On February 25, Tesla rolled out its full self-driving feature through a software update. The company labeled this update as “City Streets Autopilot.” However, the feature closely matches the FSD in North America. This marks a new phase for Tesla’s smart driving strategy in China.
This update includes the following features: City road navigation: The vehicle can navigate controlled roads and city streets. It can automatically go straight, turn, and make U-turns. It can also change lanes independently. Traffic signal recognition: The system recognizes traffic lights and stop signs. It responds accordingly. This feature was previously limited to FSD users. Non-navigation path planning: When no route is set, the vehicle can choose the optimal path based on real-time traffic conditions. Users must meet two conditions to use these features: they must purchase the FSD package and have HW4.0 hardware in their vehicle.
This year, smart driving has gained popularity due to vehicle intelligence and AI models. Many car companies see smart driving as a key to success. On February 10, BYD announced “smart driving equality,” lowering the high-level smart driving system “Heavenly Eye” to models priced at 70,000 yuan (9660$). Many brands and models in the “Huawei system,” including HarmonyOS, upgraded to the ADS 3.0 system and received market validation. New players like Xpeng and Li Auto are capturing market share with their urban NOA features. Tesla’s smart driving features generate excitement in the domestic market, but they also face significant pressure and competition. The market environment differs greatly from a few years ago. Regarding localization, Tesla’s FSD launch in China raises several questions. First, does FSD require a light touch on the steering wheel, or can drivers let go? The official stance is unclear. Domestic regulations prohibit letting go, while foreign regulations allow it. We must wait for domestic laws to change. Second, FSD relies on a pure vision approach. With the urban NOA now available, can it match the performance of Huawei’s lidar + vision system ADS 3.0? We need real-world testing. Currently, only two companies offer native pure vision urban NOA in China: Tesla’s FSD and Jidu’s ASD. Whether the latter can sustain OTA upgrades remains to be seen. Third, after launching FSD in China, will visual training data upload to overseas servers for continuous learning? How will data security be ensured? During Tesla’s Q4 2024 earnings call, Musk shared challenges faced when launching FSD in China, stating, “We face challenges because they do not allow us to transfer training videos outside China. The U.S. government also does not permit training in China. We are in a bit of a bind.” Thus, the FSD feature’s launch in China will likely use local training data and will not transfer to overseas servers for Dojo training.
Tesla FSD currently uses a buyout model for subscriptions in China. The price of 64,000 yuan (8830$) makes many people hesitate. Future adjustments remain uncertain. Tesla’s entry into China represents not just a tech competition but also a struggle between globalization and localization. In the short term, its high pricing may limit the user base. However, Tesla’s advantages in algorithm updates and brand influence could reshape the smart driving landscape in China. This could become a pivotal event in the second half of the domestic smart driving sector.