Nvidia's Worst Nightmare Just Came True (and it's not Jensen Huang) — Transcript
Full transcript
- 0:00Nvidia's second worst nightmare came true last week when it lost its CEO, Jensen Huang,
- 0:06the man who co-founded it. But its actual worst nightmare is coming true at the same time. And
- 0:11it's got nothing to do with Jensen. Their worst nightmare is China. See, a year ago, Jensen Huang
- 0:18told the Financial Times, "China is going to win the AI race." Now, Nvidia quickly scrambled to
- 0:23retract that statement, but the damage was done. He basically said the quiet part out loud. Now,
- 0:29China's tech giants are spending about 140 billion on AI this year. That's double last year. So,
- 0:36out of every top 100 AI chips that Chinese data centers are buying right now, how many
- 0:41do you think are Nvidia's? Like 50%, 25%. Well, it used to be 95%, 95 out of every 100. Apparently,
- 0:50now it's actually zero. And that isn't because America's uh of America's restrictions. The truth
- 0:56is, China doesn't want Nvidia's chips anymore because they've got a better alternative. Now,
- 1:01by the end of this video, you'll know what their alternative is, how it affects Nvidia, and
- 1:05ultimately how it affects you and me. I'm Nick, and this is Finance Stack. Now, before we dive
- 1:11into the video, our research team puts a lot of work into every one of these videos. So, there's
- 1:16a data sheet linked below in the description with the source of every number and claim I
- 1:20make in this video. Check it out for yourself. So now back in September of 2025, China's internet
- 1:25regulator told ByteDance and Alibaba, two of the biggest tech companies in China,
- 1:29to stop buying Nvidia's special China chip. A China chip is a watered-down version of the chips
- 1:36that Nvidia makes. It makes them specifically for China. Now, America bans Nvidia's best chips from
- 1:41being sold there. So Nvidia slows one down until it's just under the limit and then it ships it
- 1:47to China. It's like a car maker that's banned from selling its sports cars in a country. So, it sells
- 1:51the same car there, but with a crappy motor in it. Well, America says, "This is the best chip
- 1:56that you guys can get from us." And China said, "No thanks. You keep that crap." Now, of course,
- 2:02China hatched a plan to procure some of the best chips through alternative means. But
- 2:08that's not what today's video is about. Today's video is about China building its own chips. So
- 2:142 months after China stopped buying watered-down chips from Nvidia, they went one step further and
- 2:18they banned foreign AI chips from new state-funded data centers. And if one of those data centers
- 2:24was less than 30% built, it had to rip out any foreign uh chips that had already put in which is
- 2:30insane because you know these chips cost a lot of bloody money. And then after all this,
- 2:35America changed its mind. In December of 2025, Trump said Nvidia could sell China a much better
- 2:40chip, the H200 chip, as long as America gets a 25% cut. So about 600,000 of them worth around $16
- 2:48billion at a Nvidia's own price are now sitting in Hong Kong. And Beijing just won't let them in.
- 2:54It does not want them. Why would anyone turn down chips that good? Well, put yourself in Beijing's
- 3:00shoes for a moment. Back in 2019, America cut off Huawei, the Chinese tech giant,
- 3:06and it said, uh, you can't sell Huawei phones anymore in the US, and Huawei can't use Android
- 3:12anymore. It was crazy. Then it banned Nvidia's chips one after the other. If you're China, would
- 3:18you build your entire AI future on a supplier that America can just switch off at any moment?
- 3:25Of course you wouldn't. They don't want the US to have that sort of leverage over them. So,
- 3:29at the start of this video, I asked for every 100 top AI chips that China's data centers buy,
- 3:34how many are Nvidia's? Well, the answer is zero. And it's because China has stopped bloody buying them.
- 3:41But this begs the question, can China actually make its own chips? So, let's go back to that
- 3:47sports car analogy we were using. If you can't buy the fast car, the fast sports car, you just build
- 3:52your own car, right? If you have the ability to. So, China's answer to this is that company
- 3:56we've been talking about, Huawei, and a smaller chipmaker called Cambricon. Now, Huawei's boss,
- 4:03Eric Xu, admitted that they cannot make chips as good as Nvidia's. So, what did they do? Well, they
- 4:09make up for it by combining chips. In other words, if your car's slower, you put four weaker engines in
- 4:15it, and maybe, just maybe, it beats a supercar. But that is a pretty big maybe. The data says
- 4:23that Huawei's best chips do about half as good as Nvidia's chips from 2022. Not even Nvidia's newest
- 4:30chips, the ones from four years ago. But also, you got to imagine if you are China, you don't
- 4:37want to let the US know that your latest chips, your cutting-edge ones, are as good as Nvidia.
