How A Chinese Villager Shook Silicon Valley [DeepSeek Founder]

• Hi, welcome to another episode of Cold Fusion.
• This is Liang Fen Wang, a 40-year-old Chinese man from Meiil Ling village in the southern province of Guangdong.
• Up until a few weeks ago, he was essentially unknown.
• But now, the whole world is scrambling to gather more information about him.
• The reason? Deepseek, an AI system that he built with some of his college buddies, caused a 1 trillion dollar stock meltdown on Wall Street.
• The entire tech world has taken notice.

Hi, welcome to another episode of Cold Fusion. This is Liang Fen Wang, a 40-year-old Chinese man from Meiil Ling village in the southern province of Guangdong. Up until a few weeks ago, he was essentially unknown. But now, the whole world is scrambling to gather more information about him.

The reason? Deepseek, an AI system that he built with some of his college buddies, caused a 1 trillion dollar stock meltdown on Wall Street. And the entire tech world has taken notice. And how could they not? Deepseek as a product has rivaled the best of the United States. In fact, when the dust settled, everyone was left scratching their heads.

How had this guy seemingly come from nowhere to challenge $100 billion American giants for cheaper? And his releasing it for free, it changed the perceived power balance of the global AI race. When Liang returned to his hometown for Chinese New Year, he received a hero's welcome.

David had fought Goliath. China had joined the Americans in the AI race, and Liang Fen Wang was the guy who did it. His childhood friend Leon tells the Financial Times, "We all grew up in the same village. We're very proud of him."

So I'm sure you're wondering, what is the story of Liang Wenfng and his company Deepseek? In part two of this series, we will find out. A few weeks after its launch, the DeepSeq r1 effect continues. NVidia video stock is slowly climbing, but it's still down. Wall Street, Silicon Valley, and much of the tech community are still digesting what this all means.

America was humiliated. It seems their restrictions on China had backfired big time. The rest of the Western world had its doubts. But if anyone was praying for Deepseek to be all a fluke, it was Silicon Valley. There were frantic attempts to make sense of who Liang was: a mastermind playing the system, a lone wolf, or a part of a larger conspiracy.

Liang Finn Wang has been called a noble entrepreneur, a nationalistic hero, a government pawn, a mad genius, the anti-Silicon Valley CEO, and everything in between. Why is it so hard to get information on what he calls himself? Well, it's because he's remained silent.

As those who know him say, this behavior is part of his constitution. As a rival fund manager puts it, "For those who know him well, he can talk a lot. When it's a topic of interest, he's not being rude. When he suddenly quiet, which often happens, he's thinking. If he doesn't want to talk to you, he'll let you know."

Liang Wen Feng was born in 1985 in a small village in the Guangdong province of China. His parents were both primary school Teachers. He grew up in a time where China was adopting market capitalism. The stable state-assigned jobs of the past were gone.

Liang was surrounded by people who were trying to start their own businesses and parents who pushed their kids to excel in school. And excel he did. One of his high school teachers reflects on Liang, saying that he stood out in a class of 50 as, quote, "a top student, especially in maths."

In 2002, at the age of 17, he attended Xijiang University, one of China's top universities and home to tech giants like Alibaba. Liang majored in electronic information engineering. All of this at this point was pretty mundane.

But then something happened that would change his life and the course of history. In 2007, the financial crisis hit, and Liang, still in university, was very curious about it. While others were freaking out, Liang ran towards the crisis. He pulled together a group of classmates to investigate an ambitious question: Could machine learning predict market trends better than human traders?

They got to work, and sure enough, Liang and his team found patents that other traders had missed. From then on, they began experimenting with financial technology, and it made him a lot of money—millions. But Liang, ever the academic, did go back to university and there received his Masters of Engineering in Information and Communication Engineering.

His dissertation was on how to track moving objects using a cheap camera. So as it turns out, he's always had a long-time interest in finding cheaper ways to make powerful tools. After graduating in 2010 and using his skills in math and coding, he ventured further into automated stock trading, first as a freelancer and then with his college mates.

