Engineer, born in 1985, originally from Guangdong, Liang Wenfeng, the creator of DeepSeek At the beginning he wasn't taken too seriously: partly because of his visionary attitude, but also for its appearance with the cowlick.
In 2015, together with two of his fellow students, he founded the hedge fund High Flyer, Huanfang Quantitative, capable of becoming a giant in the IT and hi-tech sector in China in less than 10 years: it is precisely from that fund, which today manages around 8 billion dollars of assets, that in 2023 Wenfeng creates DeepSeek with the intent to invest all its resources to build its own cutting-edge AI models based on its desire for scientific research. “I can’t find a business reason,” the engineer told the Chinese 36 SEK. “Research, in essence, has a very low return on investment.” In 2021, Liang Wenfeng began purchasing thousands of Nvidia graphics processing units for his project.
“When we first met him, he was a guy very nerdy with a'terrible hairdo, who talked about building a computing system with 10.000 chips to train his models. We didn't take it seriously,” says one of his business partners. “He couldn’t clearly explain his vision, other than to say, ‘I want to build it and It will be revolutionary“We thought it was something that only giants like ByteDance and Alibaba could do,” adds the same person, quoted by Financial Times.
The Flapping of Butterfly Wings That Disrupts Tech Big
Four years later, Liang has become a source of pride for China. But since yesterday, many have cited edward lorenz and his famous metaphor of the flap of a butterfly's wings in one part of the world that can cause a tornado on the other side. And so it was: the Lilliputian with the force of his ideas made the American giants of artificial intelligence tremble, obtaining results similar to or even better than those of OpenAI, Google, Meta or Anthropic. All with a minimal fraction of the expense of the hyper-financed, made in the USA giants.
His DeepSeek caused a earthquake in silicon valley and another on Wall Street, where shares of Nvidia and other AI-related companies they fell headlong in a real Black Monday for the sector. A few days ago, Liang was the only AI leader selected to attend an entrepreneurs’ meeting with Chinese Premier Li Qiang. The entrepreneurs were told to “focus on making breakthroughs in key technologies.”
Family, studies, first difficulties
Liang is an outsider in the AI world, but he’s not one to have a makeshift resume. Son of an elementary school teacher, Liang has always shown a insatiable curiosity for technology. Graduated in artificial intelligence at theZhejiang University, developed expertise in computer vision and big data, and then began his career in finance. His High-Flyer fund experienced periods of sharp decline, linked to the volatility of financial markets, and Liang and his team used these crises to constantly refine their AI models. His team, in particular, acquired deep expertise in the use of Nvidia chips. But his true ambition was greater: to create an AI capable of competing with the best in the world.
Goal: Affordable Artificial Intelligence
DeepSeek had a clear goal: to develop an artificial intelligence that was not only competitive, but also accessible. “Our goal is not just to build a model,” he said in an interview. “We want redefining the rules of innovation”. Ambitious words, common to many entrepreneurs in the startup world. In his case, however, the vision came true in 2024, with the launch of DeepSeek V2, a model open-source which immediately attracted the attention of industry insiders, who are always very receptive to efficient and inexpensive models. “We didn’t expect such a strong reaction,” Liang admitted. “We simply followed our path, optimizing costs and setting fair prices. It was not our intention to start a price war, but we found ourselves in the middle of it.” cost theme is very relevant in the rise of DeepSeek: Emad Mostaque, founder of Stability AI, one of the most recognized companies in the field, compared the release of DeepSeek to a “Almost-iPhone which costs only $30 instead of $1.000”: the cost reduction is about 96% compared to other US AI technologies.
Even the decision to make DeepSeek-V2 open-source has sparked mixed reactions. Liang has always defended this choice. “Innovation is not protected by secrecy, but by speed and adaptability,” he said. “Even OpenAI cannot prevent others from catching up.” Then the arrival of DeepSeek R1, a model with advanced reasoning abilities, like OpenAI’s o1, completed the picture and put its chatbot at the center of the world stage, despite accusations of censorship on topics unwelcome to Beijing and of a less than rigorous treatment of privacy.
Liang Wants to Redefine the Rules of Innovation
What sets Liang apart from other entrepreneurs in the industry is his approach to innovation. At the heart of his words is a structural change in the way China participates in the evolution of AI. “We have always seen AI as something that comes from the West“, he said. “But why should we be only users and not protagonists?” he said. DeepSeek is not a company focused on immediate commercial applications, but rather (at least at the moment) on pure research. “Most Chinese companies copy and adapt, we want to create. For too long, innovation has been seen as a luxury. But today China has the resources to invest in basic research".
Recruited PhD students and graduates in the last two years
This approach is also reflected in his philosophy of reclutamento. DeepSeek has built a team entirely composed of Chinese talents, challenging the preconception that the best AI experts are all abroad. “If you want immediate results, hire those with experience,” he said. “But if you want to change the world, you look for curiosity and passion.” In his team, as he recalls Wired, did not choose expert engineers, but only PhD students from the country's top universities, from Peking University to Tsinghua University, eager to demonstrate their capabilities. "Our top technical positions are mostly filled by people who have graduates in the last two years“, Wenfeng explained to the Beijing press in 2023. “Most people, when they are young, can devote themselves completely to a goal without utilitarian considerations. What do I ask of potential employees? Solve the world’s most difficult problems.”
