“The second phase of financing is in the range of $100 million to $500 million.”

The “Automatic Driving 36” special report will talk about the innovation of the autopilot track in the form of Q&A interviews. Continue to interpret the changes, controversies and future of the industry.

The autopilot industry is at this stage of the day, and all companies realize they can’t “single-handedly”. As a result, some autonomous driving startups began to look for partners from the upstream and downstream of the industry chain, to do deep binding, and form an “iron triangle” model. PlusAI is one of them.

Intelligent Technology was established in Silicon Valley in the United States in 2016. It is an L4-class driverless of the heavy truck trunk highway. It currently has R&D and operation centers in Beijing, Shanghai and Suzhou.

The company founder and CEO Liu Wanqian told,The autopilot industry will eventually be compressed into three links: operators, vehicle manufacturers, autopilot providers, and The positioning of the technology is the technology provider. This is the same as the communications industry, with operators, equipment providers and mobile phone terminal providers, but the DNA of these three types of companies is completely different, it is difficult to cross the border to grab other people’s business. Therefore, Zhijia Technology will not operate its own fleet.

According to this idea, Zhijia Technology is bound to the Chinese head heavy truck enterprise “FAW Liberation” in the upstream —— In January of this year, FAW liberated the first batch of 10 J7 smart trucks delivered to Zhijia Technology, and announced that they will cooperate in the commercial operation of trunk logistics; downstream, it is tied The logistics capacity platform with a market share of nearly 80% in China is “full” —— In November last year, Zhijia Technology announced that it had reached an exclusive strategic cooperation with Manbang Group and got a full A+ Round financing and full round of financing.

Liu Wanqian said that the full help is to provide them with money and supply. On the other hand, the full help is equivalent to the auto-driving business layout exclusive to Zhijia Technology, but their current cooperation test The operation is still at the level of “knowing the demand for freight”, which is far from commercialization, and it is planned to achieve mass production in the next five years or so.

“If you go to see robo-taxi, the US pattern has already formed. The three major US automakers GM, Ford and Chrysler have each chosen their own autonomous driving partner. In other words, if you want to do today, Driving automatically with a car, no car factory, how do you play? You don’t have a car,You can have technology, you have nothing to lose. Liu Wanqian said.

Automatic driving 36 people | Zhijia Technology CEO Liu Wanqian: There are 5 stages of automatic driving, we just entered the second stage

Talking about whether SAIC’s investment has brought strategic cooperation resources. Liu Wanqian said that they have provided SARS solutions to SAIC, but at the end of 2017, they began to focus on L4 autopilot. Now, the two sides have cooperation at a higher technical level, but there is no product.

Intelligence Technology received the autopilot road test license issued by the California Vehicle Authority (DMV) in 2017, and announced on November 30 last year that it was the first domestic auto-driving road test license for operating trucks. , began to do road test in Jiangsu.

Liu Wanqian said that China’s social logistics costs are about 13 trillion yuan, accounting for about 15% of GDP. Among them, trunk logistics has a market scale of 3 trillion, covering 8 million heavy trucks. Industrial efficiency can be increased by 25%.

Talk about how autopilot reduces the cost of heavy truck operations, Liu Wanqian said that in the cost structure of heavy trucks, the United States is “433” – 40% of oil costs, 30% of labor, Vehicle depreciation and other expenses account for 30%, while autopilot can save 30% of labor and 10% of fuel cost, adding up to 40%; in China, the cost structure is “3322”, that is, 30% is road and bridge The fee is 30% for oil, 20% for labor, 20% for vehicle depreciation, and autopilot can save 20% for labor and 10% for fuel, which is 30%.

The following is a partial interview with Liu Wanqian (remaining and abridged without changing the original intention)

Question: How do you evaluate the company’s current business progress?

Liu Wanqian: We think that L4 should be divided into five stages. The first stage is called “prototype period”, which proves that this thing is worth doing, and puts forward the general technical direction. It is best to find some teammates; second The stage is called “engineering verification period”, which is to build “design system” and “verification system”; the third stage is “design verification period”; the fourth stage is “production verification period”, which is to build factories, assembly lines and industrial chains; The fifth stage is “production”. We just graduated from the first stage today and entered the beginning of the second stage.

To prove that autonomous driving is safer than human driving, at least 10 billion miles of road testing is required. Waymo now has 200 cars, and it took 10 years to get 10 million miles of mileage. If you want to reach 10 billion miles, it is 1,000 times worse. That is to say, if you still use the same drive test method, it will take 1000 years. Verify that today’s design is safer.