But he wanted more. So Liang branched out to start his very own hedge fund at the young age of 30. This was in 2015. There he would do something remarkable. He used deep learning models in live trading.

You see, using AI to predict market trends and make investment decisions was a novel idea at the time, as deep neural nets only came onto the scene about three years earlier with AlexNet. That was just for image recognition. So having the foresight to use deep learning and trading in China of all places was very notable.

His hedge fund grew to become one of China's "big four quantitative analysis firms." Another manager of a hedge fund in China who has known Liang for years recalls, "He stands out from the crowd because he's a true engineer. He thinks and acts like one. While we're managing people and funds, he is coding every day."

Liang's hedge fund, now called High Flyer, used AI to their advantage. They eliminated the need for human portfolio managers, and High Flyer went from strength to strength, managing billions in assets. So at this point, you could imagine everyone's shock when its co-founder, Liang Wenfang, walked away from it all.

People knew that Liang was interested in AI, but only as a hobby. So what came next was a surprise to all. In 2023, Liang announced that his hedge fund High Flyer would instead focus on artificial general intelligence or AGI—that is, human-level intelligence aimed to mimic and outperform humans on a range of tasks, not just financial trading.

It was crazy talk because AGI, if it exists at all, likely means a foundational AI model. Foundational AI models were only something that tech giants like Google, OpenAI, Meta, Baidu, and Alibaba dared to touch, specifically focused on AI.

Are there spaces where you see a startup from India building? You can build on the models, be it ChatGPT and many others. But if you want to build foundational models, how should we think about that?

Where is it that a team from India, three super-smart engineers, with not $100 million, but let's say $10 million, could actually build something truly substantial? Look, the way this works is we're going to tell you it's totally hopeless to compete with us on training foundation models. You shouldn't try, and it's your job to try anyway. And I believe both of those things.

AGI was a whole other beast on top of that, way beyond foundational models. And no one had been successful at that yet. In addition, American companies were still light-years ahead of Liang. But he wasn't scared. In fact, he recognized the role of hardware in AI development. So he started buying up thousands of Nvidia A100 GPU cards.

People thought he was reckless, but he continued and he'd create a special research lab for his AI hobby. When former President Biden limited China's access to American GPUs, it should have been game over for Liang. But it wasn't. True to form, Liang once again saw this setback as an opportunity.

He reworked the training process to reduce the reliance on GPUs. And we all know the story from here. Liang charged ahead, spinning off his research lab into a separate company that would become Deepseek. His purpose was to be a research body, no external financing, and no plans for commercialization.

In the last episode, we talked about how DeepSeq was accused of stealing IP from OpenAI, and the Internet was quick to point out the irony after OpenAI scraped the whole Internet for their model. But leaving that aside, Liang was inspired by America in some less expected ways. He describes America as China's teacher, a rare keynote on quantitative investing.

But from then until now, Liang seems to reject the American way of doing things. As American CEOs became louder and more public, Liang kept a low profile. Where the structure of Silicon Valley companies is hierarchical, Deepseek is from the bottom up.

Instead of poaching top engineers from around the world, Liang would hire bookworms. These guys often wouldn't have computer science degrees or even any work experience. The reason? Liang believed that experts were too stuck in their ways.

So why not get people that are just passionate? In the United States, exclusive access to resources plus billion-dollar investments equal innovation. But Liang proved that innovation could happen on the same scale with far fewer resources.

And while OpenAI has shifted to closed source, claiming that the move will stop AI from falling into the wrong hands, Deepseek is open source, hoping that it will get into the right hands that will advance the field of AI.

To potentially AI in China, open source is cultural and part of a growing trend. As Liang says, "For technologists, being followed is rewarding. Open source is cultural, not just commercial. Giving back is an honor and it attracts talent."

