The recent rise of Baidu’s Apollo Go, known in China as Luobo Kuaipao, has intensified the already competitive ride-hailing market.
According to the latest data released by the Hubei provincial government, the number of rides served by Apollo Go’s autonomous vehicles has surged in Wuhan, exceeding 20 per car daily—almost on par with human taxi drivers.
Baidu’s autonomous ride-hailing platform has sparked heated discussions about whether it will replace ride-hailing drivers, pushing industry competition to new heights.
The Chinese ride-hailing industry has been developing for over a decade. After several subsidy wars, a structure has formed, dominated by various platforms with Didi Chuxing occupying top spot, holding around 70% market share.
It has never been a peaceful market: order volumes may have peaked, but the number of drivers is rapidly increasing. Since last year, several regions have issued capacity saturation warnings, and negative public sentiment about unclean vehicles also erupted at the end of last year.
For ten years, the ride-hailing business has been tough and opaque, with most companies yet to achieve profitability.
Caocao Chuxing’s prospectus lists three major entrenched challenges in the industry: high user costs and poor user experience, heavy driver workload with low income, and the difficulty for ride-hailing platforms to optimize operating costs effectively, making profitability elusive.
Will the emergence of autonomous ride-hailing vehicles change all this?
The dilemma of ride-hailing drivers
In recent years, especially since Didi’s halt, small ride-hailing platforms have seized the opportunity to enter the market. According to regulatory information accessed by 36Kr, by the end of the first quarter of this year, there were 345 ride-hailing platforms nationwide in China.
However, many of these small platforms have capacity but no traffic, making integration with aggregation platforms like Alibaba-owned AutoNavi the best choice.
Caocao and other platforms are no exception. The prospectus shows that, from 2021–2023, the gross transaction value (GTV) of orders Caocao received from aggregation platforms was RMB 3.9 billion (USD 535.9 million), RMB 4.4 billion (USD 604.6 million), and RMB 8.9 billion (USD 1.2 billion), respectively, with the percentage of total GTV rising from 43.8% to 73.2%.
Essentially, aggregation platforms are in the business of reselling orders. For example, a passenger places an order on AutoNavi, which then resells it to various integrated ride-hailing platforms.
This situation places ride-hailing drivers in a dilemma. Aggregation platforms can bring them more orders, but an additional layer of commission means less income for drivers.
Gu Dasong, executive director of the transportation law and development research center at Southeast University in Nanjing, Jiangsu, found in an investigation that when a passenger paid RMB 98.11 (USD 13.4) for a ride, the driver-side platform showed the passenger paid RMB 71.46 (USD 9.8), but the driver’s actual income was only RMB 52.17 (USD 7.1), with a commission rate as high as 46.8%.
Capacity saturation exacerbates the decline in ride-hailing drivers’ income.
According to China’s Ministry of Transport, from 2021–2023, the monthly order volume for ride-hailing remained around 800 million, but the number of driver certificates increased from 2.08 million in May 2020 to 6.572 million in December 2023, a sharp increase of 216%.
Some industry insiders told Caijing that, since some regions have started freezing new driver registrations, new drivers who cannot register legally end up working for small companies and taking orders through aggregation platforms.
On July 7, the Guangzhou Municipal Transportation Bureau released data showing that the average daily income of ride-hailing drivers in Guangzhou hit a new low. From December 2023 to May 2024, the average daily order volume for ride-hailing decreased from 14.21 to 12.22, and the average daily income dropped from RMB 343.34 (USD 47.1) to RMB 311.63 (USD 42.8).
This means some ride-hailing drivers may earn less than RMB 10,000 (USD 1,375) a month even if they work without rest for the entire month, and this is before deducting costs.
The awkward situation in the ride-hailing industry today is that, although ride-hailing drivers have many complaints, companies are also facing profitability challenges.
For example, under the asset-heavy model, Caocao and OnTime have suffered significant losses.
