Amid the AI race, China’s elite “genius classes” have quietly become one of the most potent engines behind its rise in advanced technology, particularly artificial intelligence.
A recent Financial Times investigation highlights how a vast, fiercely competitive system embedded in top high schools churns out thousands of scientific prodigies each year. Many of them now power China’s leading tech firms, including ByteDance, Meituan, and AI pioneers like DeepSeek and Alibaba’s Qwen.
Inside China’s assembly line for science prodigies
The story of one Beijing mother, Stacey Tang, captures the stakes. In late 2022, amid strict COVID-19 lockdowns, she received a mysterious phone call instructing her to send her 15-year-old son to sit an entrance test for a “genius class.” The test, held inside a moving van as it drove around the city, involved university-level maths problems. “In any other country, you’d think it was lunacy,” she said. “But in China, I was weeping with joy. It was his golden ticket.”
Every year, about 1,00,000 students are funnelled into such elite “experiment” or “competition” tracks designed to prepare them for international Olympiads in maths, physics, chemistry, biology, and computer science. For many, it’s a direct route to China’s top universities, and, increasingly, to the heart of its tech revolution.
From Olympiad gold to AI gold rush
The results are staggering. Graduates of this system include the founders of ByteDance, the architects of TikTok’s recommendation algorithm, and the brains behind Taobao, PDD, and Meituan. Engineers from the same stream helped build China’s largest language models, such as DeepSeek’s R1, which stunned the world in 2025 by matching OpenAI’s best systems at a fraction of the cost.
At DeepSeek’s Beijing offices, 21-year-old Wang Zihan, a former “genius class” student from Wuhan, joined as an intern in 2024, unaware the company was about to upend global AI expectations. “The education I had was brutal, but it taught me to think like there’s no challenge I can’t overcome,” he said.
DeepSeek’s founder, Liang Wenfeng, insists that this homegrown talent is China’s real competitive edge. “We want to grow our own top talent. China will always be a follower,” he said in a rare interview.
Even Nvidia CEO Jensen Huang has acknowledged the strength of Chinese talent. “You walk up and down the aisles of Anthropic or OpenAI or DeepMind, and there are extraordinary researchers from China,” he said. “It’s no surprise their work is world-class.”
How Beijing built a scientific superpower
China’s obsession with science education dates back decades. After entering the International Mathematical Olympiad in 1985 and bringing home just one bronze medal, the nation turned the event into a national mission. By 2025, 22 out of 23 Chinese Olympiad contestants returned with gold medals, a record that underscores the country’s relentless focus on talent cultivation.
The 2000s saw reforms allowing competition winners to skip the notoriously grueling gaokao university entrance exam and gain direct admission to elite institutions under the 985 Project. This turbocharged interest in elite training tracks but also led to backlash from parents when most students failed to make the final cut.
Still, the momentum only grew. By the late 2010s, informatics had overtaken physics and maths as the top field, reflecting the rise of AI. When Beijing declared artificial intelligence a “key national growth strategy” in 2017, dozens of AI-branded “genius classes” sprang up overnight.
One of the most prestigious programmes, Tsinghua University’s “Yao Class,” founded by China’s only Turing Award winner Andrew Yao, has produced some of the nation’s leading tech entrepreneurs, including Pony.ai’s co-founder Lou Tiancheng. “We were taught to think independently, not memorise,” Lou recalled. “That’s what later shaped our belief in self-learning autonomous systems.”
Talent as China’s ultimate weapon
China now produces around 5 million STEM graduates each year, 10 times as many as the United States, according to Xinhua. Among them, tens of thousands emerge from the genius pipeline, equipped with the mathematical and coding prowess to drive the country’s AI ambitions.
For Dai Wenyuan, founder of AI firm Fourth Paradigm and a product of the system himself, talent remains China’s most significant advantage. “There are more than 1,000 registered generative AI models in China, unthinkable anywhere else,” he said. “Twenty years ago, we had zero AI experts. Now we’re mass-producing them. The real geniuses who’ll change the world are probably already in class.”
