China’s robotaxi freeze just hit the three numbers that matter most

China’s robotaxi problem just got more expensive than a traffic jam.

Reuters reported on Tuesday that Bloomberg, citing people familiar with the matter, says China has suspended new permits for autonomous vehicles after last month’s Baidu outage in Wuhan. If that holds, the damage is not limited to one ugly video of stalled cars. It means robotaxi companies could keep their current fleets on the road while losing the three ways they prove the business can still grow: adding more cars, starting new pilot programs, and entering new cities.

That is a different kind of failure. A robotaxi outage is a systems problem: the software or operations break in public. A permit freeze is a business problem: the company may still know how to run the service it already has, but regulators can stop the next step in its expansion.

Reuters said the reported halt would prevent self-driving companies from adding robotaxis to existing fleets, launching new pilot projects, or expanding into additional cities. Those are the scoreboard for this whole sector. Robotaxi companies win attention by showing they can move from one demo zone to the next, then from one city to many.

The timing matters because the Wuhan incident itself is old news. On April 1, Reuters reported that a “system failure” left multiple Baidu Apollo Go robotaxis stalled in central Wuhan, affecting at least 100 vehicles. Passengers got out safely and no injuries were reported. We already knew the machines froze. The new fact is that the freeze may now have spread from the street to the permitting pipeline.

China had already signaled a tougher posture. On April 14, Reuters reported that authorities ordered local governments to conduct self-inspections and strengthen safety oversight for road tests involving intelligent connected vehicles, China’s broad term for software-heavy cars that can handle some driving tasks. That sounded like the usual post-incident cleanup. A reported permit halt is sharper. It turns oversight into a brake.

That brake lands on a business that has been selling scale. In February, Baidu’s investor relations said Apollo Go delivered 3.4 million fully driverless rides in the fourth quarter of 2025, had topped 20 million cumulative public rides, and had reached 26 cities globally. Those numbers matter because robotaxi economics do not get interesting at five cars in a fenced district. They get interesting when companies can add vehicles, spread fixed costs, and show that one city was not a one-off.

That is what makes the history in Wuhan worth revisiting. In 2022, PR Newswire reported that Baidu had won China’s first permits to charge fares for fully driverless rides in Wuhan and Chongqing. The company said the initial approval covered five Apollo robotaxis in each city, and a 13 square kilometer service area in Wuhan. Back then, the permits read like proof that China was willing to let robotaxis graduate from showpiece to service. The current reported halt suggests that permission can narrow just as fast as it opened.

Other companies have not been pulled off the road, at least not yet. Reuters reported that Pony.ai said its robotaxi services in Beijing, Shanghai, Guangzhou, and Shenzhen are operating normally. Reuters also reported that WeRide said its China robotaxi service is still operating normally and covers more than 1,000 square kilometers. So the immediate picture is not a national shutdown. It is a market where current service may continue while future growth gets stuck in place.

That distinction is the story. For years, robotaxi coverage has treated the technical challenge as the hard part: can the car drive safely enough, often enough, cheaply enough? It still matters. But the person standing next to the machine here is a city official, not an engineer. Once a highly visible systems failure spills into traffic, the scarce resource is not just better software. It is official tolerance for putting more of these cars on the road.

The caveat belongs near the top because it changes how hard to lean on the claim. Reuters attributed the permit suspension to Bloomberg’s reporting from people familiar with the matter and said it could not independently verify the move. There is no public national notice yet in the story file. If regulators narrow the action, describe it differently, or reverse it quickly, the market-wide reading gets weaker. But if Reuters’ account holds, China has found a simple way to wound a robotaxi company without banning robotaxis outright: leave the cars that already exist where they are, and lock the next door.