It was 1950. The US government detained Qian Xuesen and slapped him with a $15,000 bail, enough to keep a family comfortable for years. Caltech and his friends eventually scraped together the money. But to this day, the exact details of how it was raised and whether the bail was ever refunded remain murky.
That episode is part of what historians call the 'Qian Xuesen Mystery.' Today, we're grappling with a new riddle: Why couldn't America keep Yang Zhilin? Eighty years apart. You have to ask: what the hell kind of place is America, anyway?
Salakhutdinov, a Carnegie Mellon professor and AI research advisor, remembers it well. Yang Zhilin could have easily stayed in the United States, but he chose to return to China and start a business. So Yang came home. In 2023 he founded Moonshot AI, and its Kimi K3 model now rivals the best America has to offer.
Some eighty years ago, Qian Xuesen was a founder of NASA's Jet Propulsion Laboratory. Then, suddenly, he was accused of being a spy and a member of the US Communist Party. He spent five years under house arrest before he was allowed to leave the US. In September 1955, he sailed back to China. There, he made monumental contributions to the nation's rocket, ballistic missile, and space programs. The most famous: the Dongfeng missile series. He changed China's destiny.
Why bring this up? Because Qian's work was about the tools of a great power. It's no exaggeration to say the next set of those tools is AI models. Their power lies in leapfrogging US technology, ensuring America can no longer single-handedly steer global development.
When Kimi K3 launched in July 2026, a Silicon Valley notable asked a pointed question: Why didn't Yang Zhilin stay in the United States? The post exploded, racking up 7.47 million views in four days. The reality is, Yang's choice is far from unique. In recent years, top AI talent has been flowing back from the US to China.
The Economist crunched the numbers for NeurIPS, a premier AI conference. The share of authors from China jumped from 29% in 2019 to 51% in 2025. Over the same period, the US share slid from 20% to 12%. Meanwhile, a US survey revealed that 73% of Chinese-American scientists feel unsafe as researchers, and 65% fear hate and violence against Asians.
China's tech sector isn't just attractive. It's lucrative. The data backs it up: China's AI ecosystem is gaining appeal in capital, policy, market size, and application scenarios）. Riding the Kimi K3 wave, Moonshot AI closed four funding rounds in the first half of 2026, pulling in over $3.9 billion.
Salakhutdinov put it bluntly: The US didn't lose someone who was forced out. It lost someone who simply didn't want to stay. Yang's return to China underscores a hard truth. If America insists on decoupling from China, the price tag keeps climbing. Right now, the US still leads in AI infrastructure like semiconductors and data centers. But talent is the real engine of AI development. And there, America's disadvantage may be far worse than it thinks.
The Cost of Xenophobia: America's Self-Inflicted Wound
The Nikkei highlights another gravitational pull: the IPO dream. If Moonshot AI and DeepSeek can list in mainland China or Hong Kong, they will become new entrepreneurial icons. That could accelerate the trend of Chinese AI talent chasing success at home rather than in the United States.
Some in the US venture capital community are already urging a course correction. They argue America should be rolling out the red carpet for Chinese talent, including helping their families relocate.
Instead, the industry sees the Trump administration's chaotic anti-China policies for what they are: a self-defeating campaign of 'driving talent away.'
Postscript: Forget the 'Qian Xuesen mystery.' The real puzzle is the 'American mystery.' A New York Times article from June 5 last year, titled 'The U.S. Deported This Chinese Scientist, in a Decision That Changed World History,' tore into the Trump administration. It reported that Secretary of State Rubio "turned up the pressure, announcing that the administration would work to 'aggressively revoke' visas of Chinese students, including those with ties to the Chinese Communist Party or who are studying in 'critical fields.'".
Consider the scale: the United States hosts one million foreign students. More than 250,000 of them are Chinese.
Qian Xuesen was unreasonably treated by the US government and finally had to leave. America learned nothing. Now, beset by troubles at home and abroad, instead of striving to improve, it reaches for xenophobia as a weapon.
And yet the United States still sees itself as the world's strongest. A real unfathomable mystery!
