Generated by Codex with GPT 5.6 Sol XHigh

China’s next export surge may arrive without a shipping container. Chinese artificial-intelligence companies are rapidly taking robotaxis, delivery vehicles, cloud services and chatbots abroad. Their expansion is often described as a sign of technological strength. It is also a response to a harder economic problem: many of these businesses struggle to earn money at home, while Beijing wants the benefits of AI without the disruption that mass automation could bring to Chinese workers.

Robotaxis show how the strategy works. A Swiss public-transport company plans to use autonomous vehicles from Apollo Go, which is owned by Baidu. More than 70 Chinese “physical AI” firms have already established operations outside the country, and roughly 20 more are preparing to do so. Autonomous-logistics companies such as Neolix operate in dozens of markets. American rivals, by contrast, spent years concentrating on their large domestic market and are only beginning to expand overseas.

The push from a difficult home market

China gave its AI companies an unusually good place to develop their technology. They could deploy fleets at scale, collect vast quantities of driving data and lower the cost of expensive hardware. Yet scale has not guaranteed profit. An abundance of drivers keeps taxi fares low, and workers who lose other jobs often become cabbies, adding still more supply. Baidu once said that its Wuhan robotaxi fleet had broken even, but has offered few updates since. Pony.ai makes a similar claim for some cities.

Foreign markets offer a wider margin. A ten-kilometre taxi ride costs about 23 yuan, or \$3.40, in Wuhan. The comparable fare in the Swiss canton of St Gallen is SFr44, or \$54. Chinese firms can therefore combine technology and data developed in a low-cost domestic market with the higher prices available abroad.

The same pressure affects other AI services. Chinese consumers are reluctant to pay for chatbot subscriptions, while Chinese companies spend less than a tenth as much on information technology as their American counterparts. Exporting becomes less an optional route to growth than a way to make the economics of the industry work.

Exporting both technology and disruption

Beijing has another reason to favor overseas expansion. The Communist Party wants China to narrow America’s AI lead, but it also fears the social consequences of visible job losses. Protests have already followed suggestions that robotaxis could displace drivers. Some local authorities have stopped issuing self-driving permits, a court has sided with an employee dismissed because of AI, and a national oversight body has warned companies against using the technology expressly to replace people. Technology firms are reportedly cutting some programming jobs but rarely advertise the fact.

Overseas sales allow Chinese companies to earn revenue while moving some of AI’s employment shock beyond China’s borders. They could also extend the country’s export-led growth model as trade barriers make manufactured goods harder to sell. Adjusted for inflation, Chinese service exports doubled to \$510bn between 2015 and 2025, while the much larger flow of goods exports grew by only a little over 40%. AI services could become an increasingly important part of that trade.

The strategy still carries risks. Robotaxis and cloud systems create regulatory, legal and compliance questions that ordinary merchandise does not. Foreign governments may resist Chinese services just as they have resisted Chinese products. But many wealthy countries face labor shortages and rely heavily on immigrant drivers, which could make automation politically and economically attractive.

Standards may be the lasting export

The biggest prize may be influence over how AI systems operate. Industries need shared technical standards so that equipment, software and infrastructure work together. Western countries have traditionally set many of those rules, but standards for services such as robotaxis are still unsettled. If a Chinese provider enters a market first, its technology can become the local default.

That gives early exports a compounding advantage. Chinese companies do not merely sell vehicles or software; they establish protocols that later entrants may have to follow. American competitors could eventually find themselves adapting products to Chinese-designed systems.

China’s AI-export drive is therefore more than a search for customers. It converts the weaknesses of the domestic market - low prices, cautious buyers and political anxiety about employment - into pressure for global expansion. If Chinese firms can overcome foreign resistance, they may capture not only revenue but also the power to shape the rules of an emerging industry.