Who is Liang Wenfeng, the founder of AI company DeepSeek?

Liang Wenfeng is a figure who has sparked a revolution in the field of artificial intelligence (AI), yet he maintains a low profile and rarely appears in public. This 40-year-old entrepreneur grew up in Guangdong Province in southern China, which pioneered the implementation of a market economy in China during the 1980s and 1990s.

Today, Mr. Liang is leading China in its participation in the artificial intelligence revolution, striving to keep pace with the United States in this field. DeepSeek's release of its new AI chatbot on Tuesday, which it claims is lower in cost and consumes fewer resources, shocked the American market. The market capitalization of several companies, including U.S. chipmaker Nvidia, consequently plummeted, losing nearly a trillion dollars in value.

Although Liang Wenfeng has maintained a low profile, he was one of nine invited guests who spoke at a closed-door seminar hosted by Chinese Premier Li Qiang on January 20th this year. Prior to this, he had given two interviews to the Chinese media outlet LatePost last year, but otherwise, he has rarely been seen in the public eye.

Despite his limited media exposure and public statements over the years, Mr. Liang is unambiguous when expressing his views on China's role in the artificial intelligence race. In an interview with LatePost last July, he stated, "China's artificial intelligence cannot always be in the position of a follower. We often say that there is a one or two-year gap between China's AI and the US, but the real gap lies in the difference between originality and imitation... If this situation does not change, China will always just be a follower—so some exploration is inevitable."

Liang Wenfeng attended the prestigious Zhejiang University at the age of 17, majoring in Electronic and Communication Engineering, and later obtained a Master's degree in Information and Communication Engineering, graduating in 2010. In 2015, he co-founded the quantitative hedge fund Gao鹄 Capital, which uses complex mathematical algorithms for trading rather than relying on manual analysis. By the end of 2021, the fund's total investment portfolio exceeded 100 billion yuan (approximately $22 billion USD).

However, in April 2023, Gao鹄 Capital announced on its WeChat public account that it would expand its business scope beyond the investment industry and focus its resources on "exploring the essence of general artificial intelligence." A month later, DeepSeek was established. The company released its first AI large language model later that year. DeepSeek's chatbot quickly gained popularity after being launched on the Apple and Google app stores this year.

Under Liang Wenfeng's leadership, DeepSeek has intentionally avoided application development, instead focusing its research talent and resources on creating a model that can rival or even surpass OpenAI. His approach is unusual in China's tech industry, which is accustomed to borrowing innovations from abroad, from smartphone apps to electric vehicles, and scaling them up rapidly, often faster than in the countries where the innovations were first created. This time, China is leading the trend, not following it, which is exactly the change Liang Wenfeng has always wanted to see.

DeepSeek's chatbot became more popular after launching on the Apple and Google app stores this year. But what truly sparked panic was a follow-up research paper released last week—on the same day as U.S. President Donald Trump's inauguration. This paper introduced DeepSeek's AI model R1, which demonstrated advanced "reasoning" capabilities, such as being able to rethink methods for solving mathematical problems, and at a significantly lower cost than similar models sold by OpenAI. This triggered a global tech stock sell-off.

Nvidia is the biggest victim of DeepSeek's dramatic disruption of the artificial intelligence market so far, yet Nvidia has welcomed the sudden emergence of this Chinese AI company. Nvidia's stock price fell 17% on Monday night after DeepSeek released its chatbot, losing $590 billion (about A$940 billion) in market value. Nevertheless, Nvidia described DeepSeek's innovation as "remarkable AI progress." Nvidia stated in a statement that more of its chips will be needed in the future to meet the demand for DeepSeek's services.

"DeepSeek's work demonstrates how to use the technology to create new models, leveraging widely available models and computing power that is fully compliant with export controls," Nvidia said. Unlike its U.S. competitor OpenAI, DeepSeek decided to open-source all of its models. In open-source models, the underlying code is public, and any developer can freely use and modify it. "Even if OpenAI is closed-source, it cannot stop others from catching up," said Liang Wenfeng. "Open source is like a cultural practice, not a business practice... A company that does this will have soft power."

Liang Wenfeng's rare public comments reveal that he believes the Chinese tech industry has reached a crossroads. It lacks confidence but does not lack the funding needed to make breakthroughs in basic research. "In the past 30 years, (the Chinese tech industry) has only emphasized making money while neglecting innovation," he said in July. "Innovation is not only driven by business, it also requires curiosity and the desire to create."

Xu Chang, an associate professor at the University of Sydney specializing in machine learning and computer vision, stated that DeepSeek is not profit-driven due to the success of its parent company, Gao鹄 Capital. "What's impressive is its cost-effectiveness... I think this is a success of extreme engineering and algorithmic innovation," Dr. Xu said. He stated that restrictions on the supply of U.S. chips in China have forced companies like DeepSeek to innovate in engineering and algorithms.

But Dr. Xu believes that DeepSeek's decision to go open-source is the biggest shift in the artificial intelligence industry. He compared this to the smartphone market, where there is competition between closed systems and open-source platforms like Android. "So I think this is actually the beginning of the Android era for large AI models, and DeepSeek has marked this," he said. "Therefore, its open-source framework can be widely accessed by the public and is very friendly to research committees, so we can freely access research to build models."

When asked, "Who is Liang Wenfeng?" DeepSeek's first response was to name another Chinese entrepreneur with the same English spelling. When asked, "Where is Liang Wenfeng from and where did he go to university?" it stated that as of October 2023—the latest knowledge cutoff date for DeepSeek's R1 AI model—"there is no public information about Liang Wenfeng's background, including his place of origin or educational experience." "If you are referring to the founder of DeepSeek, details about his personal life or academic background have not been publicly disclosed. For more information about DeepSeek, you can visit its official website," it said.