WeRide (WRD) Robotaxi Rides per Vehicle Reach 21 a Day in China

WeRide (WRD) said on its 2026-08-12 earnings call that its autonomous virtual driver reached 21 average daily rides per vehicle in China during the quarter, up 24% quarter over quarter.[1] The company did not disclose revenue for this application on its own.

What WeRide's virtual driver is

WeRide sells autonomous driving systems installed on vehicles. The product line covers robotaxis, robobuses, robosweepers, robovans and assisted-driving systems supplied to automakers. The virtual driver is an AI system that runs on the vehicle: cameras and lidar turn road conditions into data, and an end-to-end model decides in real time how the vehicle drives and which path it takes, replacing the human driver's actions.[4] The users are the passengers hailing a ride and the party operating the fleet, and a trip runs from order acceptance to passenger drop-off. Part of the revenue comes from passenger service on WeRide's own fleet, and part comes from licensing the same driving capability to local partners who operate their own fleets, billed as technology-service fees and mileage fees.[1] This driving capability is the company's core business and supports the entire product line above.

How the disclosure moved from architecture to per-vehicle output

The public record on this application starts on 2025-03-14. On that call the company first spelled out its underlying approach: from the second half of 2024 it deployed end-to-end models across both the assisted-driving and driverless product lines, using a vision-language model pre-trained on internet knowledge as the backbone, then training it on nearly one million hours of human and simulated driving data, and running it in parallel with the existing modular architecture and safety framework. The only results available then were fleet-level cumulative operating days and safety records.[4]

By 2025-05-21, the same driving capability had been packaged as WeRide One, a universal platform spanning five product lines — robotaxi, robobus, robosweeper, robovan and ADAS — with data shared across all five. The measure the company used for itself at that point was breadth of coverage: 30 cities across 10 countries.[3]

On 2026-05-13 the measure landed on the vehicle for the first time: a global robotaxi fleet of about 1,300 vehicles and 17 average daily orders per vehicle in China.[2]

By 2026-08-12 the application had not been renamed and had not been folded into another business. The change was in delivery. The company shifted to outbound licensing, with local partners owning and operating the vehicles while WeRide supplies driving capability validated by local regulators and collects technology-service fees and mileage fees. The self-operated fleet was kept in place.[1]

What changed this quarter

This is the first time the line has been measured by output per vehicle. Domestic average daily rides per vehicle rose from 17 in the prior quarter to 21, up 24% quarter over quarter — an increase in usage intensity on the same measurement basis. In the same quarter the robotaxi fleet grew by 500 vehicles to more than 1,800, and management described 40% growth over the same period.[1]

Fleet expansion normally dilutes output per vehicle. Both numbers rose together this quarter, which indicates the newly deployed vehicles picked up orders on top of the existing capacity. The Middle East fleet stood at about 400 vehicles, covering more than 70% of the core urban areas of Abu Dhabi and Dubai,[1] moving overseas activity from scattered pilots into operations at scale.

The boundaries matter. The figure of 21 refers to domestic self-operated activity; no comparable reading was given for the Middle East or other overseas markets. The 1,300 and more-than-1,800 vehicle counts come from different dates and different measurement bases, and the 40% management cited only indicates a comparison against the same prior period, so it should not be read as a precise sequential change.

Where the operating numbers meet revenue

These operating readings ultimately flow into L4 revenue, which is the company's own disclosed line item and which management said is driven mainly by robotaxi. More trips per vehicle per day, a larger fleet and a wider serviceable urban area together enlarge billable trips and mileage: the self-operated portion shows up as passenger revenue, and the licensed portion shows up as recurring charges billed through technology-service fees and mileage fees.

L4 revenue for the quarter was RMB 125 million, up 47% year over year.[1] It appears in the same quarter as the operating metrics and points in the same direction, and management explicitly attributed the main source of growth to robotaxi.

L4 revenue also includes autonomous driving businesses outside robotaxi. The company did not disclose this application's own revenue amount, its per-vehicle pricing basis or its operating costs, and gave no share of growth attributable to it. The relationship can therefore be confirmed as directional only; the size of the contribution cannot. The company's statement about steady-state annualized technology-service revenue per vehicle is an estimate under assumed conditions — a forward projection, not a basis for backing into revenue for this quarter.

What would settle the question

The operating change that can be confirmed now is that this application has passed the stage of whether the vehicles can drive on public roads and reached the stage of how much a single vehicle produces each day. Daily rides per vehicle, fleet size and serviceable urban area all rose together, and the revenue line shows a reading pointing the same way. What remains separately unquantified is the application's own economics: robotaxi revenue, per-vehicle depreciation and operating costs all still sit inside broader line items. Only if the company breaks robotaxi revenue out of the L4 line, or discloses per-vehicle pricing and per-vehicle operating cost for this business, can anyone test whether rising daily rides per vehicle actually turns into higher revenue per vehicle.

Application assessment

• WeRide Autonomous Virtual Driver | Business position: Core business | Deployment stage: Limited production | Scope: Multiple businesses or regions | Value type: Revenue growth

Sources

[1] Drillr · WeRide Inc. (WRD) · 2026-08-12 · Earnings call

Original: Average daily rides per vehicle reached 21, up by 24% quarter over quarter. While peak daily rides completed per vehicle climbed to 28, As a reading-healing platform with self-operated Robotext fleet, we recorded nearly 35% quarter-over-quarter growth in registered users in China.

[2] Drillr · WeRide Inc. (WRD) · 2026-05-13 · Earnings call

Original: On the demand side, average daily order per vehicle domestically reached 17 trips during Q1, with peak periods reaching 28 trips per vehicle.

[3] Drillr · WeRide Inc. (WRD) · 2025-05-21 · Earnings call

Original: over the past eight years, we built up a WeRide One universal platform that can be deployed to use for RoboTaxi, Robobus, RoboSweeper, Robovan, and ADAS system.

[4] Drillr · WeRide Inc. (WRD) · 2025-03-14 · Earnings call

Original: The end-to-end model employs a large VLM, i.e. visual language model, which is pre-trained with all the data available from the internet as the backbone to encode the world's knowledge. It is then trained with nearly one million hours of carefully selected driving data from both human drivers and simulated driving scenarios based on our world model.