Gasgoo Munich-In 2026, the barrier to entry for intelligent driving assistance systems was shattered again. BYD's small car Seagull added LiDAR to its options list for the first time in its annual refresh, priced at 12,000 yuan.
Almost simultaneously, Huajing S, positioned in the 150,000 yuan class, made LiDAR and Urban NOA standard across the lineup, while Bozhi 3X packaged memory parking and highway navigation into a "Smart Enjoyment Package." High-end driver assistance is flooding into the mainstream 100,000 to 200,000 yuan market.
The exclamation that "intelligent driving assist has become commodity-priced" is spreading. But is this an inflection point of inclusivity brought by industrial maturity, or a hardware race forced by falling costs and competitive anxiety? What exactly is driving this rapid trickle-down of technology?
How Did We Get to Commodity-Priced?
According to data from the Ministry of Industry and Information Technology, the penetration rate of L2 driving assist functions in China's new passenger cars reached 69.15% in the first two months of 2026. This figure is up 10 percentage points from the previous year. In other words, out of every 100 new passenger cars sold, more than 69 feature L2 capabilities.
Core functions like automatic emergency braking, full-speed adaptive cruise control, and lane centering have become widely available even in the market under 100,000 yuan. User awareness and driving habits regarding assist features have matured, clearing the psychological hurdle for higher-end L2+ and L2++ functions to trickle down. Automakers no longer need to educate the market from scratch.
The sharp drop in core sensor costs provides the most direct industrial support for this commoditization. LiDAR is a typical microcosm of this cost-reduction wave. Around 2020, the unit price of a mechanical LiDAR was still in the tens of thousands of yuan—almost comparable to an entry-level vehicle. By 2025, Hesai Technology launched the ATX semi-solid-state LiDAR designed for the ADAS market, pushing the mass-production unit price down to around $150.
According to data from Gasgoo Automotive Research Institute, in the first quarter of 2026, Hesai Technology led the domestic passenger vehicle LiDAR market with 328,246 units installed and a 34.9% market share. Huawei Technology followed closely with 303,753 units and a 32.3% share. Together, the two companies accounted for 67.2% of the market.
Gasgoo Automotive Research Institute analyzes that in the first quarter of 2026, the advantage of local enterprises in tracks like LiDAR, high-definition maps, and high-precision positioning further expanded. Industry concentration continued to climb, forming a stable local dominance. In areas traditionally held by international giants—such as driving ADAS, forward-facing cameras, and automated parking—domestic suppliers are rising strongly. They leverage proprietary technology, vertical integration, and vehicle integration advantages, with market shares climbing steadily. From a deep trend perspective, the domestic core components supply chain is undergoing a structural reshaping against the backdrop of vehicle electrification and intelligence. Import substitution is entering an accelerated deepening cycle.
In other words, the sensor and domain control markets once dominated by international Tier 1 giants are now forced to continuously lower prices in the face of close competition from local suppliers. Falling costs are not just the product of technological progress, but the result of a shift in bargaining power within the industry chain. The disappearance of hardware barriers has made installation a realistic option.
What truly drove sensor prices from low to installable was the extreme scale effect of leading automakers. BYD's full-year sales reached 4.6 million units in 2025, with e-platform 3.0 models like the Seagull and Dolphin contributing a massive base. When a single vehicle platform supports annual shipments of hundreds of thousands of units, R&D, tooling, and software adaptation costs are amortized to a negligible level per vehicle.
On the supply chain side, manufacturers like Hesai and RoboSense are willing to offer more aggressive quotes and flexible cooperation terms to lock in massive orders.
Image Source: BYD
BYD Seagull's 12,000 yuan "Intelligent Driving Assist Option Package" is precisely the product of this logic, maintaining positive gross margins even after cost amortization. It is not a marketing gimmick regardless of cost, but a natural outcome after industry chain efficiency reached a critical point.
