Guangdong courts Tsinghua's coders after losing DeepSeek, Moonshot to rivals
For a province that gave China two of its most celebrated artificial intelligence founders, Guangdong has spent the past few years watching its brightest exports build empires elsewhere. This week, the southern manufacturing heartland tried a different tack: it rolled out the welcome mat for nearly 40 computer science undergraduates from Tsinghua University, hoping that a week of factory-floor tours and policy pitches might do what hometown pride could not. The message is blunt — Guangdong wants its AI future to stay home this time.
The visitors are part of a targeted charm offensive by provincial authorities, who are staring at a hard truth in the national AI rankings. Guangdong raised Liang Wenfeng, the reclusive founder of DeepSeek, and Yang Zhilin, who built Moonshot AI, yet neither chose to launch their ventures here. Yang, a Tsinghua graduate who later earned a PhD at Carnegie Mellon, returned to China and set up Moonshot in Beijing in 2023. Liang went east to Zhejiang University, spent time in Chengdu, and ultimately planted his flag in Hangzhou, where he runs both his quantitative hedge fund and DeepSeek. Guangdong was left with the bragging rights and little else.
The numbers explain the anxiety. According to the Hurun Research Institute, Beijing hosts 19 of China's top 50 AI companies, Shanghai claims 14, and Guangzhou and Shenzhen — Guangdong's twin giants — must share just 10 between them. Meanwhile, second-tier cities are no longer waiting for scraps. Hangzhou has become a gravitational centre with DeepSeek and Unitree Robotics, while Hefei and Wuhan are building serious momentum around semiconductor champions ChangXin Memory Technologies and Yangtze Memory Technologies Corp. Guangdong, for all its manufacturing muscle and export wealth, is being outmanoeuvred in the one race that now defines China's technological future.
What makes this sting is that the province has the raw ingredients. It is China's factory floor, home to vast supply chains in electronics, autos and robotics, and it has the capital to match. But experts say the gap comes down to something harder to buy: academic talent. Beijing has Tsinghua and Peking University feeding a dense ecosystem of labs, spin-offs and venture funds. Shanghai has Fudan and Shanghai Jiao Tong. Guangdong's universities, while improving, do not yet produce the same concentrated flow of elite AI researchers, and the province has historically struggled to retain those it does train. The Tsinghua visit is an attempt to short-circuit that weakness by importing talent directly, even before graduation.
For international observers of Asian capital, this is a reminder that China's AI race is not a Beijing-versus-Shanghai story anymore. The geography of innovation is fragmenting, and provincial governments are now competing with the ferocity of private equity firms. Guangdong's move is less about hospitality than about signalling to young researchers that the province understands the stakes. The deeper question is whether a week of visits can overcome structural pull — the allure of Beijing's investor density, Hangzhou's startup culture, or the simple fact that the two most famous AI founders in China both chose to leave.
What happens next will depend on whether Guangdong can convert attention into infrastructure. The province has already shown it can build things at scale; the challenge is building the kind of ecosystem where a PhD student feels they can start a moonshot — and stay. If the Tsinghua delegation returns with a few converts, this week may be remembered as the moment Guangdong stopped exporting talent and started importing it. If not, the province will keep funding someone else's AI revolution, one brilliant graduate at a time.
