Columbia Energy Exchange Jensen Huang: The Power of Intelligence Infrastructure
Aug 26, 2025
Jensen Huang, the visionary founder and CEO of NVIDIA, shares his insights on AI's role in reshaping energy systems. He discusses the paradox of AI demanding high energy while offering solutions to reduce consumption in sectors like manufacturing and weather forecasting. Innovations like silicon photonics are highlighted for their potential to save power in data centers. Huang's concept of 'AI factories' aims to optimize energy use, revolutionizing our energy infrastructure and enhancing efficiency across industries.
AI Snips
Chapters
Books
Transcript
Episode notes
Accelerated Computing Enabled Modern AI
- Jensen Huang says accelerated computing cut energy per computation and enabled modern machine learning.
- Training is energy-intensive but reduced compute cost made large models practical.
Train Big, Run Small Models
- Jensen explains large models are trained once then distilled into many smaller, efficient models for deployment.
- Inference runs far more often and consumes far less energy than repeated training.
AI Weather Models Slash Energy Use
- Jensen gives weather forecasting as an example where AI emulators use 10,000x less energy than traditional supercomputers.
- These surrogate models enable continuous, low-energy predictions for power systems and renewables.
