"Stop Fare Evasion Now," Criticized by China... Looking Back at DeepSeek Founder’s Past Remarks

After Majoring in Electronic Engineering at University, Founded an Investment Firm
Pioneered Quantitative Investing Using Mathematics and Artificial Intelligence

"Artificial Intelligence (AI) should be accessible and usable by everyone."

Recent attention has been drawn to previous interview remarks by Liang Wenfeng (梁文鋒, 40), founder of DeepSeek, a Chinese AI startup that shocked the world by releasing a high-performance AI model at an overwhelmingly low cost.

According to CGTN, the English-language channel under China Central Television (CCTV), on the 30th, Liang Wenfeng is known for founding Highflyer in 2015 and growing it into one of China’s largest quant hedge funds within a few years. However, until DeepSeek’s latest AI model, launched in 2023, caused a sensation, he was not widely recognized by the public. Consequently, he has only given two media interviews, both with Anyong (暗涌), a media outlet under Chinese tech media 36kr, in 2023 and 2024.

Liang Yuanfeng, founder of DeepSeek (front row, right). Photo by CGTN, Yonhap News

In a July interview last year, Liang stated, "DeepSeek’s purpose is not to exploit technology innovation for excessive profits; the benefits of AI should be enjoyed by everyone." This interview took place after DeepSeek’s low-price policy for DeepSeek-V2, released in May last year, triggered a price-cutting competition. Regarding the low-price policy, Liang explained, "It was not aimed at acquiring users. Our principle is not to incur losses nor to make excessive profits. The current price level yields a slight profit above cost."

He added, "The reason we lowered prices is partly because costs decreased first during the exploration of next-generation model structures, and partly because we believe that whether it is API (Application Programming Interface) or AI, benefits should be universal and accessible to everyone."

Liang also emphasized, "The most important thing now is to participate in the global wave of innovation." He continued, "In the past, Chinese companies were accustomed to others innovating technology, and we made money by applying it, but that is not a natural right. Our starting point in this wave is not to seize opportunities to make money but to advance to the forefront of technology and promote the development of the entire ecosystem."

"China Should Be a Contributor, Not a Free Rider"

He also stated that China must move away from the perception that the U.S. leads in technological innovation while China is a latecomer that merely applies it. Liang pointed out, "As the economy develops, China should gradually become a contributor rather than remain a free rider." Regarding technological competition with the U.S., he said, "Chinese AI cannot remain forever in a trailing position. Although there is a 1-2 year gap between Chinese AI and the U.S., the actual difference is one of creation versus imitation. If this does not change, China will inevitably remain a follower forever."

As for why his company insists on open-sourcing its AI models, he said, "Nothing is lost by open-sourcing and publishing papers. Being followed by engineers brings great satisfaction. Open source is more of a cultural act than a commercial one, an additional honor. If the company does this, it will also have cultural appeal." When asked when DeepSeek’s ultimate goal of Artificial General Intelligence (AGI) might be realized, he replied, "Maybe in 2 years, 5 years, or 10 years, but it will happen within our lifetime."

Born in 1985 in Zhanjiang City, Guangdong Province, Liang Wenfeng ranked first in the university entrance exam at Wuchuan No. 1 High School in Zhanjiang in 2002 and was admitted to Zhejiang University, majoring in Electronic Information Engineering. Liang developed an interest in finance during university and, in 2013, co-founded Jacobi Investment Management Co., Ltd. in Hangzhou with alumnus Xu Jin. Two years later, he founded Highflyer to engage in quantitative investment using mathematics and AI. Subsequently, he independently built an AI training platform capable of deep learning. After researching and developing the training platform named "Firefly No. 1" in 2019, he invested 1 billion yuan (approximately 198.9 billion KRW) in "Firefly No. 2," developed in 2021, which is equipped with 10,000 NVIDIA A100 graphics cards.