Nvidia founder Jensen Huang opened the company’s artificial intelligence developer conference by stating that AI is experiencing ‘an inflection point.’ At GTC 2025, Huang focused on Nvidia’s advancements in AI and his predictions for the industry. He mentioned that demand for GPUs from top cloud service providers is increasing, expecting Nvidia’s data center infrastructure revenue to reach $1 trillion by 2028. Key announcements included details about Nvidia’s next-generation graphics architectures: Blackwell Ultra and Vera Rubin, named after the renowned astronomer. Blackwell Ultra is set for release in the second half of 2025, followed by Rubin AI chip in late 2026, and Rubin Ultra in 2027. Huang highlighted the progression of AI from perception to generative and agentic AI. He emphasized robotics as the next wave, powered by ‘physical AI,’ which understands concepts like friction and inertia. A significant aspect of this physical AI is synthetic data generation for model training. Huang also announced Isaac GR00T N1, an open-source foundation model for humanoid robot development, and the Cosmos series of AI models for generating cost-efficient photo-realistic video for robot training. General Motors plans to integrate Nvidia technology into its self-driving cars, collaborating on custom AI systems. Additionally, Nvidia unveiled its Halos system for automotive safety and an open-source physics engine called Newton for robotics simulation. The event concluded with a demonstration of a small robot named Blue. \”The age of generalist robotics is here,\” Huang declared.
