Nearly two weeks ago, TuSimple, the world's first driverless stock, issued a document announcing that it would be delisted from the NASDAQ. In April 2021, TuSimple went public on the U.S. stock market, and enthusiastic investors pushed TuSimple's market value to $12 billion with simple trust that "autonomous driving is easier than carrying people" and "Robotruck is comparable to RoboRoboti".
From "the first share of unmanned driving" to "the first share of unmanned driving to be voluntarily delisted", it took less than three years for TuSimple to change. In the past three years, drastic changes in political and economic factors have changed the fate of many driverless companies, first Robotaxi and then RoboTruck. Tucson's delisting is not a surprise, but it brings the question that has been raised again and again to the entire truck autonomous driving industryIf the stage of "good wind sends me to the clouds" passes, how will you survive in the increasingly muddy real world? The flag of the truck driverless fellThe foreshadowing of Tucson's delisting was actually laid before TuSimple went public. In March 2021, on the eve of TuSimple's listing, CFIUS launched an investigation, believing that TuSimple, as a company invested by Sina, carries out autonomous driving technology research and development in the United States, and there may be a risk of transmitting data and technology to China and affecting the world. As a result, TuSimple signed an agreement with the United States, the TuSimple team in China and the United States was decoupled, and the directors appointed by Sina were kicked out of the board, ensuring that all members of TuSimple's board of directors were American or Canadian. In addition, Tucson has established a safety committee to report regularly to CFIUS to reassure the United States**. But Tucson's doom didn't end there. In October 2022, foreign media broke the news that CFIUS, together with the SEC and the FBI, launched an investigation into TuSimple, believing that TuSimple was suspected of secretly delivering technology to the hydrogen energy heavy truck Hydron (i.e., Turing Smart Card) founded by co-founder Chen Mo. The successive investigations have clearly affected the trust of the U.S. capital markets in TuSimple, and TuSimple's stock price has plummeted after a brief spike in listing. 
During this period, TuSimple also experienced the resignation of Chairman Chen Mo and CEO Lv Cheng of professional managers - CTO Hou Xiaodi took over the power - Hou Xiaodi was removed by the board of directors and returned to the position of CEO - Chen Mo and Lv Cheng returned, and Hou Xiaodi, as the co-founder, finally resigned from the palace fight reversal plot. To some extent, Tucson is a victim of the geopolitical relationship between China and the United States affecting the field of artificial intelligence. But that's not the only reason. In the announcement, TuSimple also said that there has been a major shift in capital markets, with rising interest rates and quantitative tightening affecting investor decisions. This is difficult to explain simply by the lack of patience of capital, after allEight years into the industry, TuSimple still hasn't proven feasible on its path to commercializing driverless trucks. In the listing prospectus, TuSimple proposed a very good business model: on the one hand, TuSimple and its car company partners (mainly N**iStar) jointly develop L4 autonomous trucks and sell them to transportation companies to earn technical service fees; On the other hand, TuSimple will provide the technology and platform to build a driverless truck logistics network with third-party fleets, provide driverless freight services and charge fees by mileage, and share profits with truck asset holders. 
TuSimple's business model as outlined in the prospectus.
