Jensen Ran Nvidia Like No Other CEO. China Is Holding Up 600,000 Nvidia Chips It Still Can’t Replace. WHY?

Do you know that last month, Jensen Huang said he hoped to run NVIDIA for another 20 years? Well, Thursday morning, he resigned. 33 years, gone in one press release. You've heard enough about the affairs, probably, and the alleged CCP spy. Here's what nobody's saying. NVIDIA, the engine of the whole AI boom, Just lost the one man it never planned to live without. And meanwhile, China, the customer Nvidia fought hardest to keep, is leaving 600,000 Nvidia chips stuck in a Hong Kong warehouse. If you own an index fund or a 401k, this is your money too. So who replaces the irreplaceable? You have to keep watching because it's just going to be so good. Please give us a like and follow our channel. Alright, let's get into it. Let's start with how Jensen ran the place, because it was unusual. At most companies, an idea climbs a ladder. You tell your manager, your manager tells a director, and the director tells a VP. By the time it reaches the CEO, it's been ruined three times. It's a game of telephone. Well, Jensen hated the telephone game, so he took out the ladder. Dozens of senior leaders reported straight to him, somewhere between 36 and 60, depending on the year. No status reports. Decisions happen in one open meeting where everyone heard the same thing at the same time. Less telephone game, more group chat. Well, his reasoning? Rigid structures cannot adapt. Silos kill invention. And it worked. That's how a company that made graphics cards for video games became the engine of the entire AI boom. But here's the catch. When every road leads to one person. That person becomes a bridge. Take away the bridge. And what happens to the roads? Wall Street has a name for this, key man risk, and NVIDIA might be the biggest example in history. Tech analyst Dan Ives put it bluntly, this is the definition of key man risk. There's no public succession plan. NVIDIA had never, ever even had a board chair. This week, it named its very first one ever. Well, at a company worth trillions of dollars. And in that same interview, Jensen said the next CEO is ideally already inside NVIDIA. But instead, the board handed the keys to the CFO as interim CEO and started a search. Headhunters are reportedly already looking outside at names like AMD's Lisa Su, who happens to be Jensen's own distant cousin. Yes, really. His mother's brother is her grandfather. Well, that's not confusing. And while we're dreaming big, why not Elon Musk? Jensen once called him the ultimate GPU. Well, half joking. We heard Trump probably likes Tim Cook better. Even the family question came up. His son and daughter both work at Nvidia. Jensen's answer? Well, if they want the top job, they will have to earn it like anyone else. Well, to be fair, not everyone is panicking. Some analysts say NVIDIA's real moat isn't one man. It's the software the whole AI world already runs on. We'll get into that part. But Founder ETFs, a fund that invests specifically in founder-led companies, said it would sell its NVIDIA shares if Jensen left. Well, he left. And whoever takes this job walks into a very strange problem. And right now, about 600,000 Nvidia chips are sitting in Hong Kong, ready to go, and the people holding them back? Well, not Washington, Beijing. Wait, isn't China supposed to be desperate for these chips? Well, to answer that, we need a simpler question first. Can China actually go it alone? Short answer, sort of and not really. You know, in the first half of this year, Nvidia shipped 678,000 AI chips in China and Huawei shipped 572,000. That's 34% of the market for Nvidia and 28% for Huawei. Now look at the direction. Compared with a year ago, Nvidia's sales in China fell 44%, and Huawei's nearly doubled, a six-point gap. Close on paper, though. But that's only the official count. One alleged smuggling network run through a company in Singapore linked to the woman at the center of the affair is tied to about 240,000 more Nvidia chips. Well, rough math, that would put Nvidia near 41%. Huawei, 25. So that's a 16-point gap, not 6 anymore. And that's just one backdoor we know about. So why not copy NVIDIA's best chips and finish the job? Because designing a chip and building one are two completely different things. Think of a Ferrari. Well, you can photograph the engine, draw perfect blueprints, but building it at scale without flaws takes the factory, the tools, and the people who've done it for decades. Well, for the most advanced chips, the key tool is a machine that prints circuits thousands of times thinner than a human hair. It's called EUV lithography. And exactly one company on earth makes it, ASML, in the Netherlands. Well China cannot buy one, and you can't exactly smuggle one out because