China's southern manufacturing powerhouse of Guangdong province welcomed nearly 40 computer science students from Tsinghua University this week, part of a targeted push by local authorities to bridge a widening gap in China's artificial intelligence race by attracting top-tier talent from Beijing, the South China Morning Post reported on August 16, 2026.
The visit carries a symbolic weight that is hard to miss. Guangdong is the home province of two of China's most famous AI founders — DeepSeek's Liang Wenfeng and Moonshot AI's Yang Zhilin — yet neither built his company there. Their departures illustrate exactly the problem the province is now spending effort, and prestige, to solve, and it is one of the more quietly important subplots in the global competition for AI talent that our continuing industry coverage tracks week by week. For more context on this story, see our ongoing AI trends.
The Founders Who Got Away
The career paths of the two founders, as traced by the SCMP, show how thoroughly China's AI talent has concentrated outside Guangdong despite the province producing it.
Yang Zhilin graduated from Tsinghua University in Beijing, earned his PhD at Carnegie Mellon University in the United States, returned to China, and launched Moonshot AI — the company behind the Kimi chatbot — in Beijing in 2023. Liang Wenfeng attended Zhejiang University in eastern China, spent time in Chengdu in the southwestern province of Sichuan, and ultimately settled in Hangzhou in Zhejiang province, where he launched both his quantitative hedge fund and, later, DeepSeek.
In other words, China's two most celebrated AI startups of the past two years both emerged from a province more than a thousand kilometers from where their founders grew up — one in the capital's academic orbit, the other in Hangzhou's growing tech cluster.
The Numbers Behind the Anxiety
Industry metrics underscore Guangdong's uphill battle. According to the Hurun Research Institute, as cited by the SCMP, Beijing leads the country with 19 of China's top 50 AI companies, followed by Shanghai with 14. Guangzhou and Shenzhen — the two biggest cities in Guangdong, and two of China's most economically powerful metropolises — hold just 10 between them combined.
Meanwhile, second-tier cities are rapidly gaining ground. Hangzhou is thriving with star firms like DeepSeek and Unitree Robotics, the robotics maker whose humanoid videos have circled the globe. Hefei and Wuhan are building momentum around semiconductor giants ChangXin Memory Technologies and Yangtze Memory Technologies Corp, respectively, anchoring their own AI-adjacent ecosystems.
A core factor behind this disparity, according to experts cited by the SCMP, is the concentration of academic talent. Beijing's universities — Tsinghua and Peking University foremost among them — feed more of China's top AI researchers into the workforce than any other region, and startups cluster around that pipeline. Bringing Tsinghua students to Guangdong for firsthand exposure to local companies is a direct attempt to interrupt that gravity.
Why It Matters Beyond China
The talent contest inside China has consequences for the global AI industry. The United States has spent 2026 pressing allies to pick sides in an AI race with China, and regulators have debated restricting Chinese models in Western markets. But the underlying competition is increasingly a competition for people — researchers, engineers, and founders who decide where the next DeepSeek gets built.
Guangdong's push also reflects a broader lesson about how AI ecosystems form. The province is hardly short of resources: it is China's most populous province, home to Huawei and Tencent's home city of Shenzhen, and the heart of the country's electronics manufacturing belt. What it has lacked, the Hurun numbers suggest, is the dense academic-and-startup loop that turns a regional economy into an AI hub.
Local authorities have signaled their ambitions before. Guangdong has previously announced aims of becoming a global AI and robotics hub in the wake of DeepSeek's success, as the SCMP has reported, and the province has been pouring investment into robotics and intelligent manufacturing initiatives. Hosting Tsinghua's students is the talent-facing half of that same strategy — an attempt to make sure the next generation of founders at least considers staying.
A Long Game With No Guaranteed Return
Whether any of it works is another question. Talent agglomeration is notoriously self-reinforcing: researchers go where other researchers are, founders go where capital and cofounders are, and Beijing's lead in the Hurun rankings is the product of decades of accumulation. Cities like Hangzhou that have broken through did so with a combination of university strength, patient local capital, and at least one breakout company that recalibrated everyone's expectations — DeepSeek, in Hangzhou's case.
Guangdong's bet is that manufacturing depth plus sustained talent outreach can produce the same effect. The province's electronics supply chains and robotics factories give it an industrial base for embodied AI that few regions anywhere can match, and a founder who wants to build AI hardware still has reasons to look south.
For now, the marker is small but real: nearly 40 of Tsinghua's best computer science students spent a week in the province that produced Liang Wenfeng and Yang Zhilin, seeing what their home region has to offer. Whether any of them become the founder who finally stays is a question that will take years to answer — and one that will help decide where China's next wave of AI companies calls home.
