Holding high the banner of "equal rights for intelligent driving", BYD has made the industry prediction of "2025 will be even more competitive" become a reality. It kicked open the door of "No. 1 Intelligent Driving Store" and once again threw its "trump card" into the automobile market.
On February 10, BYD announced the launch of 21 smart driving models at one go, and announced that the "Eye of God" smart driving system will cover all models, from the million-yuan Yangwang U8 to the 69,800 yuan Seagull smart driving version, all of which will have the high-speed NOA (navigation assisted driving) function as standard.
In the past two years, expensive LiDARs have been the ticket to smart driving. Car companies compete with each other in the number of sensors like stamp collectors, making smart driving a luxury item suspended in the cloud.
Many car companies have tried to use lower-cost pure visual solutions to lower the threshold for intelligent driving, but their lethality is not as strong as BYD's this time.
Those high-end technologies that were carefully packaged as luxury goods by other competitors were now displayed by BYD on the shelves of affordable supermarkets.
BYD's entry into the market has woven a larger cost network with pure visual solutions and self-developed chips, salvaging high-end functions such as high-speed navigation and valet parking, and spreading them to the vast ocean of 70,000 yuan models. Just like when Apple made touch screens a standard feature, Nokia's keyboard empire collapsed - now BYD is reproducing this script, but the stage has been changed to a smart terminal with four wheels.
Through the self-developed BEV+Transformer algorithm architecture, BYD has broken away from its dependence on high-precision maps and supports map-free urban road navigation. At the same time, it adopts a low-cost hardware solution of "three-eye camera + 4D millimeter-wave radar", which can reduce costs by up to 60% compared to the industry average.
BYD's self-developed "Xuan Ji Architecture" realizes the dual-circulation coordination of cloud AI and vehicle-side AI for the first time, and optimizes the smart cockpit, intelligent driving and chassis control in real time through the three communication layers of Internet of Vehicles, 5G and satellite network. The architecture opens 142 vehicle control interfaces, allowing third-party developers to call intelligent driving perception data and build an Android-like car ecosystem.
In addition, BYD cooperated with Huawei Cloud to build an EB-level data storage cluster, which processes 1.2PB of data per day. It combined the DeepSeek large model to optimize the vehicle-side AI reasoning capabilities and realize fuzzy recognition of driving intentions and prediction of dynamic obstacles.
The differences in technology routes are being flattened by economies of scale. Over the past five years, car companies have been arguing over technology routes in the field of intelligent driving: the battle between pure vision and multi-sensor fusion, the game between self-developed algorithms and supplier solutions, and the battle between single-vehicle intelligence and vehicle-road collaboration. In essence, these are choices of survival strategies under different resource endowments.
BYD directly jumped out of the above context. BYD's "Eye of God" system seems to use a low-key combination of three-eye camera + 4D millimeter-wave radar, but it has achieved the ultimate in cost control: compared with the laser radar solution that the new forces are proud of, its hardware cost has dropped sharply from 20,000 yuan to 2,800 yuan, which is equivalent to using the pricing of Xiaomi mobile phones to make the iPhone's imaging system.
More secretive tricks are hidden in the data - or with annual sales of more than 4 million vehicles, BYD swallows 12 million kilometers of real road conditions data every day. The AI models fed by these "electronic feeds" are allowing rule-driven algorithms to evolve into driving intuitions close to humans.
This strategy of "trading scale for accuracy" is rewriting the direction of the industry's technological evolution: when a car company can increase the scene pass rate of a pure vision solution to 98% through a closed data loop, the "safety redundancy" value of lidar will have to be questioned in terms of cost-effectiveness.
BYD's low-cost solution of three-eye camera + 4D radar may become a new benchmark for intelligent driving in the 100,000-200,000 yuan class, forcing Tesla, Xiaopeng and others to adjust their hardware strategies. Its high-speed NOA drops to the 60,000 yuan range, which will directly impact the entry-level fuel vehicle market.
The speed at which intelligent driving functions are being introduced to the market far exceeds industry forecasts. An industry report predicts that the intelligent driving installation rate of BYD Seagull, Wuling Xingguang and other 70,000-120,000 yuan models is expected to exceed 60% in the first quarter of 2025. This change has directly impacted the traditional market stratification logic. After the intelligent driving threshold, which was originally divided by 200,000 yuan, was broken, consumers began to classify products by the completeness of the intelligent experience rather than the price range.
Even though it is impossible for BYD to rise from an unknown smart driving company to the top of the industry with just a press conference, the algorithms it has trained with its scale advantage will form a snowball advantage in iteration capabilities, and it is foreseeable that it will overtake some small and medium-sized automakers in the short term.
However, another possibility is not ruled out. If the unit price of LiDAR falls below $200, the multi-sensor fusion solution may rebound. Kaiyuan Securities pointed out in a research report released at the end of December 2024 that the key to accelerating the penetration of intelligent driving lies in cost reduction. The hardware cost of intelligent driving vehicles must drop by 20%-30% year-on-year each year. In December last year, a senior executive of Pony.ai revealed to the outside world that the kit cost of the seventh-generation robotaxi vehicle to be launched by Pony.ai will be about 30%-40% lower than that of the sixth generation.
On the other hand, industry experts believe that the reduction in intelligent driving costs due to the increase in the proportion of self-developed products may marginalize some traditional Tier 1 companies. Zheng Liyao, head of Gaogong Intelligent Automobile Research Institute, pointed out that BYD's supply chain system, with a self-developed proportion of up to 58%, has made "Bosch engineers now have longer and longer PPTs on the disassembly and analysis of BYD's domain controllers during meetings."
Analysts at Huatai Securities interpreted it from a capital perspective, believing that BYD's scale effect is rewriting the valuation logic of smart driving. "When the proportion of smart driving hardware costs drops from 25% to 18%, the capital market's judgment criteria for car companies will shift from 'technological leadership' to 'data monetization efficiency'."
Regardless of the technical path, BYD's "intelligent driving for all" strategy is forcing technological equality through scale. Its competitive barriers built through large-scale manufacturing, domestic substitution and data closed loop are pushing intelligent driving from a "technical show" to a "people's livelihood project." The ultimate goal of this change is to make intelligent driving a basic function of automobiles, just like ABS and ESP.
