Robotaxis, which are more expensive than ride-hailing services, are actually making money.
Do you know Pony.ai?
Those driverless taxis running on the streets. You may have seen them on the streets of Guangzhou and Shenzhen. In the industry, they're called Robotaxis. They've reached Level 4, and basically don't need human supervision.
01
It was founded in 2016, went public on the US stock market at the end of 2024, and also listed on the Hong Kong stock market this year. It's the first in the domestic market in this field to be listed in both places.
Currently, the fleet size has exceeded 1,700 vehicles, and they're really operating on the streets. The cash reserve on the books is 9.9 billion yuan, so there's still a relatively sufficient "ammunition".
On May 26th, the Q1 financial report came out. I looked through the figures: The single - quarter revenue was 236 million yuan, a 145% increase compared to the same period last year. Just the Robotaxi segment alone contributed 59.12 million yuan, almost four times that of the same period last year.
What does this mean?
The Robotaxi revenue in this quarter has already exceeded half of the whole year of 2025. The company has directly set a new goal: to increase the fleet size from 3,000 to over 3,500 vehicles.
The figures are really impressive, but what actually caught my attention isn't this.
You know, on May 26th last year, on the same day, Pony.ai went through something completely different - the shares of pre - IPO shareholders were unlocked. Before the news was officially confirmed, the stock price had already dropped by 15%.
The founders, Peng Jun and Lou Tiancheng, signed an agreement at that time, voluntarily extending the lock - up period by 540 days. More than 70% of the shareholders also said: "No rush, we won't sell."
Just think about the atmosphere in the market that day. Everyone had only one question in mind: How much more money will this company burn? When will it stop?
540 days later, on the same day, the revenue more than doubled. What exactly changed in the one - year interval?
I looked through the details of the financial report. An interesting thing is:
Pony.ai has achieved positive profitability in single - vehicle operations in Guangzhou and Shenzhen successively. In the industry, this is called achieving positive Unit Economics (UE). Normally, this should mean that the cost has finally dropped lower than that of ride - hailing services.
But the actual situation is not like this. After discounts, the pricing of Robotaxis is higher than that of entry - level ride - hailing services, but the user demand is still strong. That is to say, even though the price is higher than that of ride - hailing services, the business is even better.
02
When talking about Robotaxis, everyone is actually calculating the same thing.
How much has the cost per vehicle dropped? How much has the Bill of Materials (BOM) been cut? When will the per - kilometer operating cost catch up with that of ride - hailing services? The analysis framework for the entire industry boils down to one word: save.
For Robotaxis to succeed, they have to be cheaper than human drivers. When the cost comes down, the business will work. This line of thinking isn't completely unreasonable.
Take a look at the early state of Robotaxis. After modifying a vehicle, the cost could easily exceed one million yuan. Sensors, computing power platforms, and safety redundancies were all top - of - the - line. At that time, the cost of a Robotaxi could buy five or six ordinary ride - hailing vehicles. How could it possibly make money?
So, in the past few years, everyone has been focusing on one thing: When will the cost come down?
It is indeed coming down. The BOM cost of Pony.ai's seventh - generation autonomous driving kit has been cut by 70% compared to the previous generation. The target price for the whole vehicle in 2027 is to be kept below 230,000 yuan, including the vehicle, the battery, and the autonomous driving kit, which is even cheaper than the lowest - end domestic Model 3.
The sixth - generation single vehicle of Baidu's Apollo Go has also reached a price of just over 200,000 yuan. From over one million to just over 200,000 yuan, it has been cut by more than three - quarters in a few years. This progress is real.
So, everyone continues to calculate along this line: When will it reach 150,000 yuan? When will it reach 100,000 yuan? When will the per - kilometer cost be lower than that of human drivers?
It makes sense, right? But the achievement of positive UE didn't wait until that day.
I looked through the timeline. Guangzhou achieved positive UE in November 2025, and Shenzhen in February 2026. At that time, the cost was indeed dropping, but the pricing of Robotaxis didn't drop below that of ride - hailing services.
In April, Jiemian News interviewed CFO Wang Haojun. He was very specific: The pricing is based on Express and Premium services.
Express and Premium are the mid - to high - end levels of ride - hailing services, not the lowest - end economy level. The unit price in Shenzhen is about 10% higher than that in Guangzhou, and it still achieved positive UE.
Looking at some specific figures, the daily net income per vehicle in Guangzhou is 338 yuan. This is the real money after deducting all discounts and promotions. In Shenzhen, it's even more impressive. On March 22nd, the daily net income per vehicle reached 394 yuan, and each vehicle received an average of 25 orders that day.
Wang Haojun also mentioned a detail: An important factor supporting this income level is a large number of repeat users. What does this mean? Users are really taking the Robotaxis repeatedly.
Putting these things together: The price is higher than that of entry - level ride - hailing services, but users not only don't leave but also keep placing orders. During the May Day holiday, the daily average number of paid orders increased by 544% year - on - year.
If you insist on using the cost framework to explain everything, this timeline doesn't make sense. I think the entire industry has been waiting for an answer: When will it be cheaper than ride - hailing services? Maybe this question itself is off - track from the very beginning.
03
At this point, some people may think: Is the positive UE achieved by subsidies? Or did the policy provide some good channels?
I've also thought about this.
First, look at subsidies. The 338 - yuan daily net income that Wang Haojun mentioned was "after deducting all discounts and promotions". The peak of 394 yuan was also net income. It's not achieved by burning money.
Then, look at the policy. Guangzhou and Shenzhen are indeed more advanced in terms of policies, providing operating licenses and open areas. But the policy only gives the permission to "be on the road", not the guarantee that "users will pay". Road rights don't equal customer sources.
After using the process of elimination, there's a question that can't be avoided: What exactly do users want?
Pony.ai's own answer is two words: privacy and in - car environment. To be honest, when I first saw this answer, I thought it was a bit vague. Really?
Later, I looked at the data from the US.
There's a platform called Obi that does something special. It specifically compares the prices of various ride - hailing services. Since last year, they've been monitoring the pricing of Robotaxis and how users use them. So far, they've released three rounds of reports, looking at more than 90,000 real ride - hailing records in total, and also asking about the feelings of more than 2,000 people.
What does Obi's data say? It's quite interesting.
Waymo's pricing in San Francisco is higher than Uber's and even higher than Lyft's. In the initial stage, the premium was 30% to 40%. Recently, the gap has been narrowing. For medium - distance trips, it's only about 2% more expensive than Uber.
But guess what?
Even at the highest premium, 70% of the people who have taken Waymo said that they just prefer driverless cars. They're not taking it because they have no other choice, but really choose it actively.
Obi's survey asked a very direct question: How much more are you willing to pay for autonomous driving? 42.7% of the people said they were willing, and the maximum they could accept was an extra 10 dollars.
Looking at the change in attitude, in the spring of 2025, only 35% of the people said they were "comfortable" with autonomous driving travel. By January 2026, this figure had risen to 63%. In less than a year, the acceptance rate doubled.
Nearly half of the people said that autonomous driving may be their main mode of travel in the future.
So, why are users willing to pay more?
Obi asked the users, and they all mentioned very specific things: Don't have to worry about the driver's fatigue driving, don't be afraid of being taken on a detour, and don't have to make small talk. There are no strangers staring at you in the car. Especially for girls, the fact that "there are no strangers in the car" is worth the premium itself.
Do you notice? These things are completely different from the dimension of "being cheap or not".
What users want is that after the driver disappears, the act of traveling itself becomes different. The sense of security is certain, the experience is consistent, and every time you get in the car, you know what to expect.
Obi's CEO said something that I think is quite on point: "Tesla is setting the price floor, Waymo is testing premium pricing, and everyone else is being forced to respond."
What does this indicate?
The global Robotaxi market is splitting into two paths. One is to go for cheapness and grab market share with low prices. The other is to focus on experience and support the premium with differentiation.
Looking back at Pony.ai, what the CFO said about "pricing based on Express and Premium", "privacy and in - car environment", and "a large number of repeat users" is completely in line with what Waymo has verified in the US market.
This is data from two different markets (China and the US), two different companies, and two independent sets of data, all pointing to the same conclusion.
I think the question that everyone has been asking: When will Robotaxis be cheaper than ride - hailing services, has narrowed down the issue from the root.
However, I have to say one thing upfront. Obi's data covers the US market. The methodology is solid and the sample size is sufficient. In China, there hasn't been a similar independent third - party conducting such a pricing comparison study yet.
Pony.ai's statement about users being willing to pay a premium comes from the company's financial report and the CFO's public interview. The data from both sides is in the same direction and can corroborate each other.
But to say that this conclusion has been fully proven, there's still one step to go. The Chinese market also needs its own independent data.
04
We can wait for more data, but there's a question that we can think about without waiting for data: Is a Robotaxi just a cheaper taxi, or a new species?
If you look at it from the "substitute" framework, all problems will become cost problems: How much cheaper is it than ride - hailing services? When will it catch up with human drivers? How much can you save per kilometer?
Following this logic, whether it's Robotaxis or the entire driverless field, they're always competing with ride - hailing services for the same group of people and the same market share. The only way to win is to be cheaper.
But if you use a different perspective, things will be different.
If users are paying for things like "no strangers in the car", "no need to make small talk", and "the same experience every time you get in the car", then it's not competing with economy - level ride - hailing services. It's selling something that didn't exist before.
The business logics derived from these two frameworks are very different.
Under the substitute logic, the ceiling is the scale of the existing ride - hailing market. Your only weapon is to lower the price. The more you go on, the more intense the competition will be, and the thinner the profit margins will be.
Under the new - category logic, you have pricing power and don't have to compete with the cheapest segment. You're creating a new consumption scenario, and the profit margin is defined by yourself.
Of course, the story of the new category sounds more appealing, but to be honest, it's too early to conclude that Robotaxis are a new species.
Recently, there's an opinion that I think is worth thinking about seriously. Some analysis points out that the capacity of domestic ride - hailing services is already in excess. Many cities have issued saturation warnings. After the large - scale deployment of Robotaxis, no significant incremental demand has been created.
What does this mean?
Currently, Robotaxis are still competing for the same group of people who used to take ride - hailing services, and haven't attracted non - ride - hailing users. If this is the case, the term "new category" needs to be taken with a grain of salt. No matter how different you are, you're still competing in the same market.
There's also another layer of uncertainty. The current data is based on a fleet size of 1,700 vehicles. By the end of the year, when the fleet expands to 3,500 vehicles, the user structure will definitely change.
The early adopters who were willing to try are naturally more interested in new things and it's not surprising that they're willing to pay a premium. But when the fleet expands and the users shift from early adopters to ordinary people, can this premium still be maintained?
To be honest, I can't answer this question.
I think a more accurate statement is that in the Chinese and US markets, the direction of this signal is clear, but it's not a definite conclusion yet.
Looking back at the 540 - day story.
On May 26th, 2025, the essence of the market panic was a hypothesis: Robotaxis are a money - burning business, and they'll keep losing money if the cost can't come down.
The Q1 financial report 540 days later didn't refute the fact that "the cost is still coming down". The cost is indeed still dropping. But it added a variable that no one had discussed before: Users are willing to pay for "difference".
No one knows how far this variable can go. The 540 - day lock - up period signed by the founders hasn't expired yet, and the window they're betting on is still open.
However, one thing is certain: From now on, when talking about Robotaxis and driverless technology, just calculating the cost isn't enough. We have to look at things from a development perspective.
This article is from the WeChat official account "Wang Zhiyuan". Author: Wang Zhiyuan. Republished by 36Kr with permission.