Nvidia CEO Jensen Huang Urges Studying Physical Sciences for the Next AI Wave

Nvidia CEO Jensen Huang suggests that mastering physical sciences, such as physics, mechanics, and materials science, is crucial for the next wave of artificial intelligence, which he calls Physical AI. He believes that understanding the laws of physics and how data interacts with the real world will be essential for machines to interact with the physical world in meaningful ways and become more advanced. This next wave of AI will be a major focus area for robotics and human-technology collaboration.

In a significant advancement for materials science, researchers at the USC Viterbi School of Engineering have developed an AI model, Allegro-FM, capable of simulating the behavior of billions of atoms simultaneously. This breakthrough has the potential to revolutionize the field of concrete production, offering a path towards carbon-neutral and longer-lasting materials [1].

The development of Allegro-FM was driven by a desire to address the environmental impact of concrete production, which accounts for about 8% of global CO2 emissions. By simulating different concrete chemistries virtually, the AI model can test various compositions without the need for expensive real-world experiments. This capability could accelerate the development of concrete that acts as a carbon sink rather than a carbon source [1].

One of the key breakthroughs of Allegro-FM is its scalability. While existing molecular simulation methods are limited to systems with thousands or millions of atoms, Allegro-FM demonstrated 97.5% efficiency when simulating over four billion atoms on the Aurora supercomputer at Argonne National Laboratory. This represents computational capabilities roughly 1,000 times larger than conventional approaches [1].

The AI model covers 89 chemical elements and can predict molecular behavior for applications ranging from cement chemistry to carbon storage. This versatility makes Allegro-FM a powerful tool for advancing materials science and engineering. Additionally, the model can simulate mechanical and structural properties of concrete, providing insights into how different compositions affect the material's performance [1].

The implications of this technology extend beyond environmental sustainability. Modern concrete typically lasts about 100 years, but ancient Roman concrete has lasted for over 2,000 years. The recapture of CO2 in concrete can enhance its durability, making it more robust and long-lasting. By incorporating CO2 into the concrete's composition, the material can become more resilient and better suited for long-term applications [1].

The development of Allegro-FM highlights the potential of AI in advancing physical sciences. Nvidia CEO Jensen Huang has emphasized the importance of mastering physical sciences for the next wave of AI, which he calls Physical AI. This next wave of AI will be crucial for robotics and human-technology collaboration, enabling machines to interact with the physical world in meaningful ways [3].

The UK government has also recognized the importance of AI and supercomputing in driving innovation and economic growth. The launch of Isambard-AI, the most powerful supercomputer in the UK, signals a significant acceleration in the government's drive to bolster AI research nationwide. The government has pledged £1 billion to increase Britain's compute capacity 20-fold by 2030, including the establishment of AI "growth zones" and the funding of new supercomputers in Edinburgh and Wales [2].

As AI continues to advance, its applications in materials science, robotics, and other fields are expected to grow. The ability to simulate complex systems at unprecedented scales opens up new possibilities for innovation and problem-solving. The development of Allegro-FM and the launch of Isambard-AI are just two examples of how AI is transforming these industries and driving progress towards a more sustainable future.

References:
[1] https://viterbischool.usc.edu/news/2025/07/discovering-new-materials-ai-can-simulate-billions-of-atoms-simultaneously/
[2] https://ca.finance.yahoo.com/news/uk-most-powerful-supercomputer-launches-170100368.html
[3] https://blog.roboflow.com/ai-in-robotics/