Baidu faces robotaxi permit suspensions
- China has suspended new Level 4 autonomous-driving permits after Baidu’s Apollo Go fleet suffered a mass outage in Wuhan on March 31. - More than 100 robotaxis stalled mid-road, some riders were trapped for nearly two hours, and Wuhan — Baidu’s biggest hub — reportedly paused operations. - That matters because China had been scaling robotaxis fast, with Apollo Go logging millions of rides and pushing commercial expansion before this setback.
Robotaxis are supposed to fail gracefully. That is the whole pitch. A car can have a bad sensor read, a lost connection, even a software hiccup — but the system should still get the passenger somewhere safe. What happened in Wuhan was the opposite. More than 100 Baidu Apollo Go robotaxis reportedly froze in traffic on March 31, and now Chinese regulators have responded by suspending new permits for Level 4 autonomous-driving services nationwide. (bloomberg.com) ### What actually went wrong in Wuhan? The short version is ugly. Apollo Go cars stopped in live traffic, some on elevated roads, with hazard lights flashing while passengers sat inside and regular drivers tried to get around them. Local police said the vehicles were hit by a system malfunction, and several reports say some rider(bloomberg.com)t is still a nightmare failure mode for a commercial robotaxi fleet. (yahoo.com) ### Why did regulators react so hard? Because this was not one weird car. It looked like a fleet-level failure. That changes the risk math completely — one vehicle making a bad decision is a product problem, but dozens or hundreds stopping at once starts to look like a system architecture problem. Chinese authorities then called in local governments running auto(yahoo.com)bloomberg.com) ### What does the permit freeze mean in practice? Basically, robotaxi companies can keep operating where they already have approval, but they cannot easily add new vehicles, launch new pilot zones, or enter new cities while the freeze is in place. That hits the growth story directly. Apollo Go’s Wuhan operations were also reporte(bloomberg.com)i markets. (cnevpost.com) ### Why is Baidu the center of this? Because Apollo Go is not a small experiment anymore. Baidu said in its fiscal 2025 results that Apollo Go delivered 3.4 million fully driverless rides in the fourth quarter alone, with weekly rides peaking above 300,000. Earlier in 2025, the company was still talking about rapid scale, lower vehicle costs, and expansion a(cnevpost.com) biggest real-world robotaxi deployments anywhere. (ir.baidu.com) ### Is China backing away from robotaxis? Not really. The better read is that China is doing both things at once — pushing commercialization hard, then slamming the brakes when an incident makes the risks impossible to ignore. Even after this freeze, the long-term direction still looks pro-robotaxi. Analysts had been projecting ve(ir.baidu.com)s. (cnbc.com) ### So why does this matter outside China? Because Wuhan exposed a specific weakness that every robotaxi operator has to think about — not just “can one car drive safely,” but “what happens if a shared backend, network dependency, or fleet command fails all at once?” That is a different category of safety problem. And it is the kind regulators everywhere notice fast, because scale turns a software bug into a city traffic event. (carnewschina.com) ### What is the bottom line? China is still one of the world’s most aggressive robotaxi markets. But the Wuhan outage showed the catch — fast scale only works until a fleet-wide failure makes regulators question the whole operating model. Baidu is still a major player, but for now the story is not expansion. It is proving that a robotaxi network can fail safely when the network itself breaks. (bloomberg.com)
