The Next Wave: Why Jensen Huang Believes Physical Sciences Are Essential

Introduction

In a recent press briefing during his trip to Beijing, Nvidia CEO Jensen Huang dropped a surprising piece of advice for today’s students: if he were 20 again, he would major in the physical sciences rather than software. This recommendation reflects his vision for the future of artificial intelligence—what he calls the “Physical AI” era, where machines must understand and interact with the real world.

From Early Graduate to AI Visionary

Huang’s own academic journey began in the hardware realm. He earned his bachelor’s degree in electrical engineering from Oregon State University at age 20, then completed his master’s at Stanford University in 1992. A year later, he co-founded Nvidia in a San Jose Denny’s booth alongside two fellow engineers. Over three decades, the company has grown from a modest GPU startup into the world’s most valuable public firm, recently hitting a $4 trillion market cap milestone.

Despite his software-driven success, Huang now emphasizes a return to the fundamentals: “If you are a 22-year-old version of Jensen just graduated today in 2025 with the same ambition, you’d focus more on the physical sciences than the software sciences,” he told journalists in Beijing.

Understanding the Four Waves of AI

In multiple talks—including a keynote at the GPU Technology Conference—Huang has framed AI’s evolution in four distinct waves:

Perception AI: Machine vision, speech and pattern recognition (sparked by AlexNet in 2012).
Generative AI: Models that produce text, images and code, exemplified by GPT and diffusion networks.
Reasoning AI: Agentic AIs capable of solving new, unseen problems and adapting to changing conditions.
Physical AI: Systems that grasp physical laws—friction, inertia, force—and apply them within real-world objects.
Source: CNBC

What Makes Physical AI Different?

Traditional AI excels at recognizing patterns in data—classifying images, translating languages or generating prose. But Physical AI must do more: it needs to predict how objects move, estimate the force required to grip a fragile component, and even infer the presence of hidden obstacles based on partial sensory input. This capability hinges on a deep knowledge of physics.

For example, understanding object permanence (the idea that things continue to exist when out of sight) is trivial for a child but remains a research frontier for AI robots. Embedding such reasoning allows machines to navigate cluttered factory floors, assist in warehouses and eventually work alongside humans in manufacturing, logistics and beyond.

Academia: A Physics-First Curriculum

Given this horizon, Huang argues that students should ground themselves in:

Classical Mechanics: Kinematics and dynamics form the bedrock of motion planning.
Thermodynamics & Statistical Physics: Key for understanding energy efficiency and sensor fusion in robotics.
Materials Science: Critical to selecting and designing the right components for physical prototypes.
Control Theory & Electronics: Foundations for building stable, responsive embedded systems.

Such a curriculum deepens one’s intuition about real-world phenomena—an advantage as AI moves from digital to physical domains.

Industry Implications

Across sectors, the demand for robots capable of precise manipulation is surging. Amazon, Tesla and Toyota all invest heavily in robotics, while startups like Apptronik and Figure.ai are racing to deploy humanoid platforms. Yet without robust physical reasoning, these systems struggle with basic tasks such as picking oddly shaped parts or navigating uneven terrain.

Huang envisions highly automated factories where AI-driven robots handle tedious or hazardous tasks, alleviating global labor shortages. “In the next decade, we’ll build plants and factories that are highly robotic,” he said. “They’ll help us deal with the severe labor shortage we have all over the world.”

Advice for Aspiring Innovators

What does this mean for the next generation?

Pursue multidisciplinary studies: Physics plus computer science plus hands-on engineering.
Engage in research projects involving robotics, control systems or physical simulations.
Leverage open-source platforms—Omniverse, ROS and PyBullet—to prototype Physical AI applications.
Build strong foundations in mathematics—especially linear algebra and differential equations.

Broader Context: Back to Basics

Huang’s stance echoes similar calls from tech leaders. Telegram founder Pavel Durov recently urged students to master mathematics, while Elon Musk tweeted, “Physics (with math).” As AI grows more sophisticated, the ability to deploy it in the real world depends on engineers who truly grasp the laws that govern matter and energy.

For those aiming to be at the vanguard of Physical AI, academic rigor combined with practical experience will be the winning formula.

Conclusion

Jensen Huang’s counterintuitive advice—to favor the physical sciences over software—underscores where he believes the AI frontier lies. As businesses build the next generation of intelligent robots, graduates with deep knowledge of physics, materials and control systems will be best positioned to lead this transformation. In Huang’s view, tomorrow’s innovators must understand not only code, but also the real-world forces their creations will confront.

Are you ready to anchor your education in the physical sciences and ride the Physical AI wave? The future of robotics and automation beckons those who dare to master both bits and atoms.