#494 – Jensen Huang: NVIDIA – The $4 Trillion Company & the AI Revolution
Summary
Lex Fridman interviews NVIDIA CEO Jensen Huang about the company’s shift to extreme co-design, optimizing entire data center racks rather than just individual GPUs. Huang details the strategic risk of launching CUDA on consumer GeForce cards, which established the developer base crucial for today’s AI revolution. He discusses overcoming scaling blockers through four laws—pre-training, post-training, test-time, and agentic scaling—while managing complex global supply chains.
Topics discussed
Intro, sponsors, and podcast housekeeping
NVIDIA's history: From GPUs to CUDA and AI
Leadership philosophy and long-term vision
The rise of AI agents and agentic systems
Hardware innovation: Rubin, HBM, and supply chain
Energy constraints and grid infrastructure challenges
Manufacturing complexity and the TSMC partnership
NVIDIA's moat, install base, and space applications
The future of computing: Token factories and economics
Jensen's mindset, learning, and handling uncertainty
AI in gaming, DLSS 5, and job displacement fears
Adapting to AI, intelligence as a commodity, and closing
Listen ad-free on Castria