China's Robotaxi Regulatory Timeline, City by City

Beijing didn't wait for a national framework before authorizing robotaxi testing within its own pilot zones, and neither did Wuhan or Shenzhen. Each of these cities issued its own local permits, building out authorization city by city rather than through a single national rollout, and in several cases those permits eventually extended beyond testing to actual driverless commercial service. None of the three coordinated the pace of its rollout with the others; each moved on its own regulatory timeline.

The permitting inside each of these cities also tends to move through recognizable stages rather than jumping straight from no authorization to full commercial service. A company typically starts with safety-driver testing on a limited, approved road network, then petitions for driverless testing on that same network once the safety-driver phase has run long enough to build a track record, and only after that moves toward a paid commercial pilot open to the public. Full, unrestricted commercial permits sit at the far end of that sequence, and not every city that authorizes early-stage testing follows through to the commercial end of it. The pattern rewards operators willing to work through each stage methodically in a given city rather than those hoping to skip ahead based on progress made somewhere else. A city's willingness to move a company from one stage to the next often depends as much on how the earlier stage went, measured in incident-free operating time and public reception, as on anything written into the original permit.

That structure looks familiar to anyone who has followed how the United States handles the same question. Individual states, not a single federal body, issue the permits that let autonomous vehicles operate on public roads there too.

The parallel is structural, not institutional.

Chinese municipal pilot-zone authorization and American state DMV licensing sit inside very different regulatory systems, with different lines of accountability and different histories behind them. What they produce, though, is the same visible pattern: a patchwork built one jurisdiction at a time rather than approved in a single stroke.

Three companies in particular have built out multi-city operating footprints under this framework, though not at the same pace or on the same timeline. Baidu's Apollo Go expanded through the framework on its own trajectory, while Pony.ai and WeRide pursued multi-city growth as competitors working through that same city-by-city permitting structure. Neither company's expansion depended on the other's progress, which is itself a feature of a framework built one municipality at a time.

Wuhan in particular became one of the more closely watched cities in this rollout, less because its regulatory framework differed from Beijing's or Shenzhen's and more because of the scale of robotaxi fleet operations concentrated there, largely under Baidu's Apollo Go brand. A city that becomes a proving ground for fleet operations at meaningful scale draws a different kind of scrutiny than one running a handful of pilot vehicles on a limited loop, and Wuhan's experience has functioned as something close to a bellwether for how Chinese regulators respond once robotaxi service moves past the demonstration stage.

Whether that structure ends up faster or slower than a single national approval process is genuinely unclear. A city-by-city system means faster movement in whichever city says yes first, but it also means a company operating in Beijing has to run the same authorization process again in Wuhan, with no guarantee the second city moves as quickly as the first one did.