news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Burning money all the time and failing to make a profit is the true portrayal of the intelligent driving industry, especially the autonomous driving industry. Some people also say that autonomous driving is the biggest pitfall in the automotive industry.

Cruise, a global leader in Robotaxi (unmanned taxi), cannot escape the curse. GM's financial report shows that its losses in 2023 alone amounted to $2.414 billion. Cruise was founded in 2013 and acquired by GM in 2016. If calculated from 2017, Cruise's cumulative losses have exceeded $8 billion.

"The smart driving race started around 2014, and ten years have passed, but no company has really succeeded." An early investor revealed to China Newsweek that some companies in the first echelon of Robotaxi (driverless taxis) lose more than 500 million yuan each year, while some companies focusing on closed scenarios lose around 100 million yuan each year.

"Running out" means that the company has been listed and has a high market value. In recent years, the financing environment has become calmer, and the funding problem is exerting pressure on every player in the industry chain. There is no shortage of unicorns with a valuation of tens of billions, all seeking to go public to "extend their lives". Most of the companies that have been listed before are still not free from losses.

Taking the two companies that filed for listing on the US and Hong Kong stock markets this year as an example, from 2021 to 2023, the general autonomous driving unicorn WeRide’s cumulative annual losses exceeded 4.2 billion yuan, and the chip supplier Horizon Robotics’ cumulative annual losses exceeded 17.5 billion yuan.

Wang Chuanfu once blasted driverless cars: "Driverless cars are all nonsense. They are just some pretentious things to fool people. They are being controlled by capital, just like the emperor's new clothes." This slightly extreme view was refuted by the intelligent driving industry.

"But now, capital's patience is almost running out," lamented the aforementioned investor.

Picture/TuChong Creative

A money-burning marathon

Is intelligent driving the “Emperor’s New Clothes”? Perhaps we can only tell when we get close to the finish line.

In order to reach the finish line, most local startups have chosen a "two-legged" strategy: while exploring the testing and commercialization of L4 and above commercial vehicles and passenger cars, they also cooperate with OEMs to become L2-L3 intelligent driving solution providers or suppliers.

According to the purpose of the vehicle, autonomous driving can be divided into two categories: commercial vehicles and passenger vehicles. The application scenarios of commercial vehicles include trunk logistics, port scenarios, logistics parks, mining areas, airport scenarios, terminal delivery, unmanned cleaning, etc.; the main goals of the passenger vehicle field are Robotaxi and automobile pre-installation mass production.

Several company founders who have invested in L4 technology to develop complete vehicles and expand into the commercial vehicle sector used the "marathon" metaphor to China Newsweek.

Wu Gansha, Chairman and CEO of UISEE, believes that the competition and struggle in driverless cars are like a marathon and a boxing match. The boxing match means the final stage of competition, which is the future stage of all driverless travel and logistics. The previous marathon is a protracted war, and startups must run to the end to get the qualification to enter the final stage.

"During the marathon, the extreme point may come at any time. The fatigue and lack of sense of achievement caused by long-term efforts to overcome a problem without making a breakthrough may be the most difficult." He told China News Weekly, "The question that start-ups need to think about is how to build core competitiveness that giants may find difficult to possess in the future and form a competitive threshold."

How much money does it cost to develop autonomous driving? Is there an optimal solution for the game between R&D investment and commercialization? Different companies have their own considerations.

UISEE was founded in 2016, and is currently focusing on the airport and manufacturing industries to achieve commercialization. Wu Gansha said that taking the company's first unmanned driving project at Hong Kong International Airport as an example, it took a total of six years.

"Since the project was approved at the end of 2019, we have continued to invest hundreds of millions of dollars every year, released more than 50 versions, fixed hundreds of bugs, and added hundreds of functions. We need to constantly align with the airport's highest safety and operational efficiency standards and solve problems." Wu Gansa admitted that the process of algorithm tuning is actually very "painful". "You may solve one problem but another one will arise. When you solve one problem, another detail will become worse."

In addition, the high cost comes from the complex operation and maintenance system, including the management, assessment and problem-solving system. "In addition to the cloud security personnel, the on-site operation and maintenance personnel are more important. Once a problem occurs, an early warning is required within seconds, and the problem needs to be solved remotely within one or two minutes. If remote support is not possible, the problem needs to be solved on site within half an hour," he said.

