When ChatGPT dominated the horizon of artificial intelligence, the question arose: would it become a headache for China, America’s biggest competitor in the tech world?
In just two years, a new AI model from China shifted the focus of that question: Can the U.S. stop Chinese innovations?
For a moment, it seemed like Beijing had faltered in the face of ChatGPT because the app was not available in China. Users mocked Baidu’s chatbot “Ernie,” a major search engine in China. Later, versions from tech firms like Tencent and ByteDance came out but were dismissed as inferior to ChatGPT in terms of performance. Washington felt confident it was ahead and wanted to maintain that lead. Then, the Biden administration imposed bans on exporting chips and technology to China.
This is why the creation of DeepMind’s new model left Silicon Valley and the world in shock. The company claimed that its powerful models were far cheaper than the AI developed by American companies, which had spent billions of dollars.
The question now is how a relatively unknown company, whose founder is being hailed as an “AI hero” on Chinese social media, managed to achieve this. When the U.S. blocked chip companies like Nvidia from selling advanced technology to China, it was a significant blow for China.
These chips are essential for powerful AI models that can perform various human tasks, from answering basic questions to solving complex mathematical problems.
In interviews, DeepMind’s founder, Liang Wenfeng, described the chip ban as a special challenge. Even before the ban, DeepMind had required a large number of Nvidia’s A100 chips, roughly between 10,000 to 50,000 chips. Estimates suggest that around 16,000 specialized chips are used for AI models in Western countries. However, DeepMind claims to have trained its AI model using just 2,000 chips, along with thousands of lower-grade chips, making their products more affordable.
Big names in technology, like Elon Musk, have questioned this claim, noting that with the imposed restrictions, this company cannot use such a large number of advanced chips.
Experts say the U.S. sanctions have created both challenges and opportunities for China’s AI industry.
Marina Zhang, an associate professor at the University of Sydney, notes that the ban pushed companies like DeepMind to innovate and find ways to do more with less. “These restrictions pose challenges but also stimulate creativity, which aligns with China’s broader policy goals of technological independence,” she said.
From electric car batteries and solar panels to AI, China has made substantial investments in the tech world. Making China a technological superpower has been a long-standing ambition of President Xi Jinping, and the sanctions from Washington have presented a challenge that Beijing has confronted head-on.
Gregory C. Allen, an AI expert at the Center for Strategic and International Studies, pointed out that China deliberately timed the release of DeepMind’s new model on January 20, the day President Donald Trump took office. According to Gregory, the way the message was delivered was exactly how the Chinese government wanted it to be perceived.
In recent years, the Chinese government has promoted AI by fostering talent, offering scholarships and grants for research, and encouraging partnerships between universities and the industry.
Professor Zhang notes that China’s National Engineering Laboratory and other state-supported initiatives have helped train thousands of AI experts. China also had a wealth of engineers in the industry. For example, DeepMind’s team, according to Chinese media, consists of 140 people, most of whom are described as “local talent” educated at China’s top universities.
Professor Zhang believes that Western observers have missed the emergence of a new generation of entrepreneurs in China, who prioritize fundamental research and long-term technical development over immediate profits.
China’s top universities are rapidly increasing the number of AI experts, and many of their managers are under 35. This new generation is heavily influenced by the drive for innovation and self-reliance during China’s rapid technological rise.
Liang Wenfeng, founder of DeepMind, is an example of this trend. Liang, 40, studied AI at Zhejiang University. According to a piece published by the tech outlet “36Kr,” those who know Liang say he is more of an AI enthusiast than a typical boss.
Chinese media describes Liang as a “technically principled” figure. He insists on keeping DeepMind as an open-source platform, which experts believe has allowed new startups to gather resources and progress quickly.
Unlike China’s large tech firms, DeepMind prioritizes research, which, according to experts and people working at the company, has given them the freedom to experiment further.
Liang Wenfeng stated in an interview with “36Kr” that while the top 50 people in the industry may not be in China, the country can still train such talent.
However, experts are puzzled as to how far DeepMind can go. Professor Zhang mentions that the new U.S. sanctions could limit access to American consumer data, potentially affecting how Chinese models like DeepMind can expand globally.
Some experts still believe the U.S. holds the advantage, noting that the computing resources available to them are far greater, and it remains unclear how DeepMind will continue to improve its model without access to advanced chips.
But for now, DeepMind is enjoying the moment, watching as most people in China were unaware of it until recently. This sudden fame has turned Liang Wenfeng into a social media sensation in China, where he is being praised as one of the “three AI heroes” from southern Guangdong province, near the Hong Kong border. The other two are Zilin Yang, an expert from Tsinghua University, and Kaiming He, a professor at MIT in the U.S.
Before the Chinese Lunar New Year, the release of DeepMind brought excitement to Chinese internet users, especially in a tech industry that is preparing for more revenue and possibly the sale of TikTok’s U.S. business.
On Weibo, one user wrote, “This is the best gift for the New Year. May our homeland be strong and prosperous.”
AI expert Wei Sun from Counterpoint Research described the reaction in China to DeepMind’s release as a mix of “surprise and excitement,” particularly in the open-source community.
Fiona Zhou, working in the tech sector in Shenzhen, said her social media was suddenly filled with posts about DeepMind. “People are calling it ‘Made in China greatness’ and saying it shocked Silicon Valley. So, I decided to download it to check how good the app is,” she said.
Fiona tried to get her horoscope made using her birth date and time, but she was disappointed with the inaccurate results. However, she said she would still use it for her work because the app might be more useful for such tasks.