Who is Liang Wenfeng, the man who shakes the artificial intelligence industry

The last Chinese alternative in the world of artificial intelligence (AI), DeepSeekgenerated an earthquake in the technological sector worldwide. And this moved all the spotlights towards its founder, a 40 -year -old businessman named Liang Wenfeng and already qualified as “New face” of AI in China. Liang, born in 1985 in the southeast province of Canton, moved to the eastern region of Zhejiang to study electronics. There, at the University, he entered the field of artificial vision, a segment of the AI ​​that train computers to capture and Interpret information through image and video data.

And, together with a group of colleagues, he also began to explore the possibilities offered High-Flyer Quantwhich would end up becoming one of the largest coverage funds in China.

Managed with the help of the AI, the firm multiplied for more than 10 its assets in four years, from 1,000 million yuan (138 million dollars) in 2016 to more than 10,000 million yuan (1,379 million dollars) In 2019, the Hongkonese newspaper South China Morning Post points out.

In April 2023, High-Flyer Quant said he had invested “A large percentage“Of its benefits to build a “Leader Infrastructure” in AI and carry out large -scale investigations.

And, before the United States imposed restrictions on the sale of chips to China, the company led by Liang managed to get more than 10,000 NVIDIA H800 processors to train advanced artificial intelligence models.

Thus, in 2023 he launched Deepseek, which in recent weeks caused an important stir after the launch of its V3 model, whose development is claimed that it only lasted only two months and only cost less than 6 million dollars. After that, on January 20, launched the R1the one that is being downloaded these days in half the world.

In a recent interview with The China Academy, Liang was “very surprised” for Depseek’s success: “We never had intention to be disruptive, it happened by accident.”

“Our principle is neither to sell for losses or seeking excessive benefits. Our current price model gives us a modest margin of profits above our costs,” said Liang, who denied that its low prices -a 95 % cheaper than the model than the model OPENAI O1 are only a strategy to capture customers.

DEPEEEK offers application programming interface services (API) at a cost of 1 yuan ($ 0.14) per million input tokens and 16 yuan (2.2 dollars) per million departure tokens.

“We believe that API and API services should be affordable and accessible to all,” said the businessman, who maintains his commitment to the open source.

As expected, Liang not only attracted the attention of investors and entrepreneurs from the technological sector worldwide, but also of the government of his own country who recently recognized his projection by inviting him to a symposium together with the prime minister, Li Qiang.

Last week, in Beijing, a group of experts from sectors such as technology, education, science, culture, health or sports gathered with LI to offer opinions and suggestions, and Liang was the only emissary dedicated to AIdemonstrating that the authorities closely follow the Deepseek phenomenon.

And, before the technological rivalry with the USA, Beijing pointed out as a priority to AI, a market that could reach an assessment of about 5.6 billion yuan (772,313 million dollars) in the Asian giant around 2030.

In addition, according to estimates by the State Cicc Investment Bank, the Chinese IA sector could attract up to 1.4 billion dollars in capital over the next six years.

“China cannot always stay in the Ia. We often say that there is a one or two years gap between China and the American, but the true gap is between originality and imitation. If this does not change, China will always go In the rear, “says Liang.

An issue that worries in the sector is the apparent censorship in the Model of the AI ​​of China, which regulated in 2023 the services of AI to respect “the fundamental socialist values”, with what they are prohibited “generate content that threaten national security territorial unity and social stability. ”

Thus, to the question in the app on controversial matters in China as the Tiananmen massacre of 1989 or if Taiwan is part of China, Depseek refuses to comment on these issues and suggests the user to ask about problems of problems “Mathematics, coding and logic”.

Sam Altman, CEO of Openaipublished a thread in X, the social network previously known as Twitter, in which he states that the language model behind the AI ​​of Deepseek, called ‘Deepseek R1’ “It is an impressive model, especially taking into account what they can offer for the price.”

In addition, Altman has assured that “it is really stimulating to have a new competitor” and that Openai will not be left behind in the hard competition proposed by Deepseek: “Obviously we will offer much better models … we will launch some new ones”

Also, in an interview conducted by the American media CNBC in the frame take the technological advances from China seriously: “Seeing Deepseek’s new model is impressive, both for the effectiveness with which they have created an open source model that computes in time of inference and for their computer efficiency. We should take China’s advances very seriously. ”

The whole world looks closely at China for this new technology.