- 4:41So maybe they're kind of getting closer. We don't know for sure. And also back in 2025, Jensen Huang,
- 4:47the Nvidia's boss, laughed off the idea that China was years behind. He said they are nanoseconds behind
- 4:53us. So maybe China has some top-secret chips that are close to Nvidia's. But we also know that
- 4:59Huawei wires hundreds of weaker chips together. Back in 2025, its big machine did almost twice the
- 5:05work of Nvidia's top one. But it burned about four times more electricity and cost nearly three times
- 5:10as much. So yeah, it does get there. It just cost a bloody fortune to run. But I don't think China's
- 5:17worried about the cost. First of all, China is fairly rich. I mean, so they can afford the
- 5:21cost. Also, a big part of the running cost of data centers is electricity. And China is even richer
- 5:27in electricity. It builds power stations like bloody nobody else. Last year, it added about 10
- 5:33times as much power as America did. It's building 41 nuclear reactors right now. America is building
- 5:39zero. See, this is where America and democracy can't keep up. That's that sounds crazy. I I'm
- 5:45a fan of democracy, to be clear. I'm happy that I live in a democracy. I don't want a nuclear power
- 5:51plant popping up next door to me. But if we step back at it and look at it purely as competitors,
- 5:56right? In America, building a power station or a nuclear power plant means years of paperwork,
- 6:01planning approval, environmental boards, lawsuits from angry neighbors. Then when it finally gets
- 6:06through, the other side wins the election and says no more nuclear power plants and it pulls
- 6:10back all the plants. Uh, and half a decade of work is thrown out the window. In China,
- 6:14the government just decides it wants to build a nuclear power plant and kicks you out of your
- 6:17house and it kind of builds it. I'm glad I don't live there, but it is more efficient when you're
- 6:22trying to win the AI race, right? And uh they do the same with solar and wind, of course. So, yeah,
- 6:27it it's not great for the people, but this is why they added 10 times more power than America last
- 6:31year. So, even with the crappier chips that you have to chain together that cost you four times
- 6:36as much to run, it's easy when you have 10 times as much power. But if China's chips are this bad,
- 6:42why would Nvidia be scared? Well, it matters because the AI coming out of those worse chips is
- 6:47nearly as good. The best American AI now beats the Chinese AI by only 3% on the main leaderboard. So,
- 6:55think of it as a as a league table for AI, and the gap at the top is 3%. That's close to nothing.
- 7:00Going back to the car analogy, which I guess we're going to stick with, and we're just going to keep
- 7:04using it. China shoved four engines into its car and it tops out at 223 miles per hour versus the
- 7:10American car that does 230 mph. We're not talking about a big difference here. Uh, and that's what
- 7:16Chinese AI labs are good at, getting more out of less. And we've already had a taste of what
- 7:21that does to Nvidia. Back in January of 2025, a Chinese startup called DeepSeek showed off its
- 7:28AI and said that it was built for a fraction of the cost. On that day, Nvidia's stock crashed and
- 7:33$593 billion was wiped off its valuation in a single day, which is the biggest one-day loss
- 7:39for any company in Wall Street's history. To be clear though, it didn't lose any sales or
- 7:44any actual money. It's just the valuation of the company, and it obviously recovered since then.
- 7:48But it still goes to show how easily confidence can be shaken by the competition that Chinese AI
- 7:53models bring to the AI war. This year, DeepSeek's latest model was partly trained on Huawei chips.
- 7:59another Chinese lab, uh, Zhipu, I'm probably saying that wrong, built, uh, an image model entirely on
- 8:05Huawei's chips. Now, what does that even mean? You know, they built an image model entirely on Huawei
- 8:12chips. Well, it all comes down to training. So, training is how an AI model learns. It's months
- 8:16of chips crunching through data. And it's uh the job that everyone said needed Nvidia, but China
- 8:22is proving every day that you might not need Nvidia anymore. that if you put enough engines
- 8:28in the car, you get close enough to matching the top speed and then the only difference becomes
- 8:33the driver's skill. Now, to be fair, this is not a complete wipeout yet. China's best AI still leans
- 8:39partly on Nvidia. So, just last week, a former OpenAI researcher found signs that DeepSeek's
- 8:44and Kimi's newest models, Kimi being another Chinese AI, were partly trained on Nvidia chips,
- 8:50uh chips that China is not allowed to buy anymore. what uh what they're doing is possibly uh going
- 8:57into a gray area by renting GPU compute from data centers in say Singapore or maybe maybe
- 9:03they're even smuggling in some chips. But you know that's just speculation allegedly,
- 9:08right? But what is clear is that China is starting to move away from Nvidia. They've left 600,000
- 9:14Nvidia chips sitting in Hong Kong in favor of filling their data centers with Huawei chips,
- 9:19chips that are theirs. And DeepSeek reportedly plans to build its next model mainly on Huawei
- 9:25chips. China is cutting the cord with Nvidia on purpose. And from China's perspective, it makes
- 9:31complete sense. You can't rely on a supplier whose government can flick off the switch at any moment.