But why make it so open source? Couldn't US companies just take the improvements and make it for their own AI? Well, if he's to be believed, Liang says that it's not about the profit, and he's hell-bent on innovation—and most importantly, Chinese innovation.

For years, Chinese companies have been accustomed to leveraging technological innovations developed somewhere else and monetizing them through applications. But this isn't sustainable. From investors to big companies, everyone in China often says there's a one- to two-year gap between AI and China and the US. But the real gap is the difference between imitation and originality.

For many years, Chinese companies have mostly been used to letting others innovate and monetizing their inventions. But that doesn't have to be the case. "Our point of departure in this wave is not to make a quick buck, but to get to the forefront of the technology so we can drive the development of the entire ecosystem," end quote.

Consider this in the context of everything going on between China and the US, and it makes sense why people think Liang's moves are not about slowing down the US but speeding everyone up.

If investors are more reluctant to invest in US AI companies, it levels the playing field for all. But throughout all this fuss, Liang has remained silent. He's the center point of 21st-century tech war tensions between the US and China.

And there hasn't been a peep from the guy. But Liang couldn't stay silent when the Chinese government came knocking. He's been snatched up and appointed as the one and only AI leader of China.

And his first task? A joint publicized meeting with a select group of Chinese entrepreneurs and Li Kuang, who only answers to Xi Jinping himself. Liang has now been tasked with leading China into an era of technological breakthroughs.

Liang has stated that he wants China to do more than just scale existing applications of Western tech. "We believe that with economic development, China must gradually transition from being a beneficiary to a contributor rather than continuing to ride on the coattails of others. Over the past 30 years of the IT revolution, we barely participated in core tech innovation," end quote.

Liang appears to believe that chasing profit kills innovation. So the question is, will he stick to his word or will he be pulled into temptation like OpenAI? And lastly, although completely expected, do those meetings with Chinese officials mean military or other advanced AI capabilities for the CCP?

After all, not long after the release of Deepseek, American companies such as Google and OpenAI are changing their tolerance to AI military and defense applications. They're letting in the US government as a customer of Bespoke Technologies. So this is all moving very fast.

But one thing is for sure, it's definitely going to be interesting. So let's all sit back and watch the show unfold. As for Liang, maybe he'll speak to the press about Deepseek one day, but let's not hold our breath.

So that is the story of how a villager from China shocked the world and upset Silicon Valley. So what are your thoughts on this? Feel free to let me know in the comment section below.

These days everyone is talking about fake news. Getting informed is harder than it's ever been before. But who's doing something about it? Well, today's sponsor Ground News is doing just that.

Ground News is a website and app developed by a former NASA engineer. Her mission was to give readers an easy, data-driven, and objective way to read the news using the story of Deepseek being banned in Australia for government employees, for example. Their bias distribution chart shows me the political leanings of those news outlets, and I can even get a summary of how the issue is being framed.

The left often cites security risks, and they frame it as being protective. The center talks about the government's reasoning for the ban being due to other risk factors, not just as Chinese origins. It's more calculated. The right says that Australia is in line with other countries and it's more of a focus on the broader picture.

Scrolling down, I can compare every single article on the topic with convenient tags showing me context about the source, like how factual it is and who owns it. Ground News' Blind Spot feed is also great. It shows you stories that are underreported by one side of the political aisle.

If you're on the left, you may have missed the story about how the US Government is permanently classifying fentanyl as a Class I substance. Ground News is a fantastic tool for getting international views, sifting through misinformation, and identifying media biases. They provide all the tools you need to be a critical thinker.

Today, Ground News is offering 40% off, and this comes with a feature called My News Bias, so you can even check how your news viewing habits change over time. Go to Ground News/ColdFusion to get started.

Anyway, that's about it from me. My name is Deaggogo, and you've been watching Cold Fusion. If you did like that episode, feel free to subscribe. There's plenty of other interesting stuff on here about science, technology, and business.

Alright, cheers guys. Have a good one. You are watching Cold Fusion TV.