Baidu's Apollo Go and China's Robotaxi Market

Apollo Go, Baidu's robotaxi service, operates one of the largest driverless fleets in the world, and its center of gravity is Wuhan rather than any single flagship city the way Waymo's service is often associated with Phoenix or San Francisco. Baidu has built out coverage across multiple Chinese cities, but the concentration in Wuhan specifically has made it the most-cited example of Chinese robotaxi scale in Western coverage.

The regulatory structure underneath that fleet doesn't map cleanly onto the US system this site's tracker was originally built around. American AV permitting runs through state DMVs and, in California specifically, a second layer at the CPUC governing commercial passenger service. China's approach is organized differently: municipal and provincial authorities designate pilot zones, each with its own rules for what a permit holder can and can't do within that zone's boundaries. There is no single national regulator issuing a driverless authority the way a US state agency does.

Fare-charging authority is its own separate permission within many Chinese pilot zones, layered on top of the basic authorization to operate a fleet at all. A city can allow a company to run driverless vehicles carrying passengers while still restricting whether those rides can be sold commercially or have to remain free demonstration rides; moving from one status to the other is a distinct regulatory step, not an automatic consequence of getting a fleet on the road in the first place. That two-step structure has a rough parallel in the US split between a testing permit and a deployment permit, even though the institutions granting each are organized completely differently.

Baidu's robotaxi hardware has also evolved through successive vehicle generations developed in partnership with Chinese automakers, each generation aimed at lowering the cost and complexity of the sensor and compute package rather than simply scaling up the same design. That progression matters for the China-versus-US comparison as much as the regulatory structure does, since a lower per-vehicle cost changes the economics of expanding a fleet within a pilot zone in a way that is somewhat independent of how permissive that zone's rules happen to be.

That structural difference is exactly why fleet-size comparisons between the US and China get complicated fast. A company can report operating vehicles under one country's counting convention and a materially different number under the other's, depending on whether test vehicles, safety-driver vehicles, and fully driverless vehicles all get lumped into the same headline figure. Coverage that puts a Chinese fleet size next to a US one without checking which convention each side used is comparing two different things.

The pilot-zone structure itself has a longer history and more moving parts than a single Baidu fleet count suggests, and this site's China robotaxi regulatory timeline covers how that structure developed city by city. Current status for Apollo Go and its counterparts, alongside the US programs it's most often compared against, sits on the deployment tracker.

Apollo Go's ambitions haven't stayed confined to China, either. Baidu has pursued pilot programs and partnership discussions in markets outside the mainland, treating international expansion as a longer-term extension of the same platform rather than a separate business. Whether any of those international pilots reach the same operational scale as the domestic pilot-zone fleets is a separate question from whether the underlying platform is ready, since international expansion adds an entirely new layer of foreign regulatory approval on top of whatever technical readiness Baidu has already demonstrated at home.

Wuhan wasn't obvious. It isn't the city most international coverage of Chinese tech tends to default to, which is itself a small data point about how permitting geography and headline geography don't always line up.