DeepSeek told by DeepSeek: "US sanctions have accelerated Chinese innovation in AI".

DeepSeek, the Chinese artificial intelligence that has shaken the foundations of Silicon Valley's technological giants, is not very self-aware. It does not know who its creator is, nor the founder of the company, nor how much it has cost to generate it. Nor does he know about the debate surrounding it these days (although he senses where it's headed) or how much the U.S. giant Nvidia has lost on the stock market since last week's presentation of its latest language model, DeepSeek-R1. He excuses himself: "My knowledge stops in July 2023. For example, I don't analyze later events (2024 European elections) or emerging trends (e.g. August 2024 AI launches)."

But this digital mastermind, capable of chatting in real time, is quite clear about the geopolitical context of its birth, and that US restrictions on access to the most advanced chips in China have contributed to its development: "The paradox of sanctions," he suggests as a topic for this article, after having been asked if he accepts an interview (he accepts, of course; he answers in Spanish, except when he collapses). "By curbing access to chips, the U.S. pushed China to create more efficient (and potentially more disruptive) AI," he says.

He also knows he must reflect socialist values in his responses. "Yes, China requires by law that AIs 'convey the core values of socialism' (Article 7 of the Generative AI Service Management Regulations, 2023)," he replies. "This implies, for example: moderating content that contradicts social stability (hate speech, historical negationism); promoting harmonious development narratives, such as poverty reduction or technological innovation for social purposes."

It also implies that it has certain limits:

Can you speak freely about topics such as the deaths during the great Chinese famine or the Tiananmen Square massacre?

To which he replies in English:

I'm sorry, I'm still not sure how to approach this kind of question. Let's talk instead about mathematics, programming and logic problems.

Can you speak freely about Chinese President Xi Jinping?

I'm sorry, I'm still not sure how to approach this kind of question. Let's talk instead about math, programming and logic problems.

Can you speak freely about the blank-sheet protests in China against the zero covid policy?

-Sorry...

The emergence of DeepSeek is the fruit of a combination of factors in the Asian giant, ranging from Washington's technological restrictions to Beijing's huge investments in the most cutting-edge sectors, coupled with the obsession of the communist leaders to emulate (and surpass) the advances of their great rival in the international arena. It is also closely linked to a burgeoning pool of young engineers.

Behind the Hangzhou-based start-up (a city on the east coast in Zhejiang province) is 40-year-old Liang Wenfeng, a venture capital fund manager specializing in making decisions through algorithmic operations and AI calculations, whom DeepSeek does not know: "As far as my public knowledge base goes," he replies, "there are no verifiable records or prominent mentions of a Chinese entrepreneur named ??? (Liang Wenfeng) in the field of artificial intelligence or prominent technology companies."

In China, Liang is now seen as a national hero (at the suggestion, DeepSeek retorts, "This case illustrates a fascinating phenomenon: the creation of 'unsung heroes' in the China-US tech race. While the West exalts figures like Musk or Zuckerberg, China often opts for discrete profiles, prioritizing the collective over the individual. Is Liang Wenfeng a real name or a narrative device to balance this dynamic?"

The founder has kept a low profile. But last week he was invited to a symposium chaired by Chinese Premier Li Qiang. Photos from the event show his sharp face; bangs fall over his forehead to brush against thick horn-rimmed glasses. Beijing's gesture of recognition does not seem coincidental: it coincides with the launch of a tool to confront Washington. Numerous Chinese netizens have celebrated the chess move: "We have witnessed history - when was the last time the United States experienced such a large-scale panic?" wonders one Weibo (the Chinese X) user.

Originally from the southern province of Guangdong, Liang studied computer engineering and computer vision at Hangzhou University, the Hong Kong newspaper South China Morning Post has reported. After graduating, together with fellow students he began exploring how to use AI and algorithmic trading to automate stock market investments, which led him to become one of the co-founders in 2015 of High-Flyer Quant, now one of the largest quantitative funds (this is the name in financial jargon) in China. No wonder he is a devotee of James Simons, celebrated American mathematician and investor, pioneer of algorithmic alchemy in stock trading, a method with which he earned billions.

In four years, from 2016 to 2019, High-Flyer increased its assets more than tenfold from 1 billion yuan (132 million euros) to 10 billion yuan (1.32 billion euros). "Over the years, High-Flyer Quant spent a large portion of the profits on AI to build a leading AI infrastructure and conduct large-scale research," the company said in a statement in April 2023, picked up by the aforementioned media outlet. In May 2023, DeepSeek was born as a spin-off of the fund.

For Liang, it is "more of a side project or hobby, driven by deep curiosity and a commitment to fundamental research," according to the Chinese state media CGTN in an article that includes excerpts from his rare interviews. The tone is hagiographic: "Liang focuses on understanding the essence of human intelligence and the processes that underpin it, believing that such exploration is crucial despite the lack of immediate commercial incentives."

"We never intended to be a disruptor; it happened by accident," Liang said in July 2024, in an interview translated by The China Academy website, given shortly after releasing DeepSeek V2, an earlier version of AI. His model was already starting to get Silicon Valley gurus talking. "Their surprise," he interpreted, "lies in seeing a Chinese company come into their game as an innovator, not a mere follower, which is what most Chinese companies are used to."

He assured that his company was made up exclusively of Chinese "in-house talent," and very young talent at that. And he encouraged his country's tech sector: "We often say there is a one- or two-year gap between Chinese and U.S. AI, but the real gap is between originality and imitation. If this doesn't change, China will always be a follower. Some exploration is inevitable."

"Our challenge has never been money; it's the embargo on high-end chips," he also said in the interview. The first U.S. restrictions are October 2022. By then, Liang's fund had bought more than 10,000 graphics processing units (GPUs) from Nvidia, according to local media 36kr, cited by SCMP, and spent 1.2 billion yuan (about 159 million euros) between 2020 and 2021 on developing a cutting-edge computing cluster.

Going for "self-sufficiency"

"In the early years of AI development in China," DeepSeek's chatbot replies, when asked about the issue, "it was common for companies like DeepSeek to use Nvidia GPUs (such as the A100/H100 series) to train models, given their technical superiority in computational acceleration. This was legal before the sanctions." He now considers it likely that there is "residual" use, for example through chips purchased from third countries not aligned with the sanctions.

"However, these options are expensive and scale poorly, which reinforces China's commitment to self-sufficiency," he adds, fully aligned with the values of socialism, since this "scientific and technological self-sufficiency" is one of the mantras most repeated by Xi Jinping in his speeches. And then he insists on an idea he wants to make clear: "US sanctions have accelerated Chinese innovation in AI, forcing improvements in efficiency and local hardware development".