In a sweeping keynote at CES 2025, NVIDIA’s CEO Jensen Huang unveiled a bold vision for the future of artificial intelligence, showcasing innovations across gaming, autonomous vehicles, and robotics. The announcements solidified NVIDIA’s position at the forefront of AI’s evolution, from generative models to what Huang termed “physical AI” — AI systems capable of perceiving, reasoning, planning, and acting in the real world.
RTX 50 Series: A New GPU Powerhouse
Among the highlights was the debut of the GeForce RTX 50 Series, powered by NVIDIA’s Blackwell architecture. The flagship RTX 5090 GPU, boasting a staggering 92 billion transistors and achieving 3,352 trillion AI operations per second, was described by Huang as a “beast.”
“GeForce enabled AI to reach the masses, and now AI is coming home to GeForce,” Huang remarked, unveiling the GPU’s dual cooling fans and AI-enhanced real-time graphics capabilities. The RTX 50 Series rollout begins on 30 January, with the RTX 5090 and 5080 leading the charge, followed by the 5070 Ti and 5070 in February and laptop GPUs in March.
Enhancing gaming experiences further, NVIDIA introduced DLSS 4, featuring ‘Multi-Frame Generation’ technology, which can boost performance eightfold by generating three additional frames for every frame rendered. New advancements such as RTX Neural Shaders and RTX Mega Geometry promise unprecedented realism in video games, leveraging generative AI to render lifelike facial expressions and hair textures.
Cosmos: The Dawn of Physical AI
NVIDIA’s Cosmos platform aims to revolutionise robotics, industrial AI, and autonomous vehicles (AVs). Dubbed a “game-changer” by Huang, Cosmos integrates generative models, tokenisers, and video processing frameworks, enabling robots and AVs to simulate potential outcomes and optimise actions.
“The ChatGPT moment for general robotics is just around the corner,” Huang proclaimed, underscoring the platform’s transformative potential.
Cosmos, available under an open licence on GitHub, is already being adopted by leading players like XPENG, Hyundai Motor Group, and Uber. Pras Velagapudi, CTO of Agility, hailed Cosmos as a solution to data scarcity, enabling the generation of photorealistic scenarios for training robots without the cost of real-world data capture.
Empowering Developers with AI Blueprints
NVIDIA also launched new AI foundation models designed for RTX PCs. These models, packaged as NVIDIA NIM (Neural Interaction Model) microservices, aim to boost productivity, content creation, and enterprise applications.
Huang introduced AI Blueprints, open-source tools tailored for tasks such as fraud detection, content generation, and video management. These blueprints offer developers a starting point for creating bespoke AI solutions. One standout model, Llama Nemotron, provides developers with a platform to build and deploy advanced AI agents.
Ahmad Al-Dahle, VP of GenAI at Meta, noted, “Agentic AI is the next frontier of AI development. NVIDIA’s Llama Nemotron will accelerate the creation of custom AI agents, transforming enterprise workflows.”
Safer, Smarter Autonomous Vehicles
In the automotive sector, NVIDIA’s DRIVE Hyperion platform — built on the AGX Thor SoC — promises to deliver safer and more intelligent AVs. Combining NVIDIA’s DGX for AI model training, Omniverse for synthetic data generation, and DRIVE AGX for in-vehicle computing, the platform is designed to advance functional safety and autonomy.
Toyota, Mercedes-Benz, and Volvo are among the automotive giants leveraging NVIDIA’s solutions. Synthetic data, generated through NVIDIA’s AI data factories, enhances real-world datasets, accelerating the development of AVs.
Project DIGITS: AI Supercomputing in a Compact Form
Huang concluded his keynote with a surprise reveal: Project DIGITS, a compact yet powerful AI supercomputer. Powered by the GB10 Grace Blackwell Superchip, Project DIGITS offers the full power of NVIDIA’s AI stack in a portable design.
“This is NVIDIA’s latest AI supercomputer,” Huang announced, noting its potential to democratise AI supercomputing for developers and organisations alike. Set for release in May, Project DIGITS underscores NVIDIA’s commitment to accessibility and innovation.
A Vision for the Future
Reflecting on NVIDIA’s journey since pioneering the programmable GPU in 1999, Huang highlighted the transformative impact of AI over the past decade.
“Every layer of the technology stack has been fundamentally reshaped,” he said. With innovations spanning gaming, robotics, and autonomous systems, Huang predicted a future filled with groundbreaking advancements in AI and robotics.
“The best is yet to come,” he concluded, leaving the CES 2025 audience eager for what lies ahead.
