LAIKA

2024/10/31

Are domestic autonomous driving companies on the right track? Not necessarily!

After Horizon Robotics (9660.HK) officially went public in Hong Kong on October 24, it sparked strong interest from domestic and foreign investors in domestic autonomous driving companies. As the largest technology IPO in Hong Kong this year, Horizon Robotics not only attracted star cornerstone investors such as Alibaba and Baidu, but also saw its share price surge by up to 38% on the day of its listing. Moreover, BG, a top global technology stock investment institution, subscribed for over US$500 million, demonstrating the recognition of professional investors towards Horizon Robotics.

The following day, on October 25, US Eastern Time, WeRide, known as the "first global general autonomous driving stock," officially listed on Nasdaq. Just seven years after its establishment, WeRide's share price surged significantly on its first day of listing, triggering two circuit breakers.

Among the companies that have submitted prospectuses and are preparing to go public are numerous domestic autonomous driving enterprises such as Pony.ai, Momenta, Zongmu Technology, Udrive Innovation, and Xijing Technology, among others. Coupled with already listed companies such as Black Sesame Technologies, Hesai Technology, RoboSense, and Zhixing Technology, there will be over ten domestic autonomous driving companies listed simultaneously in the future.

According to Qichacha data, as of early October this year, there were 6,472 existing smart driving-related enterprises in China, with their registration volume showing a fluctuating growth trend over the past decade. In the first half of this year, 263 new enterprises were registered, representing a year-on-year increase of 9.13%. Domestic smart driving is springing up like mushrooms after rain.

01

The market is hot, policies are boosting, and there are continuous benefits for smart driving

Although TuSimple, a domestic self-driving truck brand known as the "first autonomous driving stock," delisted from Nasdaq earlier this year due to losses, it cannot stop domestic autonomous driving companies, which are being labeled as "first" in various ways, from gearing up for their IPOs.

For the capital market, the commercialization potential of autonomous driving has been fully demonstrated this year, with news about Robotaxi emerging constantly since the beginning of the year. From Baidu's Luobo Kuaipao facing driver protests and controversy in Wuhan to Google's Waymo fully opening up in San Francisco, and Tesla's long-awaited Robotaxi product, the Cybercab, being officially launched and expected to enter production in 2026 at a cost of less than US$30,000 – a series of disruptions brought about by autonomous driving technology have greatly sparked enthusiasm in the capital market.

The popularization and application of smart driving technology are also accelerating. For ordinary car consumers, smart driving is no longer a mysterious technology only seen in science fiction movies. A large number of new energy vehicles are applying cutting-edge smart driving technology. As we introduced in our previous article on Horizon Robotics' IPO, smart driving for passenger cars is currently divided into Advanced Driver Assistance Systems (ADAS) and Advanced Driving (AD). Technologies such as adaptive cruise control, lane-keeping assistance, and lane-change assistance, which are widely used in L2-level autonomous driving, fall into this category and are already applied to cars priced at around US$10,000, which have been well-received by consumers.

The appeal of Advanced Driving to consumers is also increasing. For example, XPeng's newly launched Mona M03, despite being a new energy vehicle with a wheelbase of over 2.8 meters and a rear suspension of torsion beam instead of independent suspension, has achieved remarkable sales figures due to its combination of an affordable price of around US$15,000 and XPeng's Advanced Driving technology. It is understood that XPeng's Mona M03 has accumulated over 80,000 orders, with over 10,000 deliveries in September, accounting for half of XPeng's sales that month. XPeng's strategy of bringing Advanced Driving technology, originally reserved for high-end models, down to mid-to-low-priced models has indeed translated into market sales.

Looking back, at BYD's Dream Night in January this year, Wang Chuanfu also stated that BYD was the first to shout out the slogan, "The first half is electrification, the second half is intelligence," and subsequently launched some models equipped with its self-developed "Divine Eye" system. In BYD's dominant price segment, L2-level driving assistance functions have become ubiquitous, except for the lowest and second-lowest configurations.

Another brand that cannot be overlooked when it comes to smart driving is Huawei. The Wenjie series equipped with Huawei HarmonyOS Intelligent Driving has become the most successful product series among domestic high-end new energy SUVs. Among them, the Wenjie M9, with a main selling price of US$50,000, received over 160,000 pre-orders in the first ten months after its launch and has successfully topped the sales chart for vehicles priced above US$50,000. The share price of Thalys (the manufacturer of Wenjie) has also soared from around US$1 in 2020 to US$10.9 today, and Huawei's overhaul of Thalys has become an industry legend.

As for autonomous driving, the ultimate form of smart driving, relevant supporting policies have also been implemented. According to China Economic Net, the Ministry of Industry and Information Technology and three other departments announced that China's first consortium, consisting of nine automobile manufacturers and nine users, will carry out pilot projects for intelligent and connected vehicle access and road access in seven cities, including Beijing, Shanghai, and Guangzhou. Additionally, as of the end of August this year, public security organs had issued a total of 16,000 test licenses for autonomous vehicles and opened up 32,000 kilometers of public test roads, strongly supporting the verification and iterative updates of autonomous driving technology.