The U.S. traditional trucking market is $800 billion[1], and if it has a 5% share, the market will also witness the rise of a new giant. However, even TuSimple, which ranks in the first echelon of truck autonomous driving start-ups in terms of technical strength, is trapped in many obstacles in reality, and it is difficult to achieve even one-thousandth of the market share. In December 2022, TuSimple's partnership with n**istar was terminated, and the loss of support from car companies in front-end production meant that it was difficult to define a truck model that met the requirements of safety redundancy and manufacturing costs according to the needs of autonomous driving. Political support is also unclear due to concerns about safety and employment. Last May, the Senate in California, where Tucson's U.S. headquarters is located, passed a bill that "self-driving trucks must have safety officers on the road." Once enacted, the commercialization of Level 4 autonomous trucks in California will be stuck after 2030. In 2020, 2021, and 2022, TuSimple's future revenue will be 18430,000 US dollars, 62610,000 US dollars, 936$90,000. In the first half of last year, TuSimple's revenue plummeted to $300,000. In three and a half years, Tucson lost more than $1.5 billion. In 2023, TuSimple's stock price has hovered below $2 for a long time, with a market capitalization of only hundreds of millions of dollars. In the face of the double kill of politics and capital, both the potential market space in the United States and the financing channels of U.S. stocks have lost their meaning for Tucson. Since last year, TuSimple has been seeking to shut down its U.S. operations and shift its focus to China. At last year's Shanghai Auto Show, TuSimple unveiled a domain controller for assisted driving. A few days ago, Hao Jianan, CEO of TuSimple China, said in an interview that TuSimple China "will continue to focus on Level 4 autonomous driving as its main business, and transform some of its technologies into Level 2 autonomous driving business to serve OEM customers." [2] "The unmanned first stock, which is a weather vane, has switched survival strategies. The pace of mass production is unmovingIn the truck autonomous driving industry, trunk logistics autonomous driving was once the most optimistic track in the capital market. Due to its early establishment, strong technology and early listing, TuSimple has always been a representative enterprise of driverless trucks. After TuSimple went public in 2021, its hot stock price also gave birth to a number of new companies in China. But when the flag of the industry fell, the industry players who were staring at it were in little mood to celebrate the weakness of this formidable competitor. TuSimple's autonomous driving of trunk logistics last year came with bad news. In North America, except for a few companies such as Aurora, which is backed by Amazon, the industry is not optimistic. In February 2023, Locomation, a self-driving truck start-up, was hit by large-scale layoffs; In May, Embark, a trunk logistics autonomous truck company that also completed its listing, went bankrupt and was acquired; In July, Waymo's autonomous truck division was laid off, and some technical research and development, operations, and commercial work were postponed. Overseas companies are having a hard time, and the situation of domestic trunk logistics autonomous driving companies is also bad. In 2023, Plus.ai, which had unsuccessfully planned to list on the U.S. stock market through SPAC, was revealed to have chosen to split its China team under its major shareholder, Full Truck Group[3]. Under the change of ownership, the latest news is that the Plusai China team has just undergone a layoff. In May last year, Tao Ji, CEO of Qianhang Technology, resigned, and then switched to Changan Automobile to be responsible for the research and development of intelligent driving. In June, the company was found to have entered bankruptcy and liquidation, less than two years after the company was established. The survival situation of startups has deteriorated, and the change in the preference of capital is a key factor. 
According to Gartner's 2023 Technology Maturity Curve, (unmanned) autonomous driving is at the bottom of the bubble.
In 2023, with the explosion of AGI, primary market capital will collectively turn to large language models, robots and other tracks, and the funds flowing to the field of autonomous driving will further shrink and turn conservative. The few remaining capitals pursue the pursuit of safety in the pocket, and require the invested truck autonomous driving companies to accelerate the business closed loop. However, when American companies that are theoretically less difficult to commercialize driverless trucks are unable to take the pace of mass production, it is difficult to expect domestic companies to take the lead. The domestic trunk logistics autonomous driving is facing a small and scattered pattern at both ends - on the production side, the overall scale of the truck market is small, and the annual revenue of the head enterprises is only tens of billions of yuan; On the demand side, the overall concentration of the logistics industry is low, except for the express market, the LTL and vehicle markets, which account for the majority, are highly fragmented. 
IFC**.