Speaking of "burning money", Ma Wei, co-founder and CEO of Xidi Zhijia, analyzed to China Newsweek that in terms of R&D investment, although it is necessary to cultivate young talents who are proficient in high-end algorithms to cope with the challenges of iterating trillions of data in large models, based on the team size of several hundred people, this investment is still "relatively limited."

He pointed out that the uncertainty of "burning money" lies more in the process of verifying technology. "It's like the chicken and egg problem. In order to verify that a technology can be unmanned and can be used by customers, and to achieve large-scale deployment in the future, a certain amount of vehicle and maintenance costs need to be invested in the early stage. The upper limit of the scale of investment required for verification varies in different scenarios and different time spans to repeatedly prove the feasibility, but it is often the biggest part of the cost."

Xidi Zhijia was founded in 2017 and has achieved commercial closed loops in vehicle-road-cloud integration and unmanned mining trucks. Therefore, when implementing unmanned mining trucks, the company first chose to start with small mines to verify the technology, reducing the cost of persuading customers; after entering large mines, the deployment requirements and scale increased accordingly, and the mixed operation of manned and unmanned driving also brought new technical challenges.

Zhang Dezhao, Chairman and CEO of Idriverplus, mentioned another major challenge: "The market penetration rate of intelligent equipment is still at a low level. Compared with competition from peers, it is more important now to constantly convince customers through products and technologies, educate the market, and compete with traditional equipment." At the same time, the iterative upgrades of technology and the frequent updates of customer needs also bring long-term tests to after-sales operation and maintenance.

The UnicornsurgentlyNeed blood transfusion

Another smart driving company is going public. On July 27, WeRide formally submitted IPO documents to the U.S. Securities and Exchange Commission (SEC), seeking to be listed on Nasdaq.

WeRide was founded in 2017. According to its prospectus, the company has completed 10 rounds of financing to date, with a disclosed amount of over US$1.09 billion. In November 2022, when it completed its last D+ round of financing, its valuation reached US$5.1 billion.

Two other unicorns also disclosed their plans to go public this year. On June 17 and April 22, the China Securities Regulatory Commission received the filing materials of Momenta and Pony.ai, respectively, and they planned to issue no more than 63.3529 million and 98.1495 million common shares in the U.S. stock market.

In terms of Hong Kong stocks, two intelligent driving solution companies have also appeared this year. Youjia Innovation and Zongmu Technology both "changed course" from A-shares and submitted their applications to the Hong Kong Stock Exchange on May 27 and March 28 respectively.

"Some shareholders expressed the hope to hold on a little longer and find the best time for IPO, while others, under pressure from LPs, suggested IPO as soon as possible or demanded dividends." Talking about the current IPO boom in the smart driving industry, an executive of a smart driving company told China Newsweek that from the perspective of enterprises, the autonomous driving industry must be a long-distance race rather than a 100-meter sprint, but the patience of capital may be only five or six years. "If it is still in a state of continuous losses, capital may be uncomfortable."

Intelligent driving usually includes two parts: ADAS (Advanced Driver Assistance System) and autonomous driving. The upstream of the industry chain mainly includes industries such as sensors, chips, algorithms, high-precision maps, and intelligent driving solutions. The midstream is the automobile OEMs, and the downstream is the service market derived from the upgrade and operation of intelligent driving technology.

According to data from MAX, a private equity firm under Zero2IPO, the years with the most financing events in the field of autonomous driving in China were 2021, 2022, and 2018, with 162, 139, and 108 financing events, respectively. There were 92 financing events in 2023, and as of July 29, there have been 45 financing events so far this year.

The speed of financing has declined significantly, and funding difficulties are putting pressure on every player in the smart driving industry chain. Listing may become a "life-saving straw".

The IPO prospectuses that have been submitted intensively this year have revealed the long-standing difficulties of the industry. In the past three years (2021-2023), WeRide's revenue has increased from 138 million to 402 million yuan, and its annual losses were 1.007 billion, 1.298 billion, and 1.949 billion yuan, respectively. In addition to the loss of 880 million yuan in the first half of this year, the accumulated losses in three and a half years have exceeded 5.1 billion yuan.

