Tesla Enters the Scene, BYD Provides Support: It's Time to Charge for Intelligent Driving
The world's most expensive technological gamble is about to enter its most crucial arena.
In late May, Tesla announced that Supervised Full Self-Driving (FSD) would be pushed to Chinese users after approval. Perhaps the most important thing for FSD to prove in China is not that its autonomous driving technology can outperform Chinese competitors, but whether its highly valued business model can work in markets outside the United States: selling cars is just a means to obtain data, and the real business is to get users to pay for autonomous driving software.
In the past three years, the Chinese intelligent driving industry has been fighting a different battle: intelligent driving is not an independent product but an auxiliary function of cars. Its value is reflected in sales volume, market share, and brand premium, rather than in separate software revenue.
Chinese consumers have never paid separately for intelligent driving. To maintain FSD's valuation model, Tesla has to overturn this user mindset.
Coincidentally, on the eve of FSD's entry into China, BYD, which promotes "equal access to intelligent driving" and standardizes auxiliary driving functions for free on 70,000-yuan models, is also trying to break through this hard nut.
On May 28th, at a press conference themed on intelligentization, BYD Chairman Wang Chuanfu proposed that urban navigation is an important safety configuration and should not be tied to the car price. He announced that all BYD models can be optionally equipped with the "Heavenly Eye B" auxiliary driving laser version, priced at 12,000 yuan, and promised to provide one-year safety guarantee for the urban navigation function.
The two new energy vehicle companies with the largest global sales have reached a tacit agreement at this time: they must regain the pricing power of intelligent driving.
The Free Trap
Compared with selling cars, software subscriptions can provide automakers with continuous and stable recurring revenue.
Early on, FSD mainly adopted the one-time purchase system. The one-time purchase price in the US market soared to $15,000, and the price in the Chinese market was as high as 64,000 yuan. However, the high optional purchase price kept most users away. So in 2024, Tesla gradually shifted to the subscription system, reducing the FSD subscription fee in the US market from $199 per month to $99. In February this year, it also canceled the one-time purchase option in most markets.
According to the first-quarter financial report of 2026, based on the monthly subscription fee of $99 and 1.28 million subscribers, FSD contributed about $380 million in revenue to Tesla this quarter, accounting for 1.7% of the total revenue. Although the proportion is not high, as software, FSD's gross profit margin of over 80% or even 90% far exceeds that of the vehicle sales business, which has fallen below 20% in the same period. At the same time, the number of FSD subscribers is rapidly increasing. The number of active users in the first quarter reached 1.28 million, a 16.4% increase from the previous quarter and a 51% increase year-on-year.
By the end of 2025, Tesla had produced about 9 million vehicles, and the delivery volume in the past three years has been between 1.5 million and 2 million. If the number of FSD subscribers continues to increase, combined with the upcoming Robotaxi business, FSD is likely to become another cash cow for Tesla - a profit engine with a gross profit margin that crushes the vehicle business and automatically grows as the fleet size expands.
However, Chinese consumers are not used to paying for intelligent driving. This is not because they do not accept autonomous driving. On the contrary, the Chinese market has a much higher acceptance of high-level intelligent driving than the US. A cross - national survey in PwC's "Digital Automotive Report" shows that 60% to 70% of US and German respondents said they "don't trust" high - level autonomous driving technology, while the proportion of Chinese consumers with the same attitude is only 15%.
Some have tried Tesla's software payment model but failed. In the first quarter of 2021, XPeng disclosed a cumulative intelligent driving software revenue of 80 million yuan, accounting for about 2.5% of the total revenue at that time. The payment rate of its P7 model once reached 20%, which was even higher than Tesla's level in the first quarter of 2026.
After that, XPeng has never publicly announced relevant software revenue data again. Instead, it has included core intelligent driving capabilities such as highway NOA and urban NOA in the vehicle price as standard for all models, which is also the practice of most automakers at present.
Another long - term free method is called "time - limited discount". The one - time purchase price of Huawei's ADS high - level function package is 36,000 yuan, but it has long maintained a discount of over 10,000 yuan or more. In actual sales, automakers and dealers often subsidize car owners through purchase rights and interests to reduce the price in disguise.
BYD and other automakers' "equal access to intelligent driving" campaign has promoted the popularization of intelligent driving functions, but at the same time, it has also reduced the value of intelligent driving as a single product to zero. The user mindset of getting intelligent driving for free when buying a car has gradually solidified during the three - year price war.
This is the strategic trap that Chinese automakers have collectively fallen into: they have spent a lot of money on research and development, only to find that they are making a standard accessory without pricing power.
The Signal
In the same month when FSD announced its entry into China, a meaningful signal appeared in the market.
On May 28th, BYD Chairman Wang Chuanfu announced two things at the intelligentization strategy press conference: First, all BYD models can be optionally equipped with the "Heavenly Eye B" auxiliary driving laser version, priced at 12,000 yuan; Second, BYD provides one - year safety guarantee for the urban navigation function. Once an accident occurs during the use of urban navigation, BYD will bear the compensation liability.
Urban navigation is an important safety configuration and should not be tied to the car price. This is the keynote that Wang Chuanfu set for this press conference.
On the surface, this is a continuation of BYD's "equal access to intelligent driving" strategy. But from another perspective, it is also a quiet loosening of the "free intelligent driving" strategy: the "Heavenly Eye C" is still standard and free, but the "Heavenly Eye B" that covers urban navigation starts to let users pay 12,000 yuan separately.
The logic behind this is not difficult to understand. According to data from the Ministry of Industry and Information Technology, from January to February 2026, the penetration rate of new passenger cars with L2 - level combined driving assistance functions reached 69.15%. The market has been ripened by free intelligent driving, and users' usage habits have been established. Now, it's time to charge.
First offer for free, then charge. Create dependence through experience, and then establish pricing power through dependence. This is the path that the Chinese Internet has taken. BYD is now on this path too.
There is also an opportunity for Tesla here. The prerequisite is that FSD performs well enough in China.
However, FSD's excellent performance cannot be perfectly replicated in markets outside the United States. In Australia, where there are many roundabouts, local media CarExpert found in tests that FSD failed to yield to motorcycles entering roundabouts, the turn signals responded slowly, and it even tried to turn right in the left - turn lane; in the Netherlands, users reported that FSD would ignore the set route and forcefully navigate to the highway, or start to react only after missing an intersection, and the speed limit recognition showed 30 km on a 50 - km section.
Chinese roads are more complex than those in Australia and the Netherlands: dense electric bicycle flows, highly competitive intersection cultures, and different traffic signs in each province - these scenarios that hardly exist in North American training data require FSD to start learning from scratch.
However, the evolution speed of FSD is restricted. China does not allow Tesla to transfer real - road video data collected from Chinese car owners overseas; the United States restricts Tesla from transferring advanced AI training computing power (such as top - level GPUs) to China to establish a local computing center. This makes the Chinese version of FSD have to "learn from videos" as Elon Musk said.
In a market where people are just starting to learn to pay for intelligent driving, product experience is the prerequisite for everything.
Whether FSD's actual test performance in China can make users willing to pay will determine whether FSD's entry into China will replicate its success in the United States or be redefined by the Chinese market.
This article is from the WeChat official account "Yunjian Insight". The author is Liu Yimo, and the editor is Wang Hailu. It is published by 36Kr with authorization.