it weighs about 180 tons, the size of a bus, literally. So China tried the next very best thing. According to the Reuters, it quietly recruited former ASML engineers, some, surprise, working under fake names, and built a prototype from old parts of the machines, but China's version takes up to almost an entire factory floor. Same job, far clunkier. It's still missing the ultra-precise mirrors only Germany's Zeiss can make, and it hasn't produced one single usable chip yet. Well, the target? Commercial production around 2028 and at the earliest, well, maybe 2030, who knows. In case you are curious who does exactly what, Germany's Zeiss, that company, makes the mirrors, and the Netherlands' ASML makes the machines. Taiwan's TSMC uses those machines to make the chips. Without EUV, Chinese factories use older machines and repeat steps over and over, more steps, more mistakes. In 2025, Morgan Stanley estimated only about 30% of Huawei's flagship AI chips were coming out as usable. Imagine baking 10 cakes and throwing out 7 of them. Then there's the software. You know, nearly every AI developer on earth works in Nvidia's programming language called CUDA. Here's why this matters so much. A chip is only as useful as the software that knows how to use it. So, what does CUDA actually do? NVIDIA's chip is like a giant kitchen with thousands of cooks. CUDA is the cookbook. Almost 20 years of recipes telling every cook what to chop, stir, and plate all at the same second. Millions of chefs learn to cook from it. That's the millions of engineers around the world. And it's written for one kitchen only, NVIDIA's. Chinese developers can still use CUDA, but only on NVIDIA chips. Put a Chinese chip like Huawei's in that kitchen and two things go wrong. First, the cooks are slower. That's the hardware gap. But the bigger problem is the cookbook. Nvidia's cookbook only works on Nvidia's kitchen, so Huawei had to write its own, called C-A-N-N. You know, every recipe has to be rewritten and retested. That's a month of engineering, every single time. On top of all of this, switching away from CUDA means retraining the world. Think of your iPhone. Even if another phone is just as good, your apps, your photos, your habits are all built around it. Switching is a pain. Now picture that pain for every AI lab on the planet. Well, that's CUDA. So here's the truth. Building a competitive chip is one thing. Recreating the whole ecosystem that makes it useful is another. So China does the next very best thing. If one chip is weaker, sure, use more of them, wire thousands of them together, feed them very cheap electricity, and you can get close to the same result. The catch though? More chips, more power, more space, not better, just more. Well bottom line, China can make enough chips, not the best ones, not yet. Which makes those 600,000 chips even stranger. If China still can't fully replace Nvidia, why turn them away? So here's what's going on. Those 600,000 chips are sitting in Hong Kong. Beijing is withholding approval to bring them into mainland. Well, Hong Kong doesn't even have enough data centers to switch them on. Giants like ByteDance and Tencent have reportedly gotten about 10,000 each. It trickled. So why? Well, there are three reasons. One, force the switch. If Chinese companies cannot buy Nvidia, they buy Huawei. Every chip left in Hong Kong is a customer handed to a Chinese company. And that's how Huawei went from 300,000 chips to 572,000 in one year. And number two, just no off switch. Think about it from Beijing's side. A chip supply Washington can turn off is one you cannot build an economy, a military or national security on. Better a slower chip you can control than a faster one someone can take away. And number three, credibility. Importing Nvidia by the hundreds of thousands would be an omission. You know, our own chips are not good enough. Well, that's not a message Beijing wants to send. For NVIDIA, the stakes are huge. Jensen himself put China's AI chip market at around $50 billion and warned it was slipping to Huawei. And now he's not there anymore to fight for it. And next Tuesday, November the 17th, NVIDIA reports its earnings, the first time in 33 years without its founder in charge. Watch that one closely. Here's what gets me about all this. Jensen built a company so fast, so flat, it never had to wait on anyone. And the whole time, it was waiting on one man. Beijing is learning to live without Nvidia. Nvidia is learning to live without Jensen. Well, the hardest thing to build isn't a chip, it's not needing anyone. So, How can NVIDIA really outlive its founder? Tell me in the comments. Also, we talked about the company that shorted NVIDIA six months ago in the last episode, and make sure you check it out if you haven't yet. I'll see you in the next one.