The former has led to the lack of car companies with scale and resource advantages to take the lead, driving the rapid maturity of the advanced chain and technology of truck autonomous driving; The latter is more fundamental, and the lack of enough large logistics companies to put forward real, large-scale truck autonomous driving needs, the industry lacks the driving force for mass production. The biggest bug encountered by truck autonomous driving companies in China is that they urgently need to find a willing to do itAutonomous driving technologyThe truck asset holders who pay the bill, but the largest truck asset holders in China are the truck drivers who are the most reluctant to be served by autonomous driving. In China, the capacity of trucks operated in the form of individuals (including anchored) reached 5.9 million, accounting for more than 70%. Self-employed truck drivers often buy their vehicles financially, competing fiercely in the low-freight logistics market. In long-distance freight transportation, it is not uncommon to see sibling shifts and mom-and-pop shops that "take care of men and women". In addition to driving skills, they can rely on manual labor that is close to zero cost for themselves. Individual truck drivers who trade manual labor rather than capital for profit are unlikely to be willing to pay for autonomous driving. According to the data provided to us by an autonomous driving investor, the current cost of software and hardware of the L4 truck autonomous driving system exceeds 200,000 yuan per vehicle due to the use of more high-specification sensors in autonomous driving trucks, but the lack of shipments and the lack of bargaining power of the ** chain. The average ** of a domestic heavy truck is only more than 40 yuan [4]. In an interview with Yuanchuan Auto Review, TuSimple co-founder Chen Mo said that the biggest difference between Chinese and American truck autonomous driving is the cost of drivers, which leads to the replacement of truck drivers with unmanned vehicles as well," he saidProbably only a fraction of the gains in China are in the United States”。To a certain extent, Chinese truck drivers have turned the domestic truck driverless track into a commercial depression with tenacity and diligence beyond the world level. Even if a small number of large logistics companies are willing to provide some orders for trunk logistics autonomous driving enterprises out of technical reserves, data accumulation, market value management, and vehicle life cycle operating cost optimization, it is difficult to form effective demand beyond regulations at this stage. At the regulatory level, the ** sector has maintained a relatively cautious attitude towards the autonomous driving of L4 trucks on the open road, whether for the purpose of stabilizing employment or considering the greater harm of trucks after traffic accidents. At present, the self-driving trucks driving on high speeds in China are equipped with safety officers and are in test or trial operation. At the same time, in December last year, the General Office of the Ministry of Transport issued the "Guidelines for Transportation Safety Services for Autonomous Vehicles (Trial)", proposing that Robotaxi "under the premise of ensuring safety, with the consent of the people of districted cities, remote safety officers can be used when operating in designated areas", making its commercialization one step closer - just a few years ago, the mainstream view of the industry was RoboTruck (that is, L4 trunk logistics self-driving trucks) will be mass-produced before RoboXi. Under the reality that the cost, technology, ** chain, and regulations are not supported, there are still few start-ups in China that are still adhering to the one-step path to the sky, and there are few start-ups that try to mass-produce L4 autonomous trucks directly on trunk logistics. A mid-level domestic truck autonomous driving company told us that he believes that the mass production time of L4 trunk logistics autonomous driving will take another five years. The story of L4 is slow
Two years ago, Robotaxi companies that faced commercialization difficulties on the L4 track successively put down their bodies and chose to become the leading providers of assisted driving technology for passenger car companies. The same history is repeated in the L4 trunk logistics autonomous driving start-ups: on the one hand, there is a major mountain on the road to L4 autonomous driving, and on the other hand, investors are demanding hematopoiesis as soon as possible, and the reality forces them to choose to learn from Robotaxi companies, postpone L4, and develop intelligent driving solutions that can be mass-produced and applied earlier. Among the autonomous driving companies in trunk logistics, Plus.ai Technology has "stopped telling the story of L4" at CES this year[6], while Inceptio Technology and Qianhang Technology have successively proposed ".Twin drive becomes single drive"Solution: Through intelligent driving, the driver's hands, feet and eyes are liberated, and his driving fatigue is reduced, so that one person can complete the driving task that only two people can be competent for in medium and long-distance freight, and at the same time achieve a fuel saving rate 2-10% higher than that of human drivers. Last year, Inceptio announced that it had verified the feasibility of converting a two-way cargo into a single-driver vehicle in a one-way cargo transportation of 450 800 km, and would further verify the feasibility of 1,500 km [5]. Qianhang Technology gave data that the two-driving to single-driving can save 750,000 yuan in labor costs in 5 years, and the operation of 1 million kilometers can save 200,000 yuan in fuel costs. However, the concept of changing from two drivers to single driving has also raised questions:InBraking distanceIf such a system can replace one driver, why can't it replace the other? For example, in a 20-hour long-distance freight, two drivers usually take turns working for 10 hours in the form of one driving and one sleeping, and if there is intelligent driving technology that allows one driver to undertake 20 hours of work alone, it either means that the driver is superhuman, or it means that the driver can sleep peacefully for a long time - such a system is essentially an L4 autonomous driving system that does not require a driver. 