Furthermore, the competitive strategy of adding high-end specs to entry-level models has accelerated this process. In the commuter car market below 100,000 yuan, traditional selling points like range, space, and screen size have been exhausted. Intelligent driving assist has suddenly become the most recognizable differentiator.
Huajing S boldly made LiDAR standard across the lineup in the 120,000 to 150,000 yuan price range and supports Urban NOA. Bozhi 3X packaged intelligent parking and highway navigation into a "Smart Enjoyment Package" to target young families. These moves continue to lower user price expectations for intelligent driving configurations. BYD Seagull followed this trend by offering LiDAR as an option rather than standard equipment. This responds to competitors' low-price offensives and tests market acceptance based on its own sales volume.
The trickle-down of intelligent driving assist is driven by improved supply chain efficiency and the urgency of keeping up with competitors under intense market competition.
Rationalizing the Trickle-Down
Aligning hardware specifications is far from equivalent to the equalization of intelligent driving assist experiences. This issue needs to be unpacked from three dimensions: platform capability, scenario matching, and automaker strategy.
First is the physical ceiling of platform capability. Advanced driver assistance, especially in Urban NOA scenarios, requires real-time perception, prediction, and planning. This applies to open road conditions like dense mixed traffic, unprotected left turns, and irregular traffic lights. These tasks place stringent demands on the computing platform. Currently, domain control solutions that can stably support Urban NOA generally have computing power above 100 TOPS. They must be paired with complex models like BEV+Transformer and a massive data closed loop.
Low-priced models, limited by vehicle positioning, thermal management, and overall vehicle cost, likely have intelligent driving domain control computing power in the tens of TOPS range. The addition of LiDAR can indeed significantly improve perception capabilities for irregular obstacles and night scenarios, making basic L2 functions like AEB and lane keeping more stable. However, if directly bearing heavy-load tasks like Urban NOA, the situation may arise where system computing power is stretched thin and algorithms cannot be fully deployed.
Even if functions are nominally available, the experience gap compared to vehicles priced over 200,000 yuan remains significant. This is evident in acceleration smoothness, pass rates at complex intersections, and system degradation frequency. Sensors can trickle down, but computing power, model precision, and depth of engineering adaptation still retain distinct class characteristics.
Second is the mismatch between usage scenarios and user willingness to pay. The core user profile for low-priced models is mainly urban short-distance commuting and second family cars, with core road conditions being urban low-speed, congestion, and frequent stops. In this usage scenario, the call frequency for Highway NOA is extremely low, while Urban NOA, although matching the scenario, faces the constraints of the aforementioned experience uncertainty.
BYD making LiDAR a 12,000 yuan option rather than standard equipment reveals a prudent judgment of real market demand. Automakers use this to maintain flexibility, neither absent from the technology narrative nor rashly transferring costs to all consumers.
Users in the 100,000 to 200,000 yuan bracket targeted by Huajing S and Bozhi 3X have higher expectations and willingness to pay for intelligent driving assist. Consequently, these models adopt standard or quasi-standard strategies. However, even at this price point, the transition dilemma of "hardware first, experience later" remains prominent.
Image Source: Huajing
Some models only enable basic L2 and parking assist at the initial launch, marking highway navigation as "subsequent OTA push" and Urban NOA as "function planned." After paying hardware costs including LiDAR, consumers may be unable to fully use all the advertised functions for as long as half a year or even a year. The gap between initial perception and expectations easily triggers reputational controversy.
Whether through BYD Seagull's cautious options or Huajing S and Bozhi 3X's aggressive standardization, the industry has not yet crossed the experience chasm. This chasm is defined by functional completeness, scenario matching, and user expectations. Hardware trickle-down is just the beginning; experience delivery is the truly long tug-of-war.
The Invisible Challenges Have Just Begun
The popularization of intelligent driving assist has a glamorous side, but it brings real costs. Safety redundancy, maintenance, after-sales, and user education are becoming public issues the industry must face.