Someone said that if the team is expanded by 100 times, can it be verified in 10 years? Still can’t, because every change to a line of code will affect the security of the entire system, and must be re-tested. So, The purpose of the road test fleet is not to test security, but to collect as much data and scene as possible. So,Today my test team, I hope it often find the case better, I can’t wait for me It is a problem to test the vehicle for 100 meters, because I can collect a variety of test cases.

Question: Why is it necessary to drive a road test of 10 billion miles? How did you figure it out?

Liu Wanqian: In the United States, a traffic accident occurs every 10 million miles. We believe that if there is no fatal accident if the autopilot has been opened 1,000 times of 10 million miles, it can be explained that it is well opened.

Automatic driving 36 people | Zhijia Technology CEO Liu Wanqian: There are 5 stages of automatic driving, we just entered the second stage

(Source: Zhijia Technology)

Question: How long does it take for each of these five stages, and how much can it be done?

Liu Wanqian: The project verification period is about one year to one and a half years. We have already done it for half a year, and it is estimated that it will take about one year. The design verification period is actually what Google has done in the past three years, because our scenes are special, the design verification period should be slightly shorter than Google, and Google has done it before, we can step on the pit, but Google is a very good The company, offset on both sides, I think it may be 2-3 years. The final production verification period is to build a factory and industry chain, which may take 1-2 years. Production verification period is the largestThe problem is that the investment is very large. The prototype period is about 10 million to 100 million US dollars, the engineering verification period is 200 million to 300 million US dollars, and the design verification period is 500 million to 100 million US dollars.

Waymo should be able to remove the full driver next year. Basically, the design verification process is completed. You can move on to the next stage and start mass production. Waymo is definitely our benchmark, we hope to keep a distance from Waymo, don’t be fooled by it.

Question: When can you remove the security officer?

Liu Wanqian: We are now equivalent to Waymo in 2015 and 2016.

Question: Do you think it will take another three or four years?

Liu Wanqian: Right.

Question: Can you profit if you withdraw your security officer, right?

Liu Wanqian: Profit is not our goal in the foreseeable future. As long as we graduate from mass production, the company will be worth at least $100 billion.

Question: What is the experimental cost of a car?

Liu Wanqian: The cost of the experiment, regardless of cost, costs 10 million US dollars for a car. If it can be safer than people, it will be finished. The cost of a car can not be more expensive than today, because that is the metal, IT is valuable in it.

Question: How many cars do you have now? Are you buying it yourself?

Liu Wanqian: There are ten cars in China and the United States. All are bought by myself.

Question: I also plan to buy a car with financing money in the future? Still through financial means?

Liu Wanqian: The car is not expensive, one car is 100,000 US dollars, 20 cars is one or two million dollars, mainly because of the large operating expenses, the data collected by the car must have a system to collect. If the proportion of accidents in a car is X, 10 cars are 10X. In the research and development stage, the accident rate must be zero. If an accident occurs, not only the company but the entire industry will have to stop for half a year to one year, which is very serious. So we must have two people in each car, and we are professional people.

Question: When can I have income?

Liu Wanqian: When we enter the design verification period, there will be hundreds of cars. My test team can pull money to make money, this is the benefit of doing heavy trucks. Because Waymo’s users are passengers, passengers value experience and comfort, passenger experience is not good, no wayTo collect money, and our users are shippers, just pull the container from point A to point B, so you can make money when testing.

Question: In March 2018, some media reported that you have reached technical cooperation with Chery Automobile to share resources and channels in the field of intelligent driving. How is this cooperation going?

Liu Wanqian: Chery wants to drive automatically, then find us. The goal at the time was to do robo-taxi, and finally to reduce the dimension to L3. It is okay for ADAS to make a company with two or three hundred million dollars, but how can it be explained to investors?

Ask: Will L2-L3 be done in the future? This can have both technical output and revenue.

Liu Wanqian: We are not totally opposed to licensing some of these technologies to partners for commercialization. Things that can help improve the entire industry can be considered.

Question: Some companies are doing L3. They think that L3 can replace a heavy truck driver, which is equivalent to saving costs.

Liu Wanqian: We studied it carefully two years ago. Can we replace a driver? I think it is possible to really enhance the driving experience. The problem is that the threshold for creating this value is low. How do you get and retain this value? The L4 thing can’t be done, it is estimated that a company, at best two companies, or two camps to eat the whole cake. Because there are any companies or camps that can afford billions of dollars to do this?

Automatic driving 36 people | Zhijia Technology CEO Liu Wanqian: There are 5 stages of automatic driving, we just entered the second stage

(Source: Zhijia Technology)

Question: What is your business model?

