Liang Wenfeng, the man behind DeepSeek, a nerd with weird hairstyle

Liang Wenfeng, the creator of China's DeepSeek AI model, has become a topic of discussion online due to his appearance.

The startup founder began working on the project back in 2021. His acquaintances dismissed his work on the project at the time as a strange hobby that would yield zero result. Wenfeng was nicknamed a "nerd with a terrible haircut."  

"When we first met him, he was just a really dull guy with a terrible haircut, and he was telling us about building a cluster with 10,000 chips to train his models. We didn’t take him seriously," one of the businessman’s partners recalled. "He couldn't clearly articulate his vision – he just kept on repeating, 'I want to build this, and it will be a game-changer.'"  

Liang Wenfeng had been interested in artificial intelligence since his student years. The first company he founded was High-Flyer – a hedge fund that used AI for algorithmic trading.  

At the end of January, reports about DeepSeek’s launch triggered a sell-off in US tech stocks, with Nvidia shares plunging by 10%.

Who is Liang Wenfeng

Liang Wenfeng was born in 1985 in Zhangjiang, Guangdong Province. His parents were elementary school teachers of the Chinese language.  

During his school years, Liang developed an interest in mathematics, which later led him to study at a local university, where he earned both a bachelor's and a master's degree in information and communication engineering. His specialization was in computer vision.  

According to acquaintances, as a student, Liang was "reserved and unsociable" individual. Despite this, in 2008, amid the global financial crisis, he gathered a team of his classmates to collect financial market data. Together, they attempted to apply machine learning to quantitative trading – a strategy that relies on mathematical models and algorithms.  

Liang Wenfeng always believed that "AI would change the world," but at the time, he had no like-minded peers who could support and share his ideas. Many even mocked him, saying he "watched too much science fiction."  

Algorithmic Trading and GPU Purchases

After completing his studies, Liang experimented with various AI applications before ultimately focusing on finance. In 2015, he co-founded the hedge fund High-Flyer, which used artificial intelligence to make investment decisions and predict market trends. 

In 2019, the fund established High-Flyer AI to research AI-driven algorithmic trading and invested 200 million yuan (approximately 2.6 billion rubles as per the Central Bank's exchange rate on January 29, 2024) into developing the Yinghuo model. The company purchased 1,100 Nvidia GPUs for the purpose.

By 2021, High-Flyer had invested another 1 billion yuan into Yinghuo 2, acquiring 10,000 graphics cards.

At this point, High-Flyer was managing assets worth 100 billion yuan. The company relied entirely on AI for trading, operating similarly to Renaissance Technologies, which specializes in quantitative trading using mathematical and statistical methods.  

Liang explained that his GPU purchases were driven more by curiosity than by any secret business logic – he wanted to see how far AI could go. He noted that while OpenAI's ChatGPT revolutionized the industry for casual observers, those already involved in AI had realized, with the release of GPT-3 in 2020, that artificial intelligence would require massive computational power. Yet, "most people" still couldn’t understand why he was spending so much on GPUs.  

The Birth of DeepSeek

After amassing a significant fortune in algorithmic trading, Liang Wenfeng decided to return to his original vision of researching Artificial General Intelligence (AGI). In 2023, he announced the creation of DeepSeek – a company entirely separate from trading and independent of High-Flyer. 

When hiring employees, DeepSeek prioritizes "passion" and practical skills over experience. Most of its workforce consists of recent graduates or early-career professionals.  

In May 2024, the company released DeepSeek-V2, followed by DeepSeek-V3 in November. Then, in January 2025, it introduced DeepSeek-R1 a "thinking" AI model that solves mathematical and programming tasks better than or on par with OpenAI's o1 model.

Details

DeepSeek is a Chinese artificial intelligence company with nearly 200 employees that develops open-source large language models (LLMs). Based in Hangzhou, Zhejiang, it is owned and solely funded by Chinese hedge fund High-Flyer, whose co-founder, Liang Wenfeng, established the company in 2023 and serves as its CEO. The DeepSeek-R1 model provides responses comparable to other contemporary LLMs, such OpenAI's GPT-4o and o1, despite being trained at a significantly lower cost—stated at US$6 million compared to $100 million for OpenAI's GPT-4 in 2023—and requiring a tenth of the computing power of a comparable LLM. DeepSeek's A.I. models were developed amid United States sanctions on China for Nvidia chips, which were intended to restrict the country's ability to develop advanced A.I. systems.

Subscribe to Pravda.Ru Telegram channel, Facebook, RSS!

DeepSeek founder Liang Wenfeng in meeting with Chinese Premier
Author`s name Pavel Morozov
*
Editor Dmitry Sudakov
*

Report spelling error:

Error page url:

Text containing error:

Your comment: