Today, Tesla's AI team released a product roadmap, in which it is expected to launch fully autonomous driving (FSD) in China and Europe in the first quarter of 2025, but regulatory approval is still pending.
Since 2016, Musk has been exploring Tesla's FSD autonomous driving solution. In 2024, Tesla FSD V12 will be launched and widely promoted in North America. FSD V12 autonomous driving has been upgraded to an "end-to-end" neural network from perception to vehicle control, with training computing power and data becoming the core barriers.
At present, the FSD iteration is steadily advancing, with the latest version 12.5.2 increasing its parameters by 5 times compared to 12.4, merging highway and urban technology stacks, and further improving driving fluency. Specific improvements include more natural and earlier lane change decisions, using only visual systems for driver monitoring, end-to-end driving on highways, and FSD functionality on Cybertrucks.
At the same time, Tesla insists on investing in the construction of AI supercomputing centers to meet its computing power needs for autonomous driving training. As of 24Q2, Tesla's self built supercomputing center has AI computing power exceeding 40000 Nvidia H100 equivalent computing power, and is expected to increase to nearly 90000 by the end of the year.
Musk previously expected that Tesla's FSD testing mileage would need to reach 6 billion miles to meet global regulatory requirements. According to Tesla's latest financial report, as of the end of Q2, the cumulative mileage of FSD exceeded 1.6 billion miles. The company also stated that the non intervention driving mileage of FSD will exceed that of human drivers by the end of the year, and is confident in achieving unsupervised L5 level intelligent driving next year.
Since the beginning of this year, Tesla has repeatedly lowered its FSD pricing to attract customers. China is an important market for Tesla cars, but the timing of FSD entering China continues to be postponed.
Recently, Musk stated during Tesla's financial conference call in July that FSD is expected to enter China by the end of the year. FSD will be launched in China, Europe, and other countries after version 12.6 is released, and the unsupervised version is expected to be commercially implemented in sync with Robotaxi.
At that time, he stated that Tesla's autonomous driving technology would not encounter regulatory barriers when promoted in the United States and other regions.
Although Tesla FSD has not yet been launched in the Chinese market, it was already sold with the Model 3 when it was released, and a subscription system may be introduced in the future.
Guosen Securities previously stated that in the long run, Tesla is expected to create a series of artificial intelligence terminal carriers (smart driving cars, robots, etc.), empowering new profit models through "hardware cost reduction+software markup+Robotaxi shared travel". By combining FSD software, software mall, and subscription services, we aim to increase added value and open up intelligent driving application scenarios for shared travel.
Ping An Securities further stated that the entry of FSD into China may become a "catfish" in the domestic autonomous driving field, accelerating the landing process of advanced intelligent driving in China, and leading car companies or suppliers in the field of advanced intelligent driving will benefit more.
Understand Tesla FSD
Tesla FSD iterated from the initial version to V12, and its technological innovation mainly lies in the restructuring of the architecture - the introduction of end-to-end architecture. Meanwhile, the continuous iteration of Tesla's FSD system marks the beginning of a data-driven period of rapid development for its autonomous driving technology.
The advantages of Tesla FSD are reflected in: 1. Algorithm: It achieves deep integration of machine learning throughout the entire autonomous driving process; 2. Perception: It can perform more driving actions that were previously difficult to accomplish, such as planning routes based on real-time road conditions rather than relying solely on maps; 3. Computing power: Tesla has significantly increased its investment in computing power, such as using thousands or even tens of thousands of high-performance H100 graphics cards. Tesla also plans to increase the number of GPUs on its basic training platform by two to three times, which will greatly enhance data processing capabilities and training.
Currently, Tesla FSD is implementing a series of promotional policies in the US market. According to Tesla's software tracking website TeslaFi, as of April 26th, out of 19759 Tesla vehicles registered by its paying users, 7444 were installed with FSD V12 version and 9724 were installed with FSD V11 version. The highest installation rates for FSD V12.3.4 and FSD V11.4.9 versions were 6149 and 9381, respectively, with penetration rates of 31.1% and 47.5%.