China expands robotaxi testing and scale

With no one in the driver’s seat or passenger seat, the taxi’s steering wheel still moved on its own. It signaled, changed lanes and smoothly avoided parked vehicles along the roadside. The vehicle was operating in Anting, in Shanghai’s Jiading District in northwest China, an area long regarded as a cradle of the country’s auto industry since the first Volkswagen-SAIC joint plant opened there in 1984. The self-driving robotaxi completed a roughly 10-kilometer route before returning to its starting point.

The writer recently visited Shanghai as part of the 2026 Dong-A Ilbo AI & Innovation Academy program. While business delegations have increasingly been traveling to China on so-called future innovation tours, the pace of change on the ground was still striking. Jiading offered a clear example. Once a traditional auto manufacturing base, the district has expanded into a testing ground for next-generation mobility, where fully autonomous robotaxis are now in operation.

Riding in a self-driving car is no longer particularly unusual. What stood out instead was Pony.ai, the Chinese startup operating the service, and its financing model and global strategy.

Despite expectations of heavy government backing, company representatives said Pony.ai avoids direct state investment to reduce geopolitical risk as it expands overseas. The company has taken funding from state-linked investment firms, but said direct government capital could complicate expansion into markets such as the United States and Europe.

Founded by former Baidu engineers in Silicon Valley, Pony.ai pursues a dual-track strategy: building autonomous driving data and operational experience in China while raising capital from global investors to expand overseas. Its backers include Toyota, Nvidia, Uber, Alipay, the Canada Pension Plan Investment Board and China Merchants Capital.

Scale has become a key metric for attracting investment. Chinese regulators have gradually approved fully driverless testing in cities including Beijing, Shanghai, Guangzhou and Shenzhen. As a result, Pony.ai had logged 45 million kilometers of autonomous driving mileage as of March, despite operating at a loss. That is more than three times South Korea’s total autonomous driving distance of 13.06 million kilometers as of November last year.

Baidu’s overseas push follows a similar pattern. Swiss authorities recently approved Level 4 autonomous driving trials, currently limited to rural public transport support. The move is widely seen as an early step into the European market, despite a previous incident in Wuhan in which more than 100 Baidu robotaxis came to a simultaneous halt. Even so, investors and governments appear to be prioritizing firms with proven real-world operating data.

The expansion is not limited to autonomous driving. Chinese artificial intelligence companies are increasingly turning to Hong Kong listings to raise capital and accelerate overseas growth, helping push the city’s IPO market in the first quarter of this year to its strongest level since 2021.

South Korea is also investing in artificial intelligence through initiatives such as the National Growth Fund and domestic foundation model projects. In an ecosystem that still lacks long-term patient capital, policy finance plays a key role. But equally important is creating environments where companies can compete in real markets, fail and try again, and build operational track records that attract investment.

Such testing environments remain limited in South Korea. Regulations still restrict robotaxi deployment, and AI companies are often said to focus more on securing government funding than on market experimentation. Establishing a broader regulatory sandbox may therefore be a necessary first step.