China is on track to dominate consumer artificial intelligence applications and robotics manufacturing in just a few years, but the United States will maintain its substantial lead in enterprise AI adoption and cutting-edge research. kai-fu leeOne of the world’s most prominent AI scientists and investors.
In a rare, unvarnished assessment delivered via video link from Beijing ted ai conference In San Francisco on Tuesday, Lee — a former executive at Apple, Microsoft and Google who now runs both a major venture capital firm and his own AI company — laid out a technology landscape divided along geographic and economic lines with profound implications for both business competition and national security.
"China’s robotics has the advantage of integrating AI into much lower costs, better supply chains and faster turnaround, so companies prefer unitary Leading the world in building truly affordable, embodied humanoid AI," Li said, referring to a Chinese robotics manufacturer that has undercut Western competitors on price while advancing capabilities.
The comments, made to a room packed with Silicon Valley executives, investors and researchers, represented one of Li’s most detailed public assessments yet of the comparative strengths and weaknesses of the world’s two AI superpowers — and suggested that the race for artificial intelligence leadership is becoming less a single contest than a series of parallel contests with different winners.
Why is venture capital flowing in opposite directions in the US and China?
At the heart of Lee’s analysis are fundamental differences in how capital flows through the innovation ecosystems of the two countries. American venture capitalists are pouring money into generic AI companies building large language models and enterprise software, Li said, while Chinese investors are betting heavily on robotics and hardware.
"VCs in America do not fund robotics like they do in China." Lee said. "Just like VCs in China don’t fund generic AI the way VCs in the US do."
This investment divergence reflects different economic incentives and market structures. In the United States, where companies have become accustomed to paying for software subscriptions and where labor costs are high, enterprise AI tools that boost white-collar productivity command premium prices. In China, where software subscription models have historically struggled to gain popularity but manufacturing dominates the economy, robotics offers a clear path to commercialization.
The result, Lee suggested, is that each country is leading the way in different areas – and may continue to do so.
"China has to overcome some challenges for OpenAI or Anthropic as well in financing the company." Lee acknowledged, referring to major US AI laboratories. "But I think the US, on the other hand, will have trouble developing investment interest and value creation in robotics" Area.
Why US companies dominate enterprise AI while Chinese companies struggle with subscriptions
Lee was clear about the area where the United States maintains a durable advantage: getting businesses to actually adopt and pay for AI software.
"Enterprise adoption will clearly be led by the United States," Lee said. "Chinese companies have not yet developed the habit of paying for software on subscription."
This seemingly minor difference in business culture – whether companies will pay a monthly fee for software – has become a significant factor in the AI race. explosion of spending on equipment such as GitHub Copilot, Chatgpt EnterpriseAnd other AI-powered productivity software has boosted the ability for American companies to invest billions in further research and development.
Li said China has historically overcome similar challenges in consumer technology by developing alternative business models. "In the early days of Internet software, China was also significantly behind because people were unwilling to pay for software," He said. "But then the advertising model, the e-commerce model really pushed China forward."
Still, he suggested, someone will need it. "Find a new business model that doesn’t just pay per software per use or per month. This is not going to happen in China in the near future."
The implication: US companies making enterprise AI tools have a window – perhaps a large window – where they can generate revenue and reinvest in R&D without facing serious Chinese competition in their core market.
How ByteDance, Alibaba and Tencent will overtake Meta and Google in consumer AI
Where Li sees China moving forward decisively is in consumer-facing AI applications – the kind embedded in social media, e-commerce and entertainment platforms that billions of people use daily.
"In terms of consumer use, it is likely that," Li said, referring to China’s ability to match or surpass the United States in AI deployment. "Chinese giants, such as ByteDance And alibaba And TencentCompanies like will certainly grow much faster than their counterpart in the United States meta, youtube And so on."
Li pointed to a cultural advantage: Chinese technology companies have spent the last decade honing user engagement and product-market fit in extremely competitive markets. "The Chinese giants really work hard, and they have mastered the art of finding a suitable product for the market," He said. "Now they have to add technology to it. So it’s inevitably going to happen."
This valuation is in line with recent industry comments. ByteDance tiktok It became the world’s most downloaded app through sophisticated AI-powered content recommendation, and Chinese companies have pioneered AI-powered features in areas such as live-streaming commerce and short-form video, which Western companies later copied.
Li also said that China has already deployed AI more widely in some domains. "There are many areas where China has also done very good work, such as more widely using computer vision, speech recognition and translation," He said.
