About Jensen Huang
Jensen Huang is the founder and CEO of NVIDIA, the company that accidentally became the backbone of AI. His bet on GPU computing for machine learning has made NVIDIA one of the world’s most valuable companies.
Career Highlights
- NVIDIA (1993-present): Co-founder and CEO
- GPU computing pivot (2006): CUDA launched, enabling general-purpose GPU computing
- AI bet (2012): Recognized deep learning would need massive parallel compute
- $3 trillion company (2024): Briefly the world’s most valuable company
Notable Positions
On the Five-Layer AI Stack
Jensen’s framework for AI competition:
- Energy - China has 2x US capacity
- Chips - US is “generations ahead” but 4-8x higher manufacturing costs
- Infrastructure - China can build a hospital in a weekend; US data centers take 3 years
- Models - US leads frontier by ~6 months; China dominates open source
- Applications - 80% of Chinese believe AI will do good; US is inverted
On the Energy Constraint
“Half a trillion dollars of AI supercomputers are planned for Trump’s term, but without energy growth, it’s just blueprints. The US can’t rely on the power grid - must build behind the meter.”
On the China Ban
“Nvidia is banned from China by both sides, conceding the world’s second-largest AI market. Meanwhile, Huawei is doubling revenue annually while Western semiconductors grow 20-30%.”
Key Quotes
- “Energy is layer zero.”
- “Data center GPUs have 1.5M parts and consume 200kW.”
- “The US needs to compete in China, not just defend.”
Related Reading
- Scaling Laws - What NVIDIA’s GPUs enable
- AI Infrastructure - Jensen’s five-layer stack
Video Mentions
CES 2025 keynote
Every 10 to 15 years, the computer industry resets. Except this time there are two simultaneous platform shifts—to AI and to physical AI.
Agentic AI revolution
Agentic models that can reason, look up information, use tools, plan futures, simulate outcomes—all of a sudden started to solve very important problems.
Physical AI and Cosmos
Cosmos is the world's leading world foundation model. It turns compute into data, training AVs for the longtail and robots how to adapt for every scenario.
Five-layer AI stack
AI competition is a five-layer stack: energy, chips, infrastructure, models, applications. The US leads chips and models but lags energy and infrastructure.
Infrastructure speed gap
China can build a hospital in a weekend; US data centers take 3 years. Half a trillion dollars of AI supercomputers are planned, but without energy growth, it's just blueprints.
China market dynamics
Nvidia is banned from China by both sides, conceding the world's second-largest AI market. Meanwhile Huawei is doubling revenue annually.
Corporate character philosophy
Pain and suffering - that's our secret sauce. Come work with me. That's my gift. Getting Grace Blackwell into production almost broke our company's back, but we wouldn't let it. That's 100% character, not intelligence.
AI and future of work
Instead of fewer jobs, I feel what's likely to happen is we're going to be busier than ever. We're going to think of more ideas of things we can solve now. All those things off the table are now on the table.
Democratization of software development
One of my favorite things is vibe coding. Anybody can be a software programmer now. AI is going to close the technology divide.
SaaSpocalypse counter-narrative
I think the markets got it wrong. These agentic AI will be intelligent software that uses these tools on our behalf. Nobody's going to service better than ServiceNow.