Liu Wanqian: Doing operations. For example, if you pull a box of goods and earn 1000 yuan, a driver’s car needs to spend 995 yuan to earn the 1000 yuan, and our autonomous driving can spend 700 yuan to earn this 1000 yuan, saving more than 200 pieces. money. But you will say, who wants to earn this money? Truck companies, logistics platforms and we all want to make this money, and we all have their own reasons. Since there are reasonable problems, there are reasonable solutions. Since these three families are not able to do the same, then we will take it apart and earn some money.The last is the proportional problem, and then I will say it later.

Question: Is it possible that the gang will buy you? Or do you definitely want to operate independently?

Liu Wanqian: We want to operate independently, and we hope that we will operate independently. Uber originally wanted to eat this cake himself, but he still couldn’t eat it. Because it was an industrial resource, it only saw one.

Question: Is the full help like this?

Liu Wanqian: This is what the senior executives said. We made a strategic integration last year because we were highly consistent with the development of the entire industry at the top level, and there was no way for everyone to cooperate.

Question: Does the United States have such a capacity platform? Do you have similar partners in the United States?

Liu Wanqian: The United States is not the same. The pattern of the freight industry is “811” in China, 80% is a personal merchant, 10% is a big team, 10% is a franchise team, very scattered; while the United States is “235”, 20% is a small team, 30 % is a big team, 50% is a self-operated team. Therefore, the United States does not have a large capacity platform like a full gang, but there are hundreds of big teams, and one team is 10,000 cars. We are now working intensively with two of the US commercial vehicle plants to understand their logistics systems and requirements during the test run.

Ask: Many people think that now is the capital winter, financing is getting harder and harder, what is your feeling?

Liu Wanqian: I personally feel that the capital market is very good, but the feelings of different people are different. There are five stages in the autopilot industry. During the prototype period, investors and companies are doing “validation”. It is feasible to invest millions of dollars to tens of millions of dollars, so there are a lot of companies at this time. . If you are investing millions of dollars in an autopilot company today, my advice to investors is not to play, because today you can still invest in the first stage of the company.

Question: Is there a new financing plan recently?

Liu Wanqian: The company that entered the second stage, (the money to be melted) must be the level of 100 million to 500 million US dollars. There are very few companies that can take this money, and there are very few targets that can be invested. There are only three big scenes for autopilot: robo-taxi, trunk logistics, terminal distribution. The targets that can be cast in these three big scenes, no more than two in each scene.

Question: In Zhijia TechnologyBefore, you have created three companies. Is there any entrepreneurial experience and business logic that can be applied to this company?

Liu Wanqian: The first company, InformLink, was the recommendation engine. At that time, the market share in the United States was 70%, which was later sold. The second venture was RMG Networks, the nation’s largest outdoor media network, such as the aircraft seat back and the gym’s TV show, which went on sale at the end of 2011. I served as CTO. The third time was to do mobile internet game company RedAtoms. The three companies were “technical-led industry innovations” at that time.

What I learned is that any company will definitely fall out of the pit. The key is to be able to stay alert and climb out of the pit quickly. With this awareness, what may be left is IQ, which is a network in China.

Question: What pit have you lost in Zhijia?

Liu Wanqian: There are more pits. In 2015-2016, Tesla and Waymo had factional disputes. The two of them were worse than anyone else. No matter which side they stood, they had a 50% chance of falling. Our peers, from Vision only to Vision first, have some people who have been delayed for one or two years. Because the technical direction is wrong, it will affect who you are looking for, what partners, what investment, and if you run far, you will know that it is wrong. Still have to go back to find compensation, this is the biggest pit.

We had very few PRs in the early days, because many things were uncertain. You will definitely ask me “what is your technology?” If I say “I didn’t think about it”, you will definitely feel that I can’t. Some companies came out and said, “I am doing autopilot technology, I can support all scenarios.” Even now, the company says so, I think this is a huge pit. Because if you are so positioned, it is unclear who your customers are. We feel that autonomous driving is highly scened and super-segmented. You have put in a scene, and you have been in it for the rest of your life. This can’t be found, it is zero. It’s hard to say that I used to be Robotaxi, and now I am going to do heavy trucks.

Question: Have you tried other scenes besides heavy trucking?

Liu Wanqian: We have seen a very subdivided scene at the time, there are parks, robo-taxi, trunks, distribution, ports, agricultural machinery, mines, cleaning… We have listed a table with 16 scenes The technical direction is different. The upstream and downstream of the industrial chain are different. Whoever builds a car is different, and the customers are different. For example, you have to clean up, that business