In addition, due to the lack of sufficient test mileage verification and unclear division of responsibilities, compared with the large-scale push of high-end intelligent driving in passenger car cities, the "dual-driving to single-driving" products created by the trunk logistics autonomous driving company have not been mass-produced and delivered. This also shows the huge difference between the passenger car intelligent driving market and the commercial vehicle intelligent driving market: the consumption motivation of passenger car users for intelligent driving is not completely rational, as long as there is a certain negative reduction effect or even a show-off effect, some consumers will pay for it; In the case of trucks, the utility brought by intelligent driving will be accurately quantified in every hour of operation, and the cost and benefit will be measured by the attributes of the production tool, and the ambiguity will not be tolerated. For the scenario side of trunk logistics, whether it is L4 or L3 or L2, there is no meaning in intelligent driving as a gimmick, and there is only one solution that can be selected in the end: the kind that can really reduce logistics costs. Under the core demand that cost reduction is greater than everything, a number of new truck forces were born - to build new energy intelligent driving trucks, supplemented by electric drive with intelligent driving to reduce the burden and fuel saving effect, to achieve long-term operation cost reduction in energy and labor, which accounts for the majority of truck operations, and to reduce costs in disguised form by reducing safety benefits such as accident rates. Following this line of thinking, "new energy intelligent driving truck new forces" such as Deepway, Zero-One Automobile, Weidu Technology, and Turing Smart Card have been established after 2020. Compared with passenger cars, the threshold for truck manufacturing is not high, and after the startups complete the model design of the truck, they can hand it over to the commercial vehicle company for OEM. At the same time, defining the product by itself can help adapt to the needs of autonomous driving in terms of chassis (brake-by-wire), electronic and electrical architecture, etc., and also facilitate the maintenance of better vehicle costs in the long term. However, supporting a team of car manufacturers will require higher upfront investment and expensive vehicle development expenses, making the business model heavier. It's just that compared with the L4 companies of trunk logistics that do not have much good news, capital has expressed new interest in this well-stitched new story of full-link joint cost reduction. Last year, Deepway (7400 million), 01 Automobile (1400 million) and Weidu Technology (hundreds of millions) have received financing of more than 100 million yuan. Shortcuts that don't exist
Like the autonomous driving of passenger cars, there is also a difference between "top-down" and "bottom-up" autonomous driving of trucks - one is the trunk logistics school, which is directly commercialized on the open road with higher difficulty and greater market space; The other faction is the closed scene faction, which first selects a specific closed scene as a base to obtain enough revenue to ensure survival, and then marches into other scenarios and open roads. Compared with trunk logistics, it is less difficult to achieve unmanned driving in closed scenarios, on the one hand, because the scenarios are relatively simple, and on the other hand, because closed scenarios are not open roads and are less restricted by regulations. However, in previous years, startups that started from a closed scene were not very popular with the capital market because the market value was smaller and more fragmented. Compared with the trillions of potential market space for trunk logistics, the hundreds of billions of value in closed scenarios are scattered in multiple scenarios such as mines, ports, industrial parks, and airports. For example, unmanned mining is a 100-billion-level market, and the market space in the port is about 30 billion. But in the business world, I am afraid of being foiled by my peers. At a time when the commercialization attempt of the trunk logistics faction has encountered obstacles, the silently developed closed scene faction seems to have chosen the right path to the commercialization of truck autonomous driving. Last year, the unmanned mining companies Yikong Intelligent Driving, Boonray Technology, and Tagge Zhixing, and the port autonomous driving companies Xijing Technology and Sinian Intelligent Driving all received financing, with a total amount of more than 1.5 billion. The main reason why the track has been paid more attention by the capital market isThe needs of the scene side are more real and urgent, and startups are closer to mass production. Compared with logistics companies, they hope to reduce costs and increase efficiency through self-driving trucks, and most of the operators in these scenarios of mines and ports are state-owned enterprises or large private enterprises, and they face some equally important problems: once personnel occur in mines or ports, it is a major production safety accident; The working environment is poor, the driver turnover rate is high, and the local driver shortage is more prominent. This gives them more motivation and resources to drive the deployment of driverless trucks. Last year, Tago Zhixing and Yikong Zhijia successively signed deployment agreements with miners for 300 units and 500-1000 unmanned mining trucks. At the port, Bao Xuchao, product director of Sinian Intelligent Driving, said that this year will be the first year of mass production of unmanned trucks in the port, and the number of mass production is expected to reach 500 units. 
