The robotaxi operator has even achieved single-vehicle profitability in Guangzhou and Shenzhen, but high R&D and depreciation costs are undermining its high valuation

image credit: Bamboo Works

Key Takeaways:

  • Pony AI’s robotaxi revenue surged more than sixfold in the first half of the year, as it races towards a year-end target of more than 3,500 vehicles.
  • Despite achieving single-vehicle profitability in some markets, the company’s net loss continues to widen

After years of slow advances, robotaxi commercialization has accelerated significantly this year in China. The leading trio of Pony AI, WeRide and Baidu’s Apollo Go have been putting more vehicles on the road, extending their reach to more cities. Individual fleets are crossing into the thousands of vehicles, lighting a fire under company order volumes and revenues.

Leading that charge is Pony AI Inc. (NASDAQ:PONY) (2026.HK), whose latest financial results, released last week, show its revenue nearly doubled year-over-year in the first half of 2026 to $70.47 million. Within that, revenue from its robotaxi services surged by 534% to $20.64 million from just $3.26 million a year earlier, driving its contribution of the company’s revenue pie from 9.2% to 29.3%. The growth accelerated throughout the period, with the passenger fare growth rate rising from 456.5% in the first quarter to 849.3% in the second.

Among its peers, Pony AI is notable for its rapid expansion. WeRide’s (NASDAQ:WRD) (0800.HK) revenue in the first half increased by 73.3% year-over-year to 346 million yuan ($51 million), behind Pony AI. Both companies are still losing significant money, including a first-half net loss of $98.86 million for Pony AI, and a similar-sized 789 million yuan ($116 million) loss for WeRide. WeRide’s overall gross margin stands at 36.6%, notably higher than Pony AI’s 16.9%. But WeRide’s mix includes L2, L3 and L4 autonomous driving businesses in addition to its robotaxi operation, making the margins a bit of an apples-to-oranges comparison.

Baidu (9888.HK) (NASDAQ:BIDU) doesn’t disclose revenue and profitability metrics for its Apollo Go service. But it said the platform already completed 3.2 million fully unmanned driving orders in this year’s first quarter, up over 120% year-on-year, with orders for a single week peaking at more than 350,000.

Pony AI’s global robotaxi fleet reached 1,975 vehicles by the end of June, with plans to boost that figure sharply to more than 3,500 vehicles by year-end. It has more than 1.5 million registered users in China, and its average weekly paid orders in May were more than double the rate in January. More importantly, the company said that Guangzhou and Shenzhen have achieved single-vehicle profitability on a citywide scale. But achieving single-vehicle profitability is still far cry from overall profitability.

R&D expenses of $104.1 million further weighed on the company’s profitability, outpacing its total revenue during the six-month period, while its cash used in operating activities also rose to $118.2 million. CFO Wang Haojun recently said that, based on company calculations, Pony AI will only be able to reach positive cash flow when 40,000 to 50,000 of its robotaxis are deployed domestically in tier-one and tier-two cities. That means that meeting its target of 3,500 vehicles by year-end will still only amount to less than 10% of that threshold.

Regulation remains a wild card on the road to the larger fleets necessary for sustainable operations. Rumors circulated earlier this year that approval of new licenses was suspended after an incident involving Baidu’s Apollo Go service in the spring. But state media subsequently denied a "comprehensive suspension," and Pony AI also said its business wasn’t affected. Judging from its second-quarter performance, the incident hasn’t significantly slowed the company’s pace of expansion.

WeRide disclosed that average daily orders per vehicle for its robotaxis in China exceeded 21 during the second quarter, up 24% quarter-over-quarter, with a peak of 28 orders. Pony AI doesn’t give data for that metric. However, as fleets grow from thousands to tens of thousands of vehicles, revenue generated per vehicle will become increasingly important to dilute depreciation and operating costs.

Co-built fleets and overseas expansion

Expanding a fleet to tens of thousands of vehicles requires huge capital expense for a company using a self-operated service model. Pony AI previously disclosed that vehicle depreciation accounts for about half of its total costs, and scaling up will further push up vehicle purchasing and maintenance costs.

To defray some of those costs, the company is increasingly turning to a "co-built fleet model," where partners such as ride-hailing platforms and taxi companies own and operate the vehicles, while Pony AI provides autonomous driving technology and generates revenue from vehicle sales, "virtual driver" services, and fare sharing. The company said revenue from the co-built model achieved significant sequential growth in the second quarter.

Pony AI is turning to a similar strategy overseas, as it plans to deploy over 2,000 robotaxis across five European cities in a partnership with Uber (NYSE:UBER), bringing the committed number of vehicles for overseas partnerships to more than 4,000. The model will make it easier for Pony AI to expand its fleet while better controlling its cash burn.

Despite the ongoing expansion, investors have been starting to question high valuations awarded to robotaxi operators at the time of their listings. Pony AI’s Hong Kong stock dropped more than 5% the day after its results announcement to close at HK$58.60, down by more than half from its offering price of HK$139 last November. Shares of WeRide, which debuted the same day, also currently trade about 40% below their issue price. In terms of price-to-sales (P/S) ratios, Pony AI’s multiple has fallen from approximately 100 times at the time of listing to 27 times now, while WeRide’s has dropped from 62 times to about 17.

The fact that the P/S ratios are both down by over 70% shows their IPO pricing last year incorporated aggressive expectations for their robotaxi commercialization, and now investors are re-evaluating those expectations. The declining ratio also partly stems from the rapid rise in each company’s revenue. But Pony AI’s stock is now down by nearly 60% compared to its IPO price, indicating that high-speed revenue growth is not yet sufficient to support expectations at the time of its listing.

Even after its stock decline, Pony AI’s P/S ratio of 27 times is still 60% higher than WeRide’s 17, indicating investors are still more bullish on the former for its faster robotaxi growth, aggressive fleet targets, and single-vehicle profitability in Guangzhou and Shenzhen. Now, they will be watching to see if its co-built fleet model can reduce its cash burn and drive improvements in utilization rates, gross margins, and cash flow. Positive developments on those fronts could be cause for some upside to its stock.

To subscribe to Bamboo Works weekly free newsletter, click here

Benzinga Disclaimer: This article is from an unpaid external contributor. It does not represent Benzinga’s reporting and has not been edited for content or accuracy.