Skip to main content

Meet Liang Wenfeng – the man who left a wildly successful hedge fund, turned down the fast track to billionaire retirement, and instead decided to take on the likes of OpenAI, Google, and Meta with a small-but-mighty Chinese AI startup called DeepSeek. Why? Because he believes it’s high time China stopped being the world’s tech tag-along and started calling the shots.

From Tracking Targets to Tracking AGI

Liang isn’t your typical tech bro. Sure, he’s got the degrees from Zhejiang University and an early obsession with tracking algorithms (that didn’t involve ex-girlfriends, we checked). But his rise to AI fame really kicked off when he co-founded High-Flyer, a hedge fund so smart it made $8 billion by teaching machines to trade stocks better than your average Wall Street wolf.

But Wenfeng wasn’t content with just buying low and selling high. He started buying big – investing in AI supercomputers when most hedge funds were still obsessed with yachts and cufflinks. Fast-forward to 2023: he drops the finance world like a bad stock and launches DeepSeek, declaring open war on AI mediocrity.

Originality > Copycat Culture

Liang’s philosophy? China needs to stop remixing Silicon Valley’s leftovers and start producing the next wave of original tech. For years, he says, the country has excelled at monetising other people’s breakthroughs. But that strategy has a shelf life, and it’s expiring fast.

His rallying cry? “It’s time to lead.”

DeepSeek is built on this mindset, with an R&D lab that looks more like a think tank than a tech company – one where originality, clever engineering, and a hint of rebellion are key ingredients.

Open-Source Evangelism (With Bite)

While Western AI giants are locking up their models behind NDAs and paywalls, Liang went the opposite route. DeepSeek’s R1 model? Open-source. Why? Because sharing is caring, and because he believes open-source isn’t just smart business – it’s a cultural weapon.

“Even if OpenAI is closed-source, it can’t stop others from catching up,” he said, like a man who definitely plans on catching up… and possibly overtaking with a smug grin.

Building Better AI (Without Breaking the Bank)

Forget billion-dollar budgets and warehouses full of GPUs. DeepSeek’s magic trick is doing more with less. Their models perform on par with GPT-4 – but without needing a server farm the size of a small nation. They’ve shown that clever engineering and architectural innovation (hello, Multi-Head Latent Attention!) can outsmart brute-force computing. David, meet Goliath’s GPU bill.

Talent Scouting: Poets > Programmers?

One of Liang’s spiciest moves? Hiring humanities grads to help build AGI. No, not to fetch coffee – to code, create, and innovate.

He says experience is overrated. What matters is curiosity, creativity, and passion. DeepSeek’s team includes literature lovers, history buffs, and Gen Z hackers raised on memes and Marx. Why? Because the future of AI needs more heart, more nuance, and fewer lines of code that think empathy is a syntax error.

AGI Dreams, with a Side of Maths and Multimodality

Liang’s endgame is Artificial General Intelligence. And he’s not messing around. His strategic focus is on three areas:

Math and code as the perfect sandbox for logical learning.

Multimodality, so AI can understand the world like we do (with eyes, ears, and sarcasm).

Language, because if a machine can banter in Mandarin and quote Shakespeare, we’re halfway to magic.

He also sees a future where specialised AI firms build the brains (like DeepSeek), and others build the apps. It’s like Apple and the App Store, but for sentient algorithms.

Changing the Game, and Maybe the Geopolitics

DeepSeek isn’t just building models. It’s shaking up the AI world. Its rise triggered a minor panic on Wall Street, a lot of side-eyes from Silicon Valley, and a rethink of US export controls. If they can do this with $6 million and fewer GPUs than Meta uses for lunch, what’s next?

Some are calling it the “AI Sputnik moment.” Liang just calls it a start.

Final Word: Rebels Build the Future

Liang Wenfeng’s AI philosophy is a three-part power combo:

• Build with originality.

• Share with openness.

• Think with unfiltered curiosity.

Whether he’ll beat the West at its own game remains to be seen. But one thing’s for sure – the AI race just got a lot more interesting.

And maybe, just maybe, a little more poetic.