First is the physical boundary of safety redundancy. LiDAR and surround-view cameras improve active safety capabilities. They identify sudden risks like jaywalking pedestrians and electric bicycles in advance, avoiding collisions or reducing impact speed. However, any perception and decision system has physical blind spots and long-tail scenarios that algorithms cannot avoid; once a collision occurs, it is irrevocable.
If the industry simply equates LiDAR with "segment-leading safety" in its communications, it may inadvertently weaken consumer attention to the vehicle's basic structural safety level. This itself is a cognitive risk that needs vigilance.
Second is the hidden jump in maintenance and after-sales costs. Precision sensors like LiDAR and forward-facing millimeter-wave radars are commonly placed in areas prone to low-speed collisions, such as the front bumper, intake grille, and fenders.
In urban driving scenarios, low-speed rear-end collisions, parking scrapes, and minor accidents are highly common, and these sensor components are extremely susceptible to "consequential damage." Unlike the sheet metal painting of traditional bumpers, once a LiDAR module is damaged, it usually cannot be repaired individually but must be replaced as a whole. Moreover, after replacement, multi-sensor joint calibration must be performed. The material and labor costs for the entire process often start at several thousand yuan, equivalent to or even higher than the cost of sheet metal painting for a medium-sized accident.
For users of such economy models, a seemingly minor front-end collision could bring a repair bill exceeding 1/5 of the vehicle price. Even if paid by insurance, the significant premium increase the following year becomes a continuous hidden cost. Users of higher-priced models like Huajing S and Bozhi 3X are slightly less price-sensitive. However, they cannot avoid the restructuring of after-sales costs caused by treating perception hardware as consumables.
As the penetration rate of intelligent driving assist hardware expands rapidly, insurance product design, loss assessment standards, and repair systems must keep up. This synchronization is a key variable affecting users' total cost of ownership.
Image Source: XPENG
Lagging user education is a deeper systemic cost. Commercial names like L2+, L2++, and high-end smart driving are inherently vague. Words like "automatic" and "smart driving" in some marketing rhetoric easily lead some users to form unrealistic expectations, mistakenly believing that the vehicle already possesses full-scenario autonomous driving capabilities.
In the low-price market, the proportion of first-time car buyers is high and digital literacy varies widely, making the risk from this cognitive gap more prominent. In recent years, accidents caused by drivers completely detaching their attention after enabling driver assist have been reported multiple times. In individual cases, drivers were even sleeping or playing with their phones inside the car, exposing a shocking cognitive vacuum. When such functions trickle down further to a larger user base, the accumulation of misuse risks may not be linear but could amplify exponentially.
Facing these costs, many pragmatic neutralizing approaches have emerged within the industry. Some automakers retain base versions without LiDAR on entry-level models, decoupling features with strong user perception and higher usage frequency like smart parking and panoramic imaging into independent options. This allows users with no clear need for high-end driver assist to avoid bearing extra costs and cognitive burdens for capabilities they won't use.
Huajing S and Bozhi 3X have strengthened the intervention intensity of Driver Monitoring Systems (DMS). They have also set up intelligent driving assist function learning links on infotainment and mobile systems to reduce abuse risks procedurally. These explorations represent rational self-protection at the corporate level. They also offer a direction for industry consensus: reserving a safety buffer through function decoupling and tiered supply is more important than blindly stacking hardware lists.
Conclusion
When LiDAR enters the small car market as an option, it becomes a milestone in the industrialization process of ADAS technology. It proves with a clear example that the Chinese automotive industry chain has crossed a critical threshold in capabilities regarding sensor cost reduction, scale effects, and vehicle integration. Perception hardware once exclusive to high-end models is beginning to move towards ordinary commuter cars.
However, commoditization is never synonymous with perfection. The inflection point of intelligent driving assist trickle-down is not a precise moment. It is a long-term process that is gradually unfolding.
In this process, the reduction in hardware costs has merely raised the curtain. Making functional experience match real user needs is critical. Building systemic capabilities to support a larger parc in areas like safety redundancy, after-sales guarantees, and user education is equally important. These are the core propositions determining whether the popularization of intelligent driving assist can be sustained.