Deep Blue
** 博客文章文責自負,不代表本公司立場 **
Zhuque has silenced America. SpaceX's repeated successes in rocket recovery drove home one point: American business works. And China? How could a feat achieved by marshaling the strength of an entire nation possibly compare with a private outfit pulling off a mission impossible?
Cast your mind back. The Clinton boom, built on the information superhighway, delivered four consecutive fiscal years of budget surplus. From 2015 onwards, SpaceX achieved rocket recovery, putting American commercial spaceflight at least a decade ahead of China. What more could anyone say? Behold the splendor of America's overall national power!
At 7:35 this morning, the reusable Zhuque-3 Y2 carrier rocket lifted off smoothly. Its builder is LandSpace, a Chinese commercial space company founded, coincidentally, in that very same year of 2015. The mission report reads: "About 137 seconds after liftoff, the first and second stages separated. The first stage then began its return sequence. After key maneuvers including high-altitude attitude adjustment, deceleration and aerodynamic gliding, it made a successful soft landing at the Gansu landing pad at 7:43. The flight test was a complete success."
A few points are worth adding. First, the Zhuque-3 Y1 made its maiden flight on December 3, 2025. It successfully delivered a satellite into its designated orbit, though its first-stage vertical recovery test failed.
Second, eight months on, Zhuque-3 has delivered on its mission, a task of major significance. Zhuque becomes our country's second reusable rocket after the Long March 10B.
Third, mind you, the Long March rockets belong to the "national team."
Fourth, China's state and commercial space programs are setting sail like "twin dragons out to sea." America is feeling the competitive pressure.
SpaceX has made a blockbuster stock market debut. The world is counting on America, following the internet, to lead the globe once more through space technology. The hope is a replay of the Clinton-era golden age. SpaceX pulled off the first successful recovery of a Falcon 9 rocket, then industrialized the technology. Today it launches the Falcon 9 around 150 times a year, roughly three times a week. A single booster can be reused dozens of times.
Yet even before the successful recovery of the Zhuque-3 Y2, American think tanks were already worried. China, doing things the Chinese way, has achieved recovery in both state-run and commercial spaceflight, with equally precise touchdown points. China's "catch-up" is no longer out of reach.
What deserves attention is this: China's state and commercial sectors are advancing in tandem, while NASA has widely adopted the commercial model. In return-to-the-Moon programs such as Artemis, NASA has contracted out uncrewed lunar landers, precursor satellite launches and even rocket servicing and replenishment support to multiple private space companies. Commercial players such as SpaceX, Firefly Aerospace and Rocket Lab are the mainstays.
Two years ago, analyses weighed the relative strengths of the Chinese and American space programs. Commentators argued that NASA's shift from "designing and building in-house" to "commercial contracting" would push national scientific talent out of front-line positions. Over the long run, that would weaken the state's "deep control and independent assessment capability" over space technology.
And if complex spaceflight life-support systems or new space station designs were left entirely to private leadership, NASA would face the twin hazards of technology and risk. Boeing's Starliner saga, which stranded two American astronauts on the International Space Station for a full nine months, is in its own way the same tune as today's USS Abraham Lincoln incident. Go figure.
The pair finally left the station "smoothly" aboard a SpaceX Dragon on March 18, 2025. They landed safely back on Earth the next day. Trump sang Musk's praises for a job well done.
A quick explainer: history has seen a so-called "Sputnik moment." In truth, it was a "Sputnik" crisis for America. In 1957, the Soviet Union successfully launched Sputnik 1, humanity's first artificial satellite. The phrase has since come to describe the turning point when a nation suddenly realizes it is lagging a rival in technology or a critical field. That realization triggers an acute sense of crisis and an urgent drive to catch up.
President Kennedy led America out of the Sputnik crisis with the Apollo lunar landing program. Today, can America build a crewed Mars program to overtake its rival? Musk, this round is all on you. Stop firing off low-orbit satellites like crazy to rake in cash. The day a crewed Mars landing succeeds will be the ultimate MAGA moment. Let's keep at it, together.
