A research report indicates that 2025 may be the inflection point year for automotive intelligence. The popularization of City NOA (L3 intelligence core experience) is the driving force behind making intelligence one of the top three considerations for consumers when purchasing vehicles. Currently, based on the "leading automaker new vehicle cycle - terminal grassroots research - blockbuster model validation" system, close tracking of the "L3 intelligence penetration rate" change process from 10%-50%-80%+ during 2025-2027 is underway. In the future, automakers may be divided into three types of companies: B-end Robotaxi operations + C-end personalized brands + high-end vehicle manufacturing. Attention should be paid to VLA/next-generation smart driving architecture deployment, penetration rates, and willingness to pay.
Main viewpoints are as follows:
Why recommend going all-in on automotive intelligence?
1) Automotive intelligence is essentially a mobility revolution. "L3 intelligence helps automakers sell cars - L4 Robotaxi enables automaker software monetization - domestic brands achieve global rise" represents the automotive intelligence trilogy.
2) 2025 may be the inflection point year for automotive intelligence! The popularization of City NOA (L3 intelligence core experience) is the driving force behind making intelligence one of the top three considerations for consumers when purchasing vehicles. Currently, based on the "leading automaker new vehicle cycle - terminal grassroots research - blockbuster model validation" system, close tracking of the "L3 intelligence penetration rate" change process from 10%-50%-80%+ during 2025-2027 is underway.
3) Automotive intelligence operates in an elimination tournament mode, where the strong get stronger. In the future, automakers may be divided into three types of companies: B-end Robotaxi operations + C-end personalized brands + high-end vehicle manufacturing. Component suppliers may be divided into two types: modular suppliers + single-category suppliers.
August Intelligence Summary: Keywords may reflect the iteration effects of new-generation underlying architecture
1) Regarding VLA/next-generation smart driving architecture deployment: On August 20, Li Auto's VLA new-generation smart driving architecture launched simultaneously with the i8 delivery, initially rolling out as VLA Preview version, capable of chain-of-thought reasoning and three-point turns; On August 27, Xpeng's all-new P7 was released, with all Ultra versions featuring VLA+VLM smart driving, equipped with 3 Turing self-developed chips, providing effective computing power of 2250TOPS at the vehicle end; DeepRoute has pioneered third-party smart driving supplier VLA solutions, potentially extending to 100,000 yuan level vehicle models.
2) Regarding penetration rates and willingness to pay: July's City NOA-level intelligence penetration rate reached 23.2%, up 0.9pct month-over-month, maintaining good levels. AITO M8 comes standard with urban NCA across all variants, with deliveries continuing to increase; Xpeng MONA M03 Max drives continued acceleration in supply-side smart driving, with Xpeng's intelligence penetration rate exceeding 70%; Li Auto's overall City NOA smart driving penetration rate was 59.4%, down 2.2pct month-over-month, showing slight decline.
2025 Automotive Intelligence Theme Stock Selection
Continue to be bullish on intelligent vehicles! Intelligence is the core, new vehicle cycles/monthly sales are implementation tracking, and robot deployment is the icing on the cake! Hong Kong stocks > A-shares.
Hong Kong stocks (XPENG-W + LI AUTO-W + XIAOMI-W, etc.) / A-shares (BYD + SAIC Motor + Seres, etc.).
Bullish on intelligent incremental components: AI chips (Horizon Robotics-W + Black Sesame Technologies), domain controllers (Desay SV + Hangsheng Electronics + Joyson Electronics + Jimu Technology, etc.), wire-controlled chassis (Bethel + Nexteer), automotive electronics (Baolv Technology + Hirain Technologies, etc.).
Risk Warning: Passenger vehicle price war exceeding expectations; terminal demand recovery below expectations; L3-level autonomous driving policy rollout pace not meeting expectations.