Pony.ai’s Commercialization B-Side: The Reality of Short Orders and Breaking Through by Building Joint Fleets
```
After a long period of technical validation, leading companies in the autonomous driving industry are now presenting their commercialization and mass production achievements to the market.
Financial reports published by Pony.ai show that revenue in 2025 will reach $90 million, an increase of 20% year-on-year; the loss for the same period will be $76.8 million, an over 70% reduction compared to last year.
The most noteworthy aspect of this performance report is the progress of autonomous ride services (Robotaxi).
In 2025, Pony.ai’s Robotaxi will generate revenue of $16.6 million, a year-on-year surge of 128.6%.
The main growth driver comes from increased passenger fare contributions due to fleet expansion. In Q4 2025, Pony.ai Robotaxi’s passenger fare revenue increased by more than 500% year-on-year.
Specifically, Pony.ai achieved city-wide positive single-car profitability (UE) milestone in Guangzhou and Shenzhen.
On March 22, 2026, the seventh-generation autonomous taxi in Shenzhen set a new record with daily net income per vehicle reaching 394 RMB, with 25 orders per vehicle that day.
According to All-weather Technology’s calculation based on daily income and orders, the average passenger price in Shenzhen was about 15.76 RMB per trip. Considering Shenzhen’s starting fare of 10 RMB and mileage charge of 2.7 RMB/km, this indicates current operations are still limited to short-distance trips.
In response, Pony.ai CFO Wang Haojun admitted to All-weather Technology that current operations in Shenzhen mainly consist of short trips. This is mainly due to operation areas being concentrated in Bao’an and Nanshan districts. However, as more urban areas in Shenzhen and Guangzhou open up this year, it is expected that the order structure will shift to a mix of short and long orders.
As mileage increases, the average miles per intervention (MPI) data is drawing attention.
For example, at the end of last year, a Model 3 equipped with FSD v14 departed from Los Angeles on the US west coast, crossed the continent, and reached the east coast in South Carolina in 2 days and 20 hours. The entire journey covered 2,732 miles, relied 100% on FSD, included highways, city roads, night driving, and multiple entries/exits at superchargers, and did not require any human intervention, sparking much discussion in the market.
Jim Fan, head of Nvidia’s robotics business, even exclaimed: "Tesla FSD v14 may have already passed the 'Physical Turing Test’."
But according to Wang Haojun, MPI is not applicable for L4 stage.
"In fact, when we reach large-scale L4 operations, people no longer mention MPI, it’s no longer relevant. Because there’s no human driver involved, intervention isn’t an issue. The main focus for L4 operations is still fleet scale— the larger the scale, the lower the accident rate. Additionally, attention needs to be paid to the proportion of remote assistance," Wang Haojun noted.
Wang Haojun further pointed out that, in fact, observation shows companies like Waymo no longer emphasize MPI. Many L2+ companies, however, still mention MPI in their push toward L4. When evaluating current status, the key to scaling L4 operations is fleet size and remote assistance ratio.
Looking ahead, Pony.ai has set a goal of deploying over 3,000 autonomous taxis in more than 20 cities globally by the end of 2026.
Relying solely on heavy asset investment in self-operated fleets for such scale is obviously a bottomless pit for cash flow.
Pony.ai’s solution is to simultaneously expand cities and co-build fleets with third parties.
For city expansion, Pony.ai plans to continue densifying domestic tier-one cities, and expand to new tier-one cities such as Hangzhou and Changsha.
Under the "co-built fleet" model, Pony.ai essentially transfers heavy asset vehicle purchase costs downstream to partners. Companies like Ruqi Mobility purchase vehicles and share operating income, while Pony.ai steps into the background and generates revenue by licensing its AI autonomous driving technology.
As cooperation under this model only started in Q3 last year, the number of vehicles in operation is still small. Wang Haojun expects that in the second half of 2026, with gradual introduction of Robotaxi co-built with third parties, the model will contribute more revenue.
However, overall expansion speed still depends on local policy opening pace.
Currently, China’s urban operations do not have large-scale mutual recognition mechanisms, meaning Robotaxi companies must go through stepwise progress—from safety driver road testing to fully driverless commercial operations—each time they enter a new city.
The speed of overseas policy advancement is similar.
Recently, Waymo Co-CEO Tekedra Mawakana stated in an interview that in some cases Waymo can complete the mapping-to-paid ride process in a city in just a few months. In other cases, progress is much slower, especially in cities or states lacking Robotaxi regulatory rules.
Overall, domestic Robotaxi competition is still mainly focused on deploying as many vehicles as possible to gain first-mover advantage.
As the industry leader, Waymo has already entered the stage of order-volume competition, planning to achieve over 1 million paid Robotaxi rides per week in the US market by end-2026.
In the new Robotaxi race, technology is no longer the only moat. Whoever can first scale up orders and achieve a closed commercial loop will truly remain at the table.
Risk Warning and DisclaimerThe market involves risks; investments require caution. This article does not constitute personal investment advice and does not take into account individual users’ specific investment goals, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article suit their individual situation. Investment is at your own risk. ```