Yang Zhilin completed his doctorate at Carnegie Mellon University in just four years —two fewer than usual— an achievement that opened every door in Silicon Valley. Apple, Google and Meta wanted to hire him. MIT and Stanford rolled out the red carpet. But he chose the less obvious path: he rejected the millions of dollars awaiting him in the US and returned to China to found his own company. His decision is not exceptional, but rather a reflection of a profound geopolitical shift that is reshaping the global race for artificial intelligence.

From Carnegie Mellon to Moonshot AI

Yang Zhilin was already a prominent researcher before completing his doctorate. He collaborated on one of the foundational works that laid the groundwork for ChatGPT: the paper on Transformer-XL, a milestone that marked a before and after in his career. Russ Salakhutdinov, his supervisor at CMU, later recalled what the decisive conversation was: Yang confessed that if he did not attempt to create his own company, he would regret it for the rest of his life.

Instead of founding his startup in the US —where he would have had abundant funding and a mature ecosystem— Yang went to China and created Moonshot AI. The bet paid off: he developed the Kimi family of artificial intelligence models, whose latest version, Kimi K3, has burst onto the market with competitive capacity that rivals solutions from Anthropic and OpenAI.

The brain drain nobody expected

According to xataka.com, Yang's story illustrates an increasingly serious structural problem for the United States. A Stanford Hoover Institution study analyzed the profile of researchers at DeepSeek with revealing data:

• 53.5% never studied or worked outside China
• Of the remaining 46.5% who did have international experience, 70% ended up returning to their country of origin

China has stopped being an exporter of scientific talent. Now it builds its own internal pool of world-class engineers, self-sufficient and outside the American institutions that for decades monopolized technological innovation.

Why they choose China over Silicon Valley

The explanation goes beyond patriotism. Chinese entrepreneurs in the AI sector acknowledge that a startup in the US would give them a valuation up to 10 times higher, but in China they achieve something more valuable: launching real products to market in less than five years. The density of engineers, execution speed and a less restrictive regulatory environment more than compensate for the drawbacks.

Moreover, US immigration policy favours large technology companies over entrepreneurs. A talent graduated abroad faces visa and employment barriers that push them directly to work for Google or Meta, not to found their own startups. In China, that same engineer finds fertile ground to build from scratch.

The pressure pushing people back home

The phenomenon affects not only recent graduates, but veteran professionals already integrated into American academia and industry. A Stanford University survey in 2024 among 1,304 scientists of Chinese origin residing in the US revealed a worrying figure: 73% did not feel safe working as a researcher in the country. 65% pointed directly to fears of harassment and racist violence.

Political uncertainty, institutional distrust and a climate of suspicion towards researchers of Chinese origin have made the US a hostile place for precisely the talent it most needs to maintain its technological leadership.

What it all means

Yang Zhilin's decision was not an isolated case, but a pattern. Every brilliant engineer who chooses to return to China is an artificial intelligence investment that the US loses and that will end up competing directly against its own interests. In the global race to dominate AI, that demographic imbalance could prove decisive in the coming years.