WeRide's business is quite diversified. In terms of hardware services, it provides Robobus (driverless buses), Robotaxi, unmanned street sweepers and related sensor kits; in terms of services, it includes L4 application services and ADAS software solutions.

Its R&D investment is still growing. WeRide spent 443 million, 759 million and 1.058 billion yuan in the past three years, and spent 517 million yuan in the first half of this year. In terms of gross profit, it increased from 37.4% to 45.7% in the past three years, and dropped to 36.5% in the first half of this year. As of the first half of the year, WeRide had 1.829 billion yuan in cash and cash equivalents on its account. At the above-mentioned "burning money" rate, it may only cover the losses of one or two years.

Although sales have been growing year by year, it is still difficult to cover the high R&D investment. The company has been making losses for the past three years, which is a common pain point for companies planning an IPO recently.

The prospectus shows that Youjia Innovation's revenue has increased from 175 million yuan to 476 million yuan in the past three years, with a compound annual growth rate of 64.9%, and a cumulative loss of 568 million yuan. Zongmu Technology's revenue has increased from 225 million yuan to 498 million yuan, with a compound annual growth rate of 48.7%, and a cumulative loss of 1.586 billion yuan.

Youjia Innovation mainly provides L0 to L2++ intelligent driving solutions to vehicle manufacturers. It raised a total of 1.448 billion yuan from 2015 to 2023, with a final valuation of approximately 5.348 billion yuan. It ranks sixth among China's intelligent driving solution providers in 2023, with a market share of 0.6%.

When talking about the reasons for the losses, both companies mentioned the procurement costs of raw materials and consumables, as well as a large amount of R&D expenses. Take Zongmu Technology as an example. As of the end of last year, the company had a total of 901 full-time employees, of which 58.2% were in R&D. In the past three years, R&D expenses accounted for 120%, 71.3% and 73.9% of revenue respectively.

As the upstream and downstream of the current vehicle manufacturers, another core factor for the loss of the intelligent driving business is that the vehicle market is too competitive. Challenges such as "fierce market competition" and "revenue dependence on a small number of customers" also frequently appear in the prospectuses of the above-mentioned companies.

Dai Yifan, assistant director of Tsinghua University's Suzhou Automotive Research Institute, said that because the market is too competitive, vehicle manufacturers are forcing suppliers to lower prices, and the suppliers' survival space is limited, which affects subsequent research and development.

Chai Daixuan, managing director of CIC, analyzed to China Newsweek that the products of first-tier suppliers of intelligent driving systems must be determined based on the specific needs of downstream customers and vehicle models. Sales of supporting models lower than expected and premature replacement of models will have an adverse impact on the company's revenue and profits.

"In the early days, car manufacturers might spend tens of millions of dollars on the development of a car model, but now it may be reduced to a few million yuan. The company covers the development costs, and will suffer losses if it fails to sell in large quantities in the later stage. However, the probability of increasing sales of passenger cars nationwide is getting lower and lower. The average annual sales of each model may not exceed 10,000 units, and it is becoming increasingly difficult to share the development costs." said a senior industry insider.

For companies that have already gone public, the crisis has not yet been resolved.

Zhixing Technology, the "first stock in autonomous driving", has relied on a single major customer in the past two years, with revenue from Geely Group accounting for more than 95%. In the past three years, the company's net loss attributable to the parent company has narrowed from 464 million yuan to 195 million yuan year by year, but its stock price has continued to fall. On July 17, its stock price fell from HK$84.1 per share to HK$26.4 per share at the close, a drop of 68.29%.

Hesai Technology was listed on the Nasdaq on February 9 last year. It is a lidar developer and manufacturer. As of July 26, the stock price closed at US$4.6 per share, which has shrunk by 78.2% compared with the closing price of US$21.1 per share on the first day of listing. The 2023 financial report shows that the company's annual revenue is 1.88 billion yuan, a year-on-year increase of 56.1%; but the net loss attributable to the parent company is 476 million yuan, a year-on-year increase of 58.2%.

Another SoC chip company also submitted its application to the Hong Kong Stock Exchange this year, and it also could not escape the problem of "burning money without making profits". Horizon Robotics stated in its prospectus that in the past three years, the company's revenue increased from 467 million yuan to 1.552 billion yuan, while R&D investment increased from 1.144 billion yuan to 2.366 billion yuan, with a cumulative loss of 17.523 billion yuan.

