According to Zhitong Finance APP, Guosheng Securities released a research report stating that the commercialization process of autonomous driving technology is accelerating. L2-level autonomous driving systems are gradually becoming the mainstream configuration, with most mid-to-high-end models, including Tesla, Mercedes-Benz, BMW, etc., now starting to standardize L2 systems. NOA has promoted the transition from L2 to L3, becoming one of the core competencies in the smart automobile and autonomous driving sectors. Currently, the overall penetration rate of NOA technology is on the rise but has not yet been widely adopted across all models, mainly concentrated in high-end electric vehicles and some traditional luxury cars equipped with ADAS systems. With the gradual commercialization of technologies such as L2, highway NOA, and city NOA, the demand for computing power (especially edge computing power) continues to expand.

Zhongsheng Securities' main points are as follows:

Current situation: Computing power is accelerating, and smart driving is rapidly developing.

L2 has become the market mainstream, with the penetration rate gradually rising. The rapid progress of smart driving technology, especially systems at L2 level and above, mainly stems from the improvements in computing power, mature sensors technology, and optimized AI algorithms in recent years. Since 2021, with the decrease in the cost of autonomous driving hardware, enhancements in data processing capabilities, and gradual loosening of industry regulations, the commercialization process of autonomous driving technology has accelerated. L2-level autonomous driving systems are gradually becoming the mainstream configuration, with most mid-to-high-end models, including Tesla, Mercedes-Benz, BMW, etc., beginning to standardize L2 systems.

L2+ and L2++ functions are further enhancements of L2, adding more advanced features such as automatic lane changes, automatic parking, and traffic congestion assistance, although their penetration rate is currently relatively low. L3 level can achieve fully autonomous driving under specific conditions, requiring drivers to intervene only when the system cannot handle the situation; it is still in the experimental stage, with low penetration rate globally, primarily appearing in autonomous taxis, closed areas, or certain test fleets.

Current hot topics: NOA and L2.

NOA has promoted the transition from L2 to L3, becoming one of the core competitive advantages in the smart automobile and autonomous driving fields. NOA (Navigate on Autopilot) is an upgraded feature in the autonomous driving sector, first introduced by Tesla. Tesla's Autopilot and FSD packages include functionalities for city NOA and highway NOA. The current highway NOA provides features such as automatic cruising, automatic lane changes, overtaking, and automatic entry and exit of ramps. City NOA requires higher standards in high-precision mapping, real-time data, and perception capabilities, able to handle dynamic traffic situations such as intersections, traffic lights, and pedestrians. Currently, the penetration rate of NOA technology is on the rise but has not been widely adopted across all vehicle models, primarily focusing on high-end electric vehicles and some traditional luxury cars equipped with ADAS systems.

Behind the smart driving hotspot: the landing of computing power application scenarios and the expansion of demand.

With the gradual commercialization of technologies such as L2, highway NOA, and city NOA, the demand for computing power (especially edge-side computing power) is continuously expanding. The upgrade from L2 to NOA not only reflects advancements in software and algorithms but also the ongoing enhancement of hardware computing capabilities, transitioning from traditional onboard computing platforms to the necessity for high-performance GPU, ai chip, and support from cloud computing. Computing power has become a critical factor for the implementation of smart driving technology.

Expansion of computing power application scenarios: As computing power gradually becomes popular and improves, future smart driving will no longer be limited to high-end models, but will gradually penetrate more mid-range and lower-end models. For instance, an increasing number of mid-range vehicles will start to be equipped with L2+ or highway NOA systems and city NOA systems, and application scenarios are expected to expand rapidly alongside competition among automotive companies. Smart driving is likely to become one of the core strengths in future automotive competition.

Expansion of total demand for computing power: On one hand, the acceleration of smart driving implementation requires real-time processing of large amounts of data from sensors (such as cameras, radar, lidar, etc.) and rapidly converting it into driving decisions. Remote cloud computing poses risks of network latency, thus edge computing power has become a core requirement. On the other hand, as the functions of smart driving systems gradually expand, the volume of data processed and computational complexity will also increase, especially in handling complex urban traffic scenarios, requiring real-time analysis of dynamic obstacles, road conditions, and traffic signals, which consumes considerable computing resources. Therefore, with the acceleration of smart driving, the demand for edge computing power at the vehicle end will also rise significantly.

Investment recommendation: Remain bullish on the computing power sector. In the future, as application scenarios in various downstream fields accelerate their implementation, the demand for computing power from both the AI side and edge side will show an exponential growth trend. Therefore, at this current point in time, continue to be optimistic about leading optical module companies represented by "Yi Zhongtian." Also, recommend focusing on the development of edge computing power, especially in the field of smart driving, and suggest paying attention to shanghai huace navigation technology (300627.SZ), Meige Communications (002881.SZ), guangzhou haige communications group incorporated (002465.SZ), fibocom wireless inc. (300638.SZ), and Huihan Co., Ltd. (301600.SZ).

Risk Warning: AI development falling short of expectations, computing power demand falling short of expectations, market competition risks.