Smart driving companies are already standing at the cusp of opportunity.

02

Are domestic autonomous driving companies on the right track? Not necessarily!

Although smart driving has experienced rapid development amidst the wave of new energy vehicles in recent years, it is still too early to talk about victory.

In terms of the overall technology roadmap, while China and the United States are globally leading in the ultimate form of autonomous driving, the current technology roadmap does not point to a clear winner.

From a starting point perspective, the earliest autonomous driving project in the United States was initiated by Google in 2009 and later spun off as Waymo. Baidu, which has always benchmarked against Google, is also the earliest domestic company to layout autonomous driving, establishing its Deep Learning Lab in 2013. There is only a four-year gap between the two countries in terms of autonomous driving layout.

In terms of technology roadmaps, there are mainly the following approaches. The first is the hardware school represented by Waymo, which focuses on developing its entire suite of sensors, LiDAR, millimeter-wave radars, and computing platforms in-house. The biggest drawback of this approach is the massive upfront capital expenditure. The second is Tesla's software-centric approach championed by Elon Musk, which relies on large models to provide end-to-end technical solutions. Images are input, and neural networks output decisions, delegating all tasks to AI. This is known as Full Self-Driving (FSD), a pure vision-based solution.

Currently, domestic autonomous driving companies primarily follow a practical middle ground. For example, Baidu Apollo provides the Intelligent Vehicle Infrastructure Cooperative System (IVICS), which utilizes advanced wireless communication and next-generation internet technologies to enable real-time information exchange between vehicles and roads. Based on comprehensive dynamic traffic information collection and integration, it conducts active safety control of vehicles and coordinated road management, fully realizing effective coordination between humans, vehicles, and roads, ensuring traffic safety, and improving traffic efficiency, thereby creating a safe, efficient, and environmentally friendly road traffic system. In short, it encompasses both smart vehicles and smart roads.

The reason for the different approaches between the two countries largely stems from their respective leading fields. The hardware or software-centric approaches in the United States are heavily reliant on computing power and chip support. The United States remains a global leader in integrated circuit technology, laying a solid foundation for high-end automotive chips. In contrast, China excels in infrastructure fields, such as the 5G and satellite internet initiatives launched in recent years, which can facilitate the rapid development of vehicle-road coordination.

Regardless of which technical route autonomous driving ultimately takes, the hottest topic for domestic new energy vehicle companies at present is undoubtedly "end-to-end large models."

An "end-to-end large model" simply refers to a system that inputs information at one end and directly outputs control signals at the other, significantly raising the ceiling for smart driving. Tesla's FSD is an end-to-end solution that has already been proven effective, which is why it is highly sought after by domestic automakers. A search for "end-to-end large models" on Baidu reveals that various brands are vigorously promoting their own end-to-end large models. Currently, Huawei-affiliated brands, NIO, XPeng, Xiaomi, Arcfox, IM Motors, and others have officially announced urban Advanced Driving functions based on end-to-end large models. Gu Junli, the head of Chery's autonomous driving subsidiary, also recently stated in an interview that end-to-end technology is a must-win battle for Chery.

Behind the phenomenon, we find that regardless of the specific algorithms employed, the core essence ultimately lies in data and data management capabilities. In other words, the competition ultimately boils down to computing power. Although domestic manufacturers are currently deploying computing power to cope with the training consumption of autonomous driving models, as mentioned earlier, the United States has an industrial advantage in this regard and has been strangling China's related industries in recent frictions. With hardware breakthroughs unlikely in the short term, domestic manufacturers have a long way to go to achieve world-class computing power capabilities.

Apart from computing power, another challenge facing smart driving companies lies in profitability. To date, listed domestic smart driving companies are still incurring losses. While the future of smart driving looks promising, without sufficient financial support, most smart driving enterprises may not survive long enough to witness that bright future. For automakers, profitability is essential; without it, computing power remains a pipe dream.

Therefore, domestic smart driving companies are currently rushing to secure various "first" titles and going public as quickly as possible. While profitability is fundamental, smart driving is a sector with substantial upfront investments, and it still requires substantial capital market infusions.

Conclusion

In the emerging field of smart driving, domestic enterprises are springing up like mushrooms, and the enthusiasm of the capital market continues to fuel hope. However, the pinnacle of technology is not achieved overnight, and the underlying challenges and concerns cannot be ignored. From the competition for computing power to the difficulties of profitability, the development path of domestic smart driving is fraught with uncertainties. Only through continuous technological innovation and market adaptation can enterprises truly grasp the initiative for the future. Looking ahead, domestic smart driving companies need to embrace changing market demands with firm steps, drive technological progress, and strive to go further on this broad road to achieve genuine industry leadership.