- 9:36Talking about flicking switches, how about you flick on the like and subscribe button? That is
- 9:41so cringe. Uh but it really does help the channel out. So, so please hit the like and subscribe
- 9:45button. But back to what I was actually saying. So Nvidia has lost China, right? China is no longer
- 9:51wanting to buy from them. They're slowly cutting the cord. They're pulling back. But does Nvidia
- 9:55even care? Well, it it does care and it's mostly because of the money as it as it is always with
- 10:01mega companies run by billionaires. So in 2024, China was about 19% of Nvidia sales. In 2025,
- 10:07it dropped to 9%. And the top AI chips it sells to China's data centers are now less than 1% of
- 10:13its data center sales. When America blocked even its watered-down China chip in April 2025, Nvidia
- 10:19took a $4.5 billion hit on chips that couldn't sell. And look, I know this is getting confusing
- 10:24because earlier I said that China rejected the watered-down chips. Now I'm saying that America
- 10:28blocked them. This is because America has been about as decisive as my wife when it comes to
- 10:33ordering takeaway. She says to me, "Uh, I don't care what we get. You decide." So I decide. I say,
- 10:38"I want pizza." She says, "Nah, I don't want pizza." I say, "Okay, what about Greek?" She says,
- 10:42"No, I don't feel like Greek tonight." I say, "What about some Chinese food?" She says, "Uh,
- 10:46nah." And then she's like, "Maybe we should get pizza." That's how America treats shipping Nvidia
- 10:51chips to China. Here's the order of how all of this unfolded. In April of 2025, America blocks
- 10:57the China chip. You know, the watered-down chip. August, America changes its mind and lets Nvidia
- 11:01sell it again as long as America gets a 15% cut. China says, "Nah, thanks. Keep keep your
- 11:06crappy watered-down chip." So, America offers uh a much better chip, the H200. And China says, "Nah,
- 11:13you can keep that one, too." This whole ridiculous level of indecisiveness from the US has cost
- 11:19Nvidia billions of dollars, which I really don't care about if Nvidia loses money. It's just it's
- 11:24just crazy how indecisive a government can be. But this gives you some insight as to why they care.
- 11:30Uh, and Nvidia's chips were still finding their way into China anyway, because not every Nvidia
- 11:35chip stopped going to China. Uh, some have been getting in through the back door. So, one company,
- 11:40Megaspeed, which we've all heard a lot about recently, moved about $7.5 billion worth of Nvidia
- 11:46chips through Singapore and Malaysia, according to Fortune. And this is where Jensen Huang comes
- 11:52back into this story. The reports about why he quit point to an affair with a former Megaspeed
- 11:57executive. US intelligence reportedly suspects her of spying for China. China denies it, and
- 12:03nobody has been charged, so we don't really know. Since he left, American lawmakers are pushing for
- 12:09tougher chip rules. And America is reportedly about to stop Chinese firms renting Nvidia chips
- 12:14in data centers in Singapore. Remember that gray area we talked about that still allows China to
- 12:19uh train its AI? Well, it was doing it by uh renting GPU compute from data centers abroad. Now,
- 12:25they're looking at blocking that, too. Uh and they're looking at blocking it in Singapore,
- 12:28Malaysia, and Thailand. So, now even the back door is closing. And I know this is a lot of
- 12:33information, but basically first the legitimate routes of selling chips to China were killed and
- 12:38that hurt Nvidia. Now the gray area routes are collapsing and that's going to hurt them even
- 12:43more. But at least Nvidia is having no trouble selling chips in Western countries. So, you know,
- 12:48at least there's that. Chip sales have doubled in the last year. So, they still they're still
- 12:52doing incredibly well, right? Right. Well, about that. See, Western countries are exactly where
- 12:59China's AI is heading next. Which brings us to the actual nightmare. Finally, what is Nvidia's worst
- 13:06nightmare? It's cheap Chinese AI. Remember, it's China's four-engine car that tops out at 223 mph. Now,
- 13:14China's running it as a taxi. It gets you there nearly as fast for about the fair. And there's one
- 13:21more thing that China's labs do that America's don't. Most of their best AI is open source.
