries.
Huang delivered a 90-minute keynote to around 6000 attendees, revealing Nvidia's roadmap for 2025. He also emphasized the company's crucial role in the AI revolution, "We're entering the era of physical AI— AI that can perceive, reason, plan, and act."
With Nvidia's valuation at $3.66 trillion and its stock trading at a record high, Huang showcased cutting-edge advancements that solidified the company's gaming, robotics, and autonomous systems dominance.
RTX 50-Series GPUs: A Leap in Gaming Technology
Nvidia introduced its next generation RTX 50-series GPUs, built on the Blackwell architecture. The lineup includes the RTX 5070, 5070 Ti, 5080, and flagship 5090, all priced between $549 and $1999. The RTX 5090, with 92 billion transistors, provides unsurpassed AI and gaming capabilities.
It's equipped with DLSS 4, Multi-Frame Generation technology that increases frame rates by up to 8x while improving realism with neural shaders and superior rendering capabilities.
DLSS 4: Revolutionizing Graphics Rendering
The fourth iteration of the Deep Learning Super Sampling (DLSS) introduces Multi-Frame Generation, which generates three additional AI-predicted frames for each traditionally produced frame.
This breakthrough enhances his performance, and RTX Neural Shaders improves texturing and lighting.
Cosmos Foundation Models: Transforming Robotics and AVs
Nvidia introduced Cosmos, a platform for creating synthetic, physics-based videos to train robotics and self-driving vehicles. Cosmos is designed to imitate real-world situations, reducing the need for expensive real-world data.
Some of the earliest adopters include robotics and automotive leaders like Toyota, Uber, and Hyundai.
Project DIGITS: AI Computing on Every Desk
Project DIGITS, Nvidia's first desktop AI computer, integrates the Grace Blackwell Superchip with the complete Nvidia AI stack.
This $3000 workstation designed for AI researchers, students, and developers can handle models with up to 200 billion parameters and will be available in March.
Llama Nemotron Models: Enterprise AI at Scale
Nvidia's open-source Llama Nemotron family provides enterprises with tools for creating AI agents.
These models, which may function across clouds, data centers, and workstations, ensure capability and efficiency for applications ranging from customer care to fraud detection.
Nvidia Mega: Robotics Fleet Management Redefined
Mega, Nvidia's robotics split management technology, enables autonomous robots to work together seamlessly in warehouses.
Their integration of AI, Omniverse, and Issac technologies enables businesses to optimize operations through enhanced modeling and testing.
AI Game Characters and RTX-Powered Virtual Assistants
Nvidia uses its Ace technology to create autonomous game characters that act and strategize like human gamers.
The RTX G-Assist virtual assistant will be released in February, revolutionizing gameplay aid.
Advancing Autonomous Vehicles
The Nvidia DRIVE Hyperion AV platform, powered by the AGX Thor SoC, has AI-enabled safety and autonomous features.
Nvidia's partnership with Toyota intends to speed up the development of next-generation AVs using synthetic data from Omniverse and Cosmos.
Empowering Developers with AI Blueprints
Nvidia released an AI blueprint to help developers build and use AI agents for enterprise use cases ranging from fraud detection to content creation. These designs, driven by NIM Microservices, are intended to simplify AI integration.
Jensen Huang's keynote set the stage for a transformative year in computing, solidifying Nvidia's position as a leader in the AI-driven future of gaming, robotics, and beyond.