Physik AI as the next evolution - Nvidia CEO demonstrates breakthrough in AI development

Nvidia-CEO Jensen Huang presented a clear picture of modern Artificial Intelligence at the Davos Forum: the technology is at a turning point. After years of overpromising and hallucinations, we are now experiencing a phase where AI models actually solve real problems. Huang identified three key advancements driving this transformation — and one of them could fundamentally change the industry.

From Illusions to Practical Application - Agentic AI Sets New Standards

Huang first emphasized a crucial shift that the AI industry has undergone over the past year. While AI models initially struggled with massive hallucinations, they now demonstrate real competence in research environments. These models master logical reasoning, can plan tasks, and answer complex questions — all without specialized training. This development gave rise to a completely new paradigm: the so-called Agentic AI, which autonomously analyzes problems and develops solutions.

Open-Source Ecosystem as a Game-Changer

The second significant advancement concerns the democratization of AI technology. The introduction of the first open-source inference model DeepSeek marked a critical point for various industries, according to Huang. Since then, a vibrant ecosystem of open-source models has emerged, enabling companies, research institutions, and educational organizations to implement advanced AI systems themselves. This development significantly reduces dependence on proprietary solutions and accelerates practical AI integration across all sectors.

Physics AI - AI Conquers the Physical Reality

The third and possibly most revolutionary advancement lies in Physics AI. Here, artificial intelligence reaches a new dimension: it understands not only language but also captures the physical world in its full complexity. AI models today demonstrate understanding of biological proteins, chemical processes, and physical laws. In the field of classical physics, it has been shown that AI systems can grasp and apply fundamental concepts such as fluid dynamics, particle physics, and even quantum physics. This opens up entirely new possibilities for materials science, drug development, and technological innovation.

These three pillars — intelligent autonomous agents, open AI ecosystems, and Physics AI — mark, in Huang’s view, the transition from experimental AI to practical, world-changing technology. Especially Physics AI stands at the boundary between theoretical feasibility and real-world applicability.