A large number of sensors are standard in driverless trucks.
But closer to mass production does not mean a "better business model".The reason for this is precisely the high fragmentation of the market. Unlike the "technology is applicable to a wide range of roads once it is run", it is not easy to migrate and reuse the technical solutions in closed scenarios. On the one hand, even in the same scenario, the surrounding environment and fleet operation requirements will be different; On the other hand, truck autonomous driving technology companies are currently tackling tough problems one mine after another, one port after another, and the revenue of a single project is not necessarily high, but the demand for customized development and on-site development is not low. Si Nian Zhijia said that even in a container port with a high degree of standardization, it will take 3-6 months for them to deploy their existing technical solutions in another port, and it will take 3-6 months to meet the production operation standards. This is already a high level in the industry, and some companies take a year. It is more difficult for unmanned truck startups to expand from their respective bases to other subdivision scenarios, because different subdivision scenarios also have their own unique difficulties and know-how: mines are usually in remote areas with difficult conditions, which can easily cause the loss of technical talents; There is a lot of dust and mechanical vibration in the mine, which poses a great challenge to the reliability of autonomous driving sensors. Container ports require centimeter-level positioning accuracy, and must develop high-performance and high-reliability positioning software and hardware; Some factories are partially narrow, and old human drivers need to reverse and "move slowly", which will naturally test the regulatory algorithm of unmanned trucks... In the face of the small and fragmented situation of unmanned trucks in the closed scene market, some companies have chosen to "do everything they can do in the scene", which is the above-mentioned Xijing Technology. In June last year, Nishii signed a cooperation agreement with Port Felisdu in the United Kingdom for 100 new energy driverless trucks. As an autonomous driving company based on the port, Xijing not only develops unmanned driving technology in the port, but also builds new energy unmanned commercial vehicles, develops a battery swap station by the way, and provides a set of intelligent production scheduling solutions to the port side. 
Westwell Technology's unmanned commercial vehicle + battery swap station.
Now Xijing is not so much an unmanned truck company as a leading provider of intelligent comprehensive solutions for ports. The benefit of doing more is higher revenue. Its investor and high capital said that among the port unmanned companies, Xijing Technology has the highest revenue. It's just that when Nishii chooses a large number of diversified businesses, it also means saying goodbye to the most respected business model in the autonomous driving industry: a technology company conquers the market with a large-scale replicable intelligent driving software system, and becomes a new giant with a very high gross profit margin. Last year, Westwell Technology was selected as one of the "Top 500 Unicorn Enterprises in China" by the China Economic Development Research Association, and the valuation of Westwell, which ranked first, was "only" 7 billion yuan. Compared to the concept of unmanned driving, this is obviously not a high enough valuation. But perhaps it shouldn't have been a pity in the first place. After several years of ups and downs, probably startups will gradually accept that the simple and best business model does not exist in the field of autonomous driving. References: [1] TuSimple IPO Prospectus.
2] TuSimple Future Responds to Post-Delisting Development: The U.S. business is already in a frozen state, and will focus on the Chinese market, Titanium**.
3] Plus.ai will spin off its China team to prepare for another impact on U.S. stocks.
4] The truck market is back in smoke, China Automotive News.
5] * Two days, truck intelligent driving ignites the commercialization of torches, 36 krypton.
6] Looking at the 2024 CES, what are the new directions for Chinese and foreign unmanned driving companies, Cyber Automobile.