- 13:26That means anyone can download it for free and run it themselves. It's like a car m car maker
- 13:32publishing the full design of its car. So any garage in the world can just build one. Whereas
- 13:37America's top AI companies keep theirs locked up and they charge you to use it. DeepSeek charges
- 13:41about a fifth of what Anthropic does. Uh Anthropic is the American company behind Claude in case you
- 13:47don't know. Uh and it's uh it charges a fifth of what Claude charges for its main models. And
- 13:52remember the difference in performance is only 3%. Most people would happily pay 80% less for 97% of
- 13:59the performance. Of course, they would. I mean, who bloody wouldn't? And a great example of this
- 14:03is OpenRouter. This is a site where developers can use whatever AI model they need to complete
- 14:08tasks. The amount of work that developers are sending to Chinese models is increasing. It's
- 14:13now more than half of total work being sent to models. An American company called Lindy moved all
- 14:19of its AI work from Anthropic's Claude to DeepSeek. its boss said you could see the cost crash to the
- 14:25ground. Airbnb's boss, Brian Chesky, said that they rely on uh Alibaba's AI Qwen because it's
- 14:32fast and cheap. That work used to go to American AI companies, and American AI companies are the
- 14:38ones paying for Nvidia's chips. So, the people that were buying Nvidia's sports cars are now
- 14:43hopping in a cheap Chinese taxi. And cheap makes it sound bad. It's actually really bloody good,
- 14:48right? It's like a sports car taxi. Uh, and Nvidia says in its uh, own filing,
- 14:54"Being shut out of China helped our competitors build larger developer and customer ecosystems to
- 15:00challenge us worldwide." In plain English, losing China helped build its rivals. And we've seen
- 15:05China do this before. Huawei makes the tech that phone networks run on, like 5G, etc. It went out
- 15:12into the world about 30% cheaper than Ericsson and Nokia, the big European makers, with Chinese state
- 15:17banks lending its customers the money. By 2012, it was bigger than Ericsson. Ericsson, by the way,
- 15:22is Swedish. In 2020, Sweden banned Huawei from its 5G phone networks. The next year, China Mobile,
- 15:30China's biggest phone network, which uh the Chinese state owns, handed out its 5G
- 15:36contracts. Ericsson's share fell from about 11% to about 2%. You can guess why. Now, outside North
- 15:43America, Huawei makes up 41% of all the phone network tech that is sold worldwide, right? well,
- 15:49outside of America. Uh, and it's it's cheap. It's backed by the Chinese state, and it's good enough.
- 15:55So, the world bloody bought it. For now, China's AI chips have barely left China. It can't even
- 16:00make enough to satisfy its own demand. And America has warned companies that using Huawei's AI chips
- 16:05could land them in a lot of trouble. But see, it doesn't need to sell the chips. It just uses
- 16:10them to train AI that's almost as good and one day might even be better. And then it just sells
- 16:15the AI instead. So basically, China is just using the same playbook it used to dominate
- 16:20mobile phone network tech to dominate electric cars and uh next in line is AI. So what does this
- 16:26mean for your money? So this year Amazon, Google, Microsoft, Meta, and SpaceX are spending about
- 16:31$828 billion on AI data centers and chips. And Nvidia keeps about $75 of every $100
- 16:38it sells after the cost of making chips. Those companies buy Nvidia's chips so they can sell AI.
- 16:44The whole bit only works if nobody can do it much cheaper. Because if Chinese AI models uh do nearly
- 16:51the same job for about a fifth of the price, who keeps paying top dollar? And if fewer people pay
- 16:56for American AI, those companies need fewer of Nvidia's chips. And if you've got an index fund or
- 17:02a workplace pension, that bet is probably in it, right? You've probably got exposure. In the S&P
- 17:08500, the index of America's 500 biggest companies uh that a lot of pensions and funds follow,
- 17:13about $60 of every thousand is Nvidia, just Nvidia alone. And the seven big tech companies together
- 17:19hold about a third of the total. So remember DeepSeek Day, you know, when Nvidia lost $593 billion
- 17:26off of its valuation in a single day. That was just a taste. And remember what Jensen Huang said?
- 17:31I told you at the start, China is going to win the AI race. Nvidia walked it back. They retracted the
- 17:37statement, but after everything we've just seen, it's hard to say that he was wrong. And you know,
- 17:42let's be honest, he'd know more about China than most people. Now, look, the the chips that China
- 17:46uses are worse and they cost more to run, but it doesn't really matter because the AI that comes
- 17:52out of them is nearly as good and it cost about a fifth of the price. So, obviously, Nvidia
- 17:56is scared. If you found that useful, hit the like and subscribe button, and I'll see you in the next one.
About this transcript
This page contains the full transcript of Nvidia's Worst Nightmare Just Came True (and it's not Jensen Huang) by Finance Stack, generated from the public captions YouTube serves with the video. The transcript has 3,438 words across 195 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.