China's auto market is on the eve of the explosion of intelligent driving, and the penetration rate of L2 and above driving functions is expected to soar to 62% in 2025 (according to Canalys). This is not only a milestone in the process of automotive intelligence, but also a deep test for the semiconductor industry. The wave of intelligent driving is changing every link of the automotive industry chain at an unprecedented speed, and semiconductors, as the core of intelligent driving, are standing at the forefront of this change.
1. Technological progress: Transformation from functional to user-friendly.
The maturity of technology is the fundamental driving force for the improvement of the penetration rate of intelligent driving. In 2023, the L2 penetration rate of domestic passenger cars will exceed 40%, and some new energy models such as BYD and Xpeng will be close to 80%, with a CAGR of 25%-30% from 2022 to 2025. The Level 2+ NOA (Navigate on Autopilot) function is becoming the standard for high-end models.
The progress of technology is not only reflected in the enrichment of functions, but also in the improvement of user experience. For example, the application of end-to-end large models enables intelligent driving systems to better cope with complex road conditions. Systems such as Tesla's FSD V12 and Huawei's Qiankun ADS have been trained on massive amounts of data to significantly improve decision-making capabilities. At the same time, the declining cost of hardware is also driving the adoption of the technology. The price of lidar has dropped to about 1,500 yuan, and the starting price of models that support urban NOA has dropped to 150,000 yuan, such as Xpeng MONA M03. This double breakthrough in technology and cost has laid the foundation for the rapid popularization of intelligent driving.
2. Market demand: the migration from "high-end" to "mass market".
Consumers' expectations for intelligent driving are changing from "usable" to "easy to use", transformation, especially in high-speed and urban commuting scenarios. Canalys predicts that the penetration rate of high-speed NOA will reach 10.8% in 2025, and the growth rate of urban NOA will exceed that of high-speed NOA by 2027-2028. This trend shows that the complexity and high-frequency use requirements of urban driving scenarios are becoming the main battlefield of intelligent driving technology.
Changes in market demand are also driving the spread of intelligent driving functions to low- and mid-range models. In 2023, the intelligent driving loading rate of new energy vehicles below 160,000 yuan will continue to grow, and some models have supported urban NOA. Mainstream OEMs, represented by BYD, Geely, Chery and Hongqi, are introducing mid- and high-end ADAS functions into their main models in the price range of 10-200,000 yuan. Through cross-platform driver assistance solutions, these car companies have systematically launched ADAS solutions with clear functional levels and coherent configurations, realizing differentiated functional configurations on the same or multiple SoC platforms, and gradually promoting the standardization of ADAS functions. This strategy not only satisfies the needs of models at different price points for balancing performance and cost, but also accelerates the universalization of intelligent driving technology.
image: Canalys predicts that the penetration rate of L2 and above driving functions in the Chinese market may soar by 62% in 2025
3. Industrial pattern: the game from "differentiation" to "reshaping".
The popularization of intelligent driving technology is reshaping the competitive landscape of the automotive industry. Models with a value of more than 200,000 yuan generally adopt a dual-domain controller architecture for cabin/driving, while in the range of 100,000-200,000 yuan, the integration of cabin and parking and cabin driving and parking is accelerated to balance cost and experience. This differentiation of architecture reflects the differences in the intelligent path of models at different price points.
In the future, the "one-chip" trend will further reduce BOM costs, reduce system latency, and improve cross-domain data collaboration capabilities. However, this trend also faces challenges such as high development complexity, unification of cross-domain standards, and high platform design thresholds. This means that semiconductor manufacturers need to find a new balance between computing power, power consumption and cost, while OEMs need to innovate at a deeper level in functional integration and user experience.
4. The "stubbornness" and "rebirth" of fuel vehicles
Although the penetration rate of new energy vehicles continues to rise, it is expected to be close to 60% in 2025, with annual sales reaching 13.5 million units, but fuel vehicles have not withdrawn from the historical stage. Canalys expects the gasoline vehicle market to remain around 8 million units by 2050. Joint venture brands, as well as traditional Chinese fuel vehicle brands such as Geely and Chery, are gradually promoting the path of "intelligent fuel vehicles", making the fuel vehicle market a potential new driving force for the growth of driving assistance.
This "intelligent" transformation is not only a strategy for fuel vehicles to cope with the competition of new energy vehicles, but also an effort to find a new positioning in the era of intelligent driving. The intelligence of fuel vehicles will provide consumers with more choices and open up new market space for semiconductor manufacturers.
5. Opportunities and Challenges of Semiconductors
The rapid development of intelligent driving has put forward unprecedented requirements for the semiconductor industry. According to the "2025 China ADAS SoC Supplier Leadership Matrix" released by Canalys, seven manufacturers, including NVIDIA, Qualcomm, Horizon, Texas Instruments, Mobileye, Ambarella and SiEngine, were rated as "Champion Manufacturers". These manufacturers have performed well in chip technology research and development, manufacturing, marketing and cooperation with upstream and downstream enterprises, providing strong power support for the development of intelligent driving technology.
However, as the complexity of intelligent driving functions continues to increase, semiconductor manufacturers need to find a new balance between computing power, power consumption, and cost. For example, high-performance computing platforms such as Xpeng's self-developed "Turing chip" and Horizon Journey series chips provide computing power support for L3/L4 autonomous driving. At the same time, car companies are improving safety through redundant drive-by-wire systems. This kind of technological breakthrough requires not only strong R&D capabilities, but also a deep understanding of market demand.
image: Canalys released the 2025 China ADAS SoC Evaluation, and seven manufacturers were awarded "Champions"
6. Conclusion: The sea of stars of intelligent driving
In 2025, China's intelligent driving market will usher in the milestone of 62% penetration rate of L2 and above functions, which is not only the result of technological progress and market demand, but also the beginning of the reshaping of the industrial pattern. For the semiconductor industry, this is both an opportunity and a profound change. In this era of intelligent driving, whoever can take the lead in breaking through the double barriers of technology and market will be able to take the lead in this revolution and lead the future.