The surprising open-source makeover that has Chinese models beating the llamas of meta
Perhaps Lee’s most notable data point relates to Open-source AI development – An area where China has snatched leadership from American companies in a very short time.
"The 10 highest rated open sources (models) are from China," Lee said. "These companies have now overtaken Meta’s Llama, which used to be at number one."
This represents a significant change. of meta llama model As recently as 2024 it was seen as the gold standard for open-source large language models. But Chinese companies – including Li’s own firm, 01.AITogether alibaba, Baiduand others – have released a flood of open-source models that, according to various benchmarks, now outperform their American counterparts.
The open-source question has become a flashpoint in AI development. Lee made a comprehensive case for why open-source models will prove essential to the future of technology, even though closed models from companies like OpenAI command higher prices and, often, better performance.
"I think there are several major advantages of open source," Lee argued. With the open-source model, "You can test it, tune it, improve it. It’s yours, and it’s free, and it’s vital for building if you want to build an application or tune the model to do something specific."
He made an analogy of an operating system: "People working in operating systems liked Linux and that is why it was adopted rapidly. And I think that in the future, open source will allow people to design a sovereign model for a country, making it work better for a particular language."
Nevertheless, Lee predicted that both approaches would co-exist. "I don’t think the open source model will win," He said. "I think like we have Apple, which is closed, but offers a somewhat better experience than Android… I think we’ll see more apps using the open-source model, more engineers wanting to build open-source models, but I think more money will remain in the closed model."
Why China’s manufacturing advantage makes the robotics race ‘not over, but almost certain’
On robotics, Li’s message was clear: China’s combination of manufacturing capacity, low costs, and aggressive investment has created an advantage that will be difficult for American companies to overcome.
When asked directly whether the robotics race was already over with China’s victory, Li hedged a bit. "It’s not over, but I think America is still able to come up with the best robotic research ideas," He said. "But VCs in America do not fund robotics like they do in China."
The challenge is structural. Building robots requires not only software and AI, but also large-scale hardware manufacturing — exactly the kind of integrated supply chain and low-cost production that China has spent decades perfecting. While American laboratories in universities and companies like boston dynamics To continue producing impressive research prototypes, China needs the manufacturing ecosystem it has in place to turn those prototypes into affordable commercial products.
companies like unitary has convincingly demonstrated this benefit. The company’s humanoid robots and quadrupedal robots cost a fraction of their American-made counterparts, while offering comparable or superior capabilities – a price-to-performance ratio that could prove decisive in commercial markets.
What worries Lee most: Not AGI, but race itself
Despite his generally measured tone regarding China’s AI development, Li expressed concern about one area where he believes the global AI community faces real danger – not the far-future risk of superintelligent AI, but the near-term consequences of moving too fast.
When asked about agi riskLee repeated the question. "I am less afraid of AI becoming self-aware and posing a threat to humans in the short term," He said, "But the greater concern is that it’s being used by bad people to do terrible things, or by the AI race driving people to work so hard, work so fast and furiously and move fast and break things that they create products that have problems and holes that can be exploited."
He continued: "I am very worried about this. In fact, I think something terrible will happen that will be a warning about this kind of problem."
Li’s perspective holds unusual weight because of his unique vantage point spanning both Chinese and American AI development. In a career spanning over three decades, he has held senior positions in Apple, MicrosoftAnd GoogleEven during installation Sinovation VenturesWhich has invested in more than 400 companies in both the countries. His AI company, 01.AIFounded in 2023, has released several Open-source model He is one of the most capable people in the world.
For American companies and policymakers, Lee’s analysis presents a complex strategic picture. The United States appears to have a clear advantage in enterprise AI software, fundamental research, and computing infrastructure. But China is moving rapidly into consumer applications, manufacturing robotics at lower costs, and potentially leading the way in open-source model development.
Bifurcation shows that instead of single "winner" In AI, the world is moving towards a technology landscape where different countries excel in different areas – all with economic and geopolitical implications.
In form of ted ai conference As Wednesday continued, Lee’s assessment depended on subsequent discussions. His message seemed clear: The AI race is not one competition, but multiple competitions — and the United States and China are each winning separate races.
Standing in the conference hall afterward, a venture capitalist, who spoke on condition of anonymity, summarized the mood of the room: "We are no longer competing with China. We are competing on parallel tracks." Whether those tracks ultimately converge – or turn into an entirely different technology ecosystem – may be the defining question of the next decade.
