The new frontier in the US-China AI race: Talent

BEIJING: When Will Wang left Apple in Silicon Valley in 2018, he was not chasing the next tech startup.

He was looking for a place to build hardware.

A graduate of Shanghai Jiao Tong University and UC Berkeley, Wang worked on the Apple Watch team before returning to China.

After stints at Chinese consumer electronics companies including Anker Innovations and Oppo, he founded Shenzhen-based smart glasses startup Even Realities in 2023.

The reason, he said, was simple.

“Silicon Valley doesn't really (reward) making hardware anymore,” Wang told CNA, adding that the industry’s focus had shifted to software, data - and now artificial intelligence.

For Wang, China offered the engineers, supply chains and manufacturing ecosystems needed to build the next generation of AI hardware.

“If I wanted to (lead) in AI, I needed to be in Silicon Valley,” Wang said.

“But if I want to innovate in hardware, I need to be located in the heart of hardware.”

Wang’s decision reflects a broader shift unfolding at the centre of the global AI race.

For decades, talent moved freely between China and the United States - but today, that once-familiar talent flow is becoming increasingly complicated.

As Beijing and Washington compete for AI leadership, talent is increasingly viewed as a strategic resource alongside advanced semiconductors, computing power and data.

Governments have been tightening scrutiny over technology transfers, investment and research collaboration, while companies on both sides are competing for a relatively small pool of elite engineers and scientists capable of pushing the frontiers of AI.

The contest is no longer just about who builds the best models, experts said - it is also about where the world’s most valuable researchers, founders and engineers choose to live, work and build.

Interviews with founders, recruiters and researchers suggest the result is not a simple reversal of talent flows from the US to China.

Instead, a more selective and politically charged pattern is emerging, one that could reshape how innovation moves between the world's two leading AI powers.

TWO AI ECOSYSTEMS, DIFFERENT STRENGTHS

For years, Chinese scientists and engineers attended US universities, joined Silicon Valley firms and research labs - and in many cases, returned home with expertise, networks and experience to build companies and institutions of their own.

Li Yaqi, a research assistant at the S Rajaratnam School of International Studies (RSIS) in Singapore who studies US-China AI governance and tech ecosystems, describes the phenomenon as “brain circulation” - not “brain drain”.

“America's AI advantage has been institutionally American but demographically global,” Li told CNA.

Chinese returnees have also played a significant role in developing China’s own tech ecosystem. Notable figures like Taiwanese computer scientist Lee Kai-Fu and Turing Award winner Andrew Yao helped train generations of entrepreneurs and researchers who would go on to shape China’s AI industry.

The educational pipeline also remains substantial. China was the second-largest source of international students in the US during the 2024 to 2025 academic year, according to Open Doors data.

For years, researchers moved between the two countries - carrying with them not just technical skills but networks, research practices and essential experience.

“The relationship was never only a flow of people,” Li said.

“It trained researchers, built companies, and kept each side legible to the other.”

Kelvin Sun, co-founder of talent intelligence platform DINQ, said China’s pull was no longer anecdotal.

A growing number of researchers and executives have taken on senior roles at Chinese firms and laboratories, reflecting the maturation of the country's AI ecosystem.

Yet Sun argued that decisions about where top AI talent chooses to work are rarely driven by a single consideration.

“For the top tier, my top three are compensation, compute, and research freedom,” he told CNA.

“Cash gets people in the door; compute and impact are what keep them,” said Sun.

For elite researchers, access to computing resources has become especially important as AI development grows more expensive.

Training cutting edge models requires sizeable amounts of capital and specialised hardware - resources that remain concentrated among a relatively small number of firms and laboratories - and those strengths continue to favour the US.

The San Francisco Bay Area still retains the densest concentration of elite AI talent, frontier research and computing resources needed to train advanced models.

But at the same time, China now offers something it could not provide a decade ago - a robust and mature tech ecosystem of its own with world-class companies, engineers and domestic markets large enough to build and test ambitious products at scale.

“Talents move to wherever the future is, not wherever the government is paying the most,” said Will Wang.