According to data from Gasgoo Automotive Research Institute, in the top ten ranking of China's intelligent driving domain control chip installation volume in 2023, Horizon ranked third among suppliers with 402,000 chips, accounting for 11.4% of the market share; Tesla and Nvidia ranked first and second with 1.208 million chips and 1.147 million chips respectively.

"Due to the need to continuously spend money on research and development, it is necessary to capitalize quickly. When a large amount of funds are raised in the primary market, it is necessary to go public as soon as possible to continue raising funds in the secondary market." Zhang Chi, chairman of Xinding Capital, commented.

Commercialization is the hard truth

After spending so much money, the biggest concern for startups is whether they can successfully run a commercial closed loop.

Zhang Dezhao believes that for ToB high-tech companies, a gross profit margin of 20% is the "bottom line."

"If we calculate it at 20%, and further amortize the pre-sales and post-sales sales expenses, ideally it can be controlled at around 5%. If it is slightly uncontrollable or reaches 10%, and then deduct 3%-4% of management fees, and the rest is used to amortize the research and development expenses, it may be difficult to support further investment," he said.

In order to verify step by step whether the needs of the scenario are "rigid needs" or "pseudo needs", Ma Wei said that the company needs to have a set of risk reduction methodologies to keep getting closer to the answer.

"We will use four closed loops to verify business value three times. In the first one, we will only spend 10% on R&D to quickly achieve a single-point closed loop; in the second integrity closed loop, we will confirm the function and use value; in the third perfection closed loop, we will confirm the product value, such as cost and profit; in the fourth reliability closed loop, we will confirm the quality and mass production value." he said.

Wu Gansa revealed the company's set of general standards for assessing whether the business has formed a commercial closed loop, including: the vehicle scale reaches several hundred vehicles; the number of remote operation and maintenance personnel per 100 vehicles is less than 1; the operating mileage excluding safety officers exceeds 1 million kilometers; orders worth hundreds of millions are obtained; there are more than 20 large customers, achieving stable gross profit and improving year by year.

Around 2022, as the progress of technology implementation was slower than expected and it was difficult to achieve large-scale growth in the short term, investment in L4 level products dropped significantly.

Self-driving companies in the United States also faced a wave of bankruptcies. For example, in October 2022, Argo AI, a star company, announced its closure and dissolution, with its employees and some parts being taken over by investors Ford and Volkswagen. In March 2023, Embark, which focuses on the truck business, announced layoffs and the end of operations. Founder Alex Rodrigues said in an internal letter that "the capital market has abandoned unprofitable companies."

For local manufacturers, new adjustments have taken place: companies pursuing L4 goals cooperate with OEMs to "reduce dimensions" and engage in L2-L3 pre-installed mass production; companies focusing on Robotaxi have also increased their investment in commercial vehicles and segmented scenario businesses.

Take WeRide as an example. Robotaxi is its first scene to enter autonomous driving, and the first fleet was established in 2019. In the following years, it has gradually entered the commercial vehicle scene and cooperated with Bosch to promote intelligent driving software solutions for passenger cars. The prospectus shows that in 2021-2022, its product revenue accounted for 73.5% and 64.0% of the total revenue respectively; while in 2023 and the first half of this year, service revenue has dominated, accounting for more than 86.0%.

"The advantages of passenger cars are that assisted driving is easier to scale up and obtain more high-value data; the verification scenarios are wider and the technical ceiling is higher; and the established supply chain is stronger." Wu Gansa said that the pre-installed mass production of passenger cars adopts a one-time fee model, and the cash flow model of commercial vehicles is similar to subscription services or SAAS.

"Robotaxi is an asset-heavy industry. Companies need to take into account the functions of taxi operating companies. The operation difficulty and investment intensity are higher," Zhang Chi commented.

However, this year, with the introduction of key policies and the expansion of trial operation scope in cities, L4 Robotaxi is accelerating its road access. On July 3, the Ministry of Industry and Information Technology and other departments jointly announced the list of pilot cities for the application of "vehicle-road-cloud integration" of intelligent connected vehicles, which includes a total of 20 pilot cities (consortiums). According to Pacific Securities' forecast, the scale of China's Robotaxi market is expected to exceed 1.18 trillion yuan and 2.93 trillion yuan in 2025 and 2030 respectively.

