WHO IS BETTER at the technology of the future: America or China? Speculation is at fever pitch. The world’s most famous artificial-intelligence company, OpenAI, is American. Models produced by DeepSeek, a Chinese competitor, are almost as good—and cheaper.
The true winner of the AI race, however, may not be the country that invents the best models. It is more likely to be the country where governments, businesses and ordinary people use AI at scale every day. For everything from economic growth to military power, technological diffusion ultimately matters more than technological innovation. On that front, the race is closer than many in America believe.
A popular view in the West holds that China is better at inventing new technologies than implementing them. The country has invested heavily in research institutes, generating scientific breakthrough after breakthrough. In the 2010s China filed about half of the world’s new patents. But it has long been bad at putting them to use. In a recent paper, Jeffrey Ding of George Washington University draws on data from 2020 from the World Intellectual Property Organisation (WIPO), a trade group, to estimate that China is the world’s 14th-best innovator of technology, but only the 47th-best adopter.
“A diffusion-centric approach reveals that China is far from being a science and technology superpower,” says Mr Ding. China has been slow to develop links between academia and industry. It has remained closed to foreign expertise. A turgid political system, risk aversion and blunted market incentives are among the culprits. America, by contrast, has historically excelled at both innovating and diffusing new tech, from electricity to cars.
Many assume that China will be slow to adopt AI, too. Its economy, more dependent on manufacturing and agriculture than America’s, seems to have fewer companies which can benefit from the technology. Cloud computing is used less widely by Chinese firms than American ones, depriving them of scalable computing power. A consensus is forming. America’s score on the IMF’s “AI preparedness index” is 20% higher than China’s. In October Capital Economics, a consultancy, prepared a similar ranking, placing America first and China miles behind. Goldman Sachs, a bank, predicts that 30% of Chinese firms will have adopted AI by 2030, compared with 40% of American ones.
Yet America’s advantage is not as clear as many think. China is catching up fast at adopting technology. According to the latest figures from WIPO, China ranks 32nd globally for technological diffusion, 15 places higher than in 2020. From electric vehicles to QR codes and robots which deliver room service, the Chinese consumer is exposed to more sophisticated technology than the American one. Some tech figures now believe that China, not America, is the quintessential adopter. As Han Jizhong of the Chinese Academy of Sciences argued in 2023, “We have surpassed Americans in many technologies not because of pioneering technological advances, but because of the application capabilities formed from our huge market.”
Investors are starting to wonder if China will be quicker to apply AI, too. A number of banks, including Goldman Sachs and Morgan Stanley, have produced share-price indices of companies that should see higher productivity as a result of AI adoption. Goldman’s index for America, for instance, includes Walgreens Boots Alliance, a pharmacy which hopes to use AI for things like managing prescriptions. Morgan Stanley’s index for China includes China Literature, a publisher with an AI model to help writers, and Transsion Holdings, a phone-maker which has added an AI assistant to its devices. Whereas Goldman’s AI beneficiaries listed in America have underperformed the local market recently, in China the putative AI beneficiaries have outperformed (see chart 1).
Cross-country data on AI adoption are scarce. But looking at the amount spent by large companies on the technology suggests that America is ahead. American firms are big buyers of enterprise software—things like human-resources and accounting apps. Many sellers of this software, such as Salesforce and Microsoft, are infusing AI into their products. China’s enterprise-software market is about a tenth the size of America’s, and “not really that vibrant”, as Chi Ping Lau, the president of Tencent, a Chinese tech giant, said in November when explaining why his firm’s AI sales are smaller than American rivals’.
Yet AI sales may exaggerate America’s lead, for a dollar spent on Chinese AI services gets much more than a dollar spent on American ones. Prices in China are lower thanks to a price war raging among the country’s cloud-computing firms, as well as the fact that most Chinese models are open-source. Last year ByteDance, another big Chinese tech company, slashed the price of its Doubao chatbot, making it 99% cheaper than OpenAI’s ChatGPT.
A different way to gauge adoption is survey data. Last year a study by IBM, an American tech firm, found that 50% of Chinese companies used AI, versus a third of American ones. Research by McKinsey, a consultancy, suggests that 19% of Chinese people use AI at work, whereas 12% of North Americans do. Work by Japan’s government finds that a much higher share of Chinese firms than American ones have established a policy for generative-AI use.
Top of the bots
China’s adoption of AI seems to be concentrated in three parts of the economy: the public sector, consumer tech and business hardware. Take the state first, which by one estimate accounts for roughly half of the demand for DeepSeek’s model. With the Communist Party’s encouragement, local governments are already using AI for all manner of things, from rationalising hospital records to answering citizens’ questions and finding missing persons.