Wang Li, co-founder and chief operating officer of Even Realities, told CNA that young engineers and researchers are primarily looking for opportunities to work on cutting-edge products and learn from top teams.

“They just want to find the best company and the most innovative company that they can work for, and also grow with,” she said.

Others see the US and China ecosystems as being increasingly complementary, even as global geopolitical tensions grow.

Deng Honghao, co-founder and chief executive of Butlr, a US-based company that grew out of research at the MIT Media Lab, said his career had been shaped by both systems.

Born in China and raised in Beijing, Deng studied architecture at Tianjin University before moving to the US for graduate studies at Harvard University and later conducted research at MIT.

He stressed that talent decisions are rarely as simple as choosing between China and the US.

“Both sides have this strong magnetic field for talent and also resources,” he told CNA.

The result, observers said, is not convergence but specialisation.

The US remains the stronger draw for those seeking frontier models, elite research teams and deep pools of AI capital.

China is increasingly attractive for those focused on hardware, commercialisation and the physical deployment of AI-enabled products.

IS THE REVOLVING DOOR SLOWING?

The US has tightened China’s access to advanced chips, investment flows and some AI-related services, while China has become more alert to the possibility that strategically important technology, data and talent could move beyond its reach and control.

China’s new outbound investment rules, which take effect from Jul 1, are part of that broader shift.

The regulations tighten oversight of overseas deals involving Chinese investors, especially in sensitive sectors and transactions linked to technology, data and national security.

One closely watched case was Manus, an AI startup with Chinese roots that relocated to Singapore last year before attracting regulatory scrutiny in China.

After reports emerged of a Meta deal involving the company, Chinese authorities reviewed whether the deal violated investment rules.

Two Manus co-founders were also reportedly barred from leaving China while the matter was under review.

The episode highlighted how talent, ownership and technology transfer are becoming harder to separate.

Visa restrictions, export controls, national security reviews and travel scrutiny have added layers of uncertainty to decisions about where to live and work that were once framed largely as career choices.

Li of RSIS described the trend as one of segmentation.

Elite researchers, founders and executives may still have the networks and resources to move across borders - but the broader pipeline of students, mid-career engineers and early-stage entrepreneurs may become less fluid.

That could have wider consequences beyond a simple talent redistribution from one country to another.

“What stops moving is the tacit layer,” Li said.

The deeper loss would be the movement of research instincts, management habits, product experience, laboratory culture and informal knowledge that do not travel easily through papers, patents or public datasets, he added.

TALENT AS A STRATEGIC ASSET

The changing politics of talent is also visible in how such movements are portrayed.

Chinese media reports have increasingly highlighted overseas-trained scientists and AI researchers who return to China and take up senior roles at Chinese firms - presenting them as contributors to national technology ambitions.

But several interviewees have warned against reducing individual choices to geopolitics alone.

Wang Li of Even Realities said young engineers and entrepreneurs should first focus on learning and problem solving rather than geography.

“They do not need to optimise geography - they need to optimise for the learning, where they can find the most talented people to work with, and also to choose those problems that actually matter to solve,” she said.

Deng of Butlr argued that competition between the US and China did not have to eliminate collaboration, especially in technologies meant to solve practical problems for people.

“The right tech for (human beings) should be borderless,” he said.

Yet that tension captures the changing nature of tech talent mobility.

Today talent movements are drawing greater scrutiny because they carry more than just skills, industry observers said - but data, networks, technical judgment and institutional knowledge.

The revolving door has not closed - it just no longer swings as freely as before.

The deeper risk, Li argues, is not simply that one side gains talent at the expense of the other.

It is that both become less able to understand how the other works.

Talent mobility was one of the few channels through which China and the US could understand each other from the inside.

Without that human exchange, both sides risk relying increasingly on official narratives, media coverage and worst-case assumptions.

“The real risk here is not that one side loses the race,” Li said.

“It is that both run a slower race and call it winning.”

As AI competition between Washington and Beijing intensifies, the battle is no longer only over chips, models and capital.

It is increasingly over people - and where they choose to build the future.

Source: CNA/lg(ht)