At GTC 2025, Nvidia CEO Jensen Huang introduced the company’s next-generation Rubin AI chips, marking a significant step forward in artificial intelligence development. In a keynote address, Huang highlighted how AI has evolved rapidly over the past decade, progressing from computer vision to generative AI and now to agentic AI—systems capable of reasoning and understanding context.
New GPU Architectures and Industry Growth
Huang announced Blackwell Ultra, the latest iteration of Nvidia’s GPU architecture, set for release in late 2025. Its successor, the Rubin AI chip, is expected in 2026, followed by the Rubin Ultra in 2027. These developments come as Nvidia anticipates data center infrastructure revenue to hit $1 trillion by 2028, driven by growing demand from major cloud providers.
Huang described this phase of AI as an “inflection point,” with advancements in robotics and reinforcement learning shaping the industry’s future. He emphasized that AI systems are now capable of not only responding to prompts but also learning through synthetic data generation—a method that reduces dependence on human-driven training.
Robotics and the Rise of Physical AI
A major highlight of the event was Nvidia’s focus on robotics and physical AI—the next frontier in artificial intelligence. Huang introduced Isaac GR00T N1, an open-source foundation model for humanoid robot development, alongside an updated Cosmos AI model designed for simulated training.
The use of synthetic environments for reinforcement learning, he explained, allows AI to develop understanding of physics-based concepts such as friction and inertia. This approach accelerates training for robots and lowers costs compared to traditional data collection methods.
Experts in AI research welcomed the move. Benjamin Lee, a professor at the University of Pennsylvania, noted that simulated training environments provide an accessible way for both industry leaders and academic researchers to refine robotics models.
Also read: Nvidia AI Surge Continues Despite Market Doubts
Nvidia’s AI in Automotive and Beyond
Huang also revealed Halos, an AI system dedicated to enhancing automotive safety and autonomous driving. He stated that Nvidia is the first company to have every line of code in its AI safety system fully assessed. General Motors is set to integrate Nvidia’s technology into its upcoming fleet of self-driving cars, using Omniverse and Cosmos AI to train models for advanced manufacturing and automation.
A Glimpse Into the Future of Robotics
To conclude the event, Huang showcased Newton, an open-source physics engine for robotics simulation, developed in collaboration with Google DeepMind and Disney Research. In a lighthearted moment, a small, boxy robot named Blue appeared on stage, following Huang’s commands and symbolizing the dawn of generalist robotics.
“The age of generalist robotics is here,” Huang declared, signaling a future where AI-powered machines will play a central role in everyday life.