WeRide stated in its prospectus that as the company tests, experiments and commercializes its autonomous driving technology, R&D expenditures are expected to increase. Among them, Robotaxi will be a key investment project, "commercial production of Robotaxi will begin this year and next year, and preparations will be made for large-scale commercialization."

The reshuffle is about to begin

When talking about the IPO boom, the experts interviewed generally said that the intelligent driving industry is accelerating into the second half of "survival of the fittest", and listing can be regarded as an "entry ticket".

"In recent years, the industry characteristics of large investment, long cycle and high technical barriers have gradually begun to converge. Mainstream car companies have begun to conduct self-research, and powerful Internet technology companies and ICT companies have entered the market, which has brought a great impact on the industry." Wang Xianbin, vice president of Gasgoo Automotive Research Institute, told China News Weekly: "In the future, only companies with a large number of software technicians, a large amount of capital, and experience in engineering mass production projects can continue to invest in development for a long time."

"On the one hand, there is the revolution brought about by ChatGPT, from traditional labeling to rapid learning with large models, and the underlying technology and module algorithms have also undergone iterations; on the other hand, Tesla is open source, and the threshold for autonomous driving has been lowered." Zhang Chi said.

He said that innovative companies need to build financial barriers through listing. "It is time for the industry to reshuffle, especially for the leading companies. Companies that start from L2 and L3 levels and work in closed scenarios may be relatively safe, but companies that directly engage in L4 and L5 autonomous driving still face risks."

Dai Yifan said that the competition in the second half will depend on whether the technology and products have entered the mass production and application stage. "L2-L3 assisted driving has begun to be installed on a large scale and is in the early stage of explosion."

Talking about the biggest changes in the next few years, he believes that there are too many players in the fields of complete vehicles, parts and autonomous driving, and there is overcapacity. They will go through a period of mergers and elimination, and the number of technology companies in complete vehicle and supplier brands will decrease. "Capital is no longer chasing new concepts like before, but is more concerned about commercialization, such as pre-installed mass production capacity and Robotaxi operation."

If we want to break through from "marathon" to "boxing match", what is the key to the future breakthrough?

"For startups, there must be things they should do and things they shouldn't do." Wu Gansa said that the future challenge will still be to continue to refine the versatility of automatic solutions and reduce operating and maintenance costs.

"Just like a marathon race requires controlling the pace and constantly finding supplies to run to the end." Zhang Dezhao mentioned that stable operation is the key at the moment, and the company's business strategy this year is to "go from disorder to order."

Ma Wei said that unmanned operations have broken the old barriers to competition. In addition to "surviving", companies must also make good use of the "dragon-slaying sword" brought by artificial intelligence, find suitable scenarios, and continue to break through the traditional market.

"Because technology is iterating too quickly, it is not realistic to focus on too many markets. Companies need to clearly define the focus of their main products, and each company may only have one or two main products," Dai Yifan suggested.

"The biggest challenge still lies in whether the technological route can keep up, such as the investment in pure visual end-to-end large model technology, the selection of multi-sensor fusion technology, the construction of supercomputing capabilities and data closed-loop capabilities, etc." said Wang Xianbin.

Since the beginning of this year, capital attention has warmed up. According to statistics from the New Strategy Low-Speed ​​Unmanned Driving Industry Research Institute, in the first half of the year, nearly 103 major investment and financing events were disclosed in the field of autonomous driving at home and abroad, with a total disclosed financing amount of nearly 26.1 billion yuan (including IPO fundraising), both higher than the same period in 2023, involving 82 domestic companies and 16 foreign companies.

"The growth rate of the intelligent driving market is relatively optimistic, and there is a lot of room for domestic substitution. The proportion of domestic capital is increasing, leaving ample room for capable manufacturers with commercialization capabilities." Chai Daixuan predicts that in terms of intelligent driving solution revenue (including software, hardware, algorithms and subscription service revenue, etc.), the Chinese market will reach 71.74 billion yuan in 2023 and 212.80 billion yuan in 2028, with a compound growth rate of 24.3%.

Published in the 1151st issue of China Newsweek magazine on August 5, 2024

Magazine title: Has the smart driving unicorn become a money-eating beast?

Author: Wang Shihan