State backing may be inspiring others to adopt. Consider the second group, consumers, who seem keener to experiment with AI in China than in America (see chart 2). One reason, again, is price: fierce competition has made most Chinese chatbots free. Another is trust. “People in China have been more optimistic about how technology can improve their lives…in part because China doesn’t have many nostalgic memories of pre-technological eras,” argues Tilly Zhang of Gavekal Dragonomics, a consultancy. Trust is reinforced by the fact that some popular AI models in China, such as DeepSeek’s R1, show their reasoning, reducing fears of hallucinations, according to Wei Sun of Counterpoint, a firm of analysts.
Rolling out consumer-facing AI features is also riskier in America than in China. One entrepreneur trying to sell AI features to American firms complains that lawyers get in the way, with fears about data protection and copyright infringement. “There is no case law on generative AI…so lawyers take a CYA [cover-your-ass] approach,” he says. Such objections are less common in China, where copyright and data protection are weakly enforced. Technologies such as payment by QR code, which often requires access to personal data, have been widely adopted, with Chinese consumers seemingly happy to hand over their data to buy a coffee. The loose regulatory environment could quickly change, however, if an AI product provokes a government crackdown.
With more demand and easier supply, Chinese consumers have lots of AI options. In the past year ByteDance has released at least seven apps with AI functions. Tencent has promoted its chatbot within its WeChat platform. They are hardly alone: American tech firms, from Google to Meta, are stuffing their products with AI-powered features. But Chinese consumers seem more aware of these changes than those in the West: 81% of them know which types of products use AI, compared with 39% of Americans, according to Ipsos, a pollster. Lifelike “sales bots” on streaming platforms have proved so effective that China’s millions of human influencers are worried they will lose income.
After the state and consumers, the manufacturing sector is the third big customer for Chinese AI services. This marks a radical departure from America. According to our analysis of information from PitchBook, a data provider, last year 3% of America’s venture-capital dollars in AI went into manufacturing. In China the equivalent share was 43%. Some industrialists in Asia see manufacturing as the area where China may break away most from America in terms of AI adoption.
Li Qiang, China’s premier, has called for the country to “combine digital technologies with the country’s manufacturing and market strengths”. Already China has a nearly 30% share of global manufacturing, up from 20% in 2011 (see chart 3). It has more industrial robots per manufacturing worker than almost any other rich country. Now it is pushing ahead in new fields. In February BYD, an electric-vehicle maker, launched an advanced driver-assistance technology at no extra cost (Tesla, America’s EV champion, charges about $9,000 per car for its equivalent). Companies like Unitree and EngineAI make robots which dance and do kung-fu. Investors have been floored. Stocks of the biggest Chinese firms involved in humanoid robotics have risen by a third this year.
Software and hard power
AI is especially helpful given America’s sanctions on Chinese purchases of high technology. In 2019-23 (the latest available data) Chinese imports of American capital equipment fell by more than 20% in real terms, depriving Chinese firms of the latest hardware. But better software, powered by AI, allows them to get more out of old kit. AI sensors can detect when hardware is about to break, for instance; other tools help to reduce energy consumption. From Chinese economic data, we estimate that China’s manufacturing sector buys twice as much software as a decade ago.
America still has a big advantage: brute force. This year Amazon, Alphabet, Microsoft and Meta will invest some $300bn in new technologies, with much going on AI. Alibaba, Baidu, ByteDance and Tencent will invest about one-sixth of that. According to data from Dell’Oro Group, a firm of analysts, American spending on AI servers outstripped Chinese investment almost fourfold last year. America has ten times China’s number of data centres. All this could help America’s AI capabilities to get ever more sophisticated than China’s.
But the example of DeepSeek, which achieved cutting-edge capabilities at a fraction of the price of its American rivals, suggests that financial firepower only goes so far. Many in Chinese AI, including Joe Tsai, the chairman of Alibaba, reckon that America is overspending in any case. In March Mr Tsai warned that the pace of data-centre construction in America may outstrip initial demand for AI services.
The AI race will not be won in the high-tech clusters of Palo Alto or Hangzhou. Instead it will be won in places like Dayton and Zhengzhou, where ordinary companies and consumers harness the technology to do extraordinary things. China still faces many obstacles to adoption: data can be hard to access and combine, and many of its companies are not yet digitised. But the widespread belief that America will be faster to take advantage of the new technology is open to question. Even if America has the best AI models, what good are they if not enough people use them? ■
To stay on top of the biggest stories in business and technology, sign up to the Bottom Line, our weekly subscriber-only newsletter.