According to the prospectus, from 2021–2023, although Caocao’s revenue grew from RMB 7.153 billion (USD 982.9 million) to RMB 10.668 billion (USD 1.4 billion), and its gross margin improved from -24.4% to 5.8%, it still had a net loss of RMB 1.981 billion (USD 272.2 million) in 2023, with a cumulative loss of nearly RMB 7 billion (USD 961.9 million) over the three-year period.
During the same period, OnTime’s total net loss was about RMB 2 billion (USD 274.8 million). The company expects to remain unprofitable until 2026.
Essentially, in the ride-hailing industry, platforms and drivers share the same pool of income. The amount passengers pay is fixed, so the income of platforms and drivers is inversely related. This means that, to increase revenue, ride-hailing platforms must either increase the order volume or raise the commission rate.
Therefore, the lengthening of working hours for ride-hailing drivers and the reduction in their income is not just due to the rapid increase in driver numbers but also due to the commission rates of the platforms.
Caocao’s prospectus shows that, from 2022–2023, its commission rate increased from 16% to 21%. Over the past three years, OnTime’s ratio of driver service fees to revenue dropped from 93.5% to 77.6%.
Robotaxis are not a major threat (yet)
The industry is collectively anxious, with second- and third-tier ride-hailing platforms flocking to list in Hong Kong to seek rescue.
At the end of June, after three failed IPO attempts, Dida Chuxing finally got listed on the Hong Kong Stock Exchange (HKEX). On July 10, OnTime also went public on the HKEX. Additionally, Caocao submitted its prospectus in May.
Caocao and Dida are national platforms, while OnTime mainly provides services in the Greater Bay Area. Caocao and OnTime have relatively comprehensive businesses, while Dida started offering ride-sharing services earlier than Didi.
However, judging by the already listed platforms, regardless of the business model, the secondary market is currently pessimistic about ride-hailing.
Take Dida, which was listed at the end of June, as an example. Its issue price was HKD 6 per share, but it plunged continuously after listing, recently closing at HKD 2.62, half of the issue price. Similarly, OnTime, which was listed on July 10, fell over 3% on the first day.
Dida has been one of the few profitable shared mobility platforms. Over the past three years, Dida has consistently been profitable, with adjusted net profits of RMB 238 million (USD 32.7 million), RMB 85 million (USD 11.6 million), and RMB 226 million (USD 31 million), respectively, and a stable profit margin above 70%.
The launch of autonomous ride-hailing in cities like Wuhan has heightened concerns about job security, not only among drivers but also in the capital markets, which clearly favor autonomous driving companies.
The rollout of Apollo Go briefly boosted Baidu’s stock price. Initially rebounding from a low of USD 85 in early July to over USD 100 the following week, the stock price subsequently retreated closer to the USD 90 mark.
Zhang Ning, vice president at Pony.ai, told 36Kr that the emergence of robotaxis is driven by the need for new productivity and the rising labor costs in the transportation sector, which increase operational burdens and user travel costs. According to Zhang, autonomous driving can significantly reduce the demand for human drivers, thus compensating for labor shortages while effectively lowering the occurrence of traffic accidents.
However, the notion of autonomous ride-hailing vehicles replacing human drivers still has a long way to go. Behind the operation of unmanned taxis, companies need to invest heavily in scheduling and maintenance systems.
According to Zhang’s estimates, breakeven is attainable in cities like Beijing, Shanghai, Guangzhou, and Shenzhen, once city-wide deployment reaches 1,000 vehicles. Beyond this point, Zheng believes that costs will further decrease with each additional vehicle, increasing gross margins and entering a positive self-reinforcing cycle.
According to Shanghai Observer, the daily cost of each Apollo Go vehicle exceeds RMB 370 (USD 50.8). Even without considering upfront research and vehicle insurance costs, its current pricing model in Wuhan is insufficient to cover operating costs.
Compared to the more costly autonomous driving, the simpler and more direct methods to achieve profitability might still be to focus on commissions, subsidies, and competition.
This article was written by Song Wanxin in Chinese language and was originally published by 36Kr.
