We Left the Field. They Didn't.

Jensen Huang's warning isn't about chips. It's about what happens when you concede a market and call it strategy.

Nvidia is banned from selling to China. China has also banned Nvidia from entering. Huang's joke: they're the first company in history banned from both sides simultaneously.

He's not actually laughing.

The Five-Layer Problem

Huang maps AI as a five-layer stack: energy, chips, infrastructure, models, applications. Most people only watch the model layer — ChatGPT vs DeepSeek, Claude vs Gemini. That's the wrong place to look.

Layer by layer, the picture is uncomfortable.

Energy: China has twice America's capacity, and their economy is smaller. Infrastructure: America takes three years to break ground on a data center. China built a hospital over a weekend. Models: America leads frontier models by about six months. But of the 1.4 million AI models in the world, most are open source — and China is way ahead there. Applications: polling shows 80% of Chinese citizens believe AI will do more good than harm. In America, it's the opposite.

Six months ahead on one layer. Behind on four.

What Conceding a Market Actually Means

America pulled out of China's AI market. The assumption was that this would set China back.

It didn't. Their semiconductor industry has been doubling every year while the rest of the world grows at 20-30%. Huawei — which Huang calls one of the most formidable technology companies the world has ever seen — built its entire ecosystem in isolation, then exported it everywhere.

This is the 5G playbook repeated. Get in early, build the ecosystem, become essential. By the time anyone notices, switching costs are too high.

Huang built CUDA 25 years ago. Nobody paid attention for a long time. That gave him time to connect every industry, every field, every application to the platform. That's the game China is now playing — and playing with urgency.

His conclusion: whoever applies the technology first and most broadly wins the industrial revolution. Electricity was invented in Britain. America applied it faster. That's why we're here.

We are not applying it faster right now.

The War Nobody Saw Coming Was Predicted a Year Ago

Palantir's CEO said it in August 2024. Nobody listened. Then it happened. In August 2024, Palantir CEO Alex Karp gave an interview to the New York Times. He described simultaneous pressure on three fronts — Russia, China, Iran — and argued that America's real vulnerability wasn't

Why Jensen Huang Has 60 Direct Reports and No One-on-Ones

The org chart isn't bureaucracy. It's architecture. And most companies get it wrong. Jensen Huang has more than 60 direct reports. He doesn't do one-on-ones with any of them. Most people hear this and assume it's chaos. It's the opposite.

The Best Investor You've Never Heard Of Admits He's Been Chickening Out

Stan Druckenmiller: 30% annualized returns for 30 years. No losing years. His edge isn't what you think. Stan Druckenmiller ran Duquesne Capital from 1981 to 2010. Thirty percent annualized returns. Zero losing years. When asked what his advantage is, he doesn't say intelligence. He says trigger-pulling.

The Best Investors Dream Less Than the Founders They Back

Doug Leone's biggest mistake at Sequoia: not believing enough. Doug Leone was an early investor in Nvidia, Google, and Apple. In each case, the company became something far beyond what anyone at Sequoia imagined. That's not a success story — it's a confession. His biggest