Ideal CEO Lee wants to try to teach you how to build cars

"Do a stupid thing every round, and do another upgrade as long as you are alive."

Text | Cheng Manqi Zhang Jiahao Editor | Song Wei

At yesterday's ideal car spring media sharing meeting, Li Xiang, chairman and CEO of Ideal Motor, spent more than 100 minutes explaining from the beginning why to create ideal, how to disassemble needs, meet needs, and further fall on R&D, products, delivery and management and organization. He seems to be trying to teach field reporters how to build cars.

Li Xiang said that every wave of new employees come in, he has to give lectures, and he gives one-on-one lessons for employees at P8 level and above.

As for the information of the annual report and quarterly report released just 3 days ago, Li Xiang did not take the initiative to mention it. Compared with business issues such as sales volume and market share targets, and whether it can be profitable this year, Li Xiang wants to explain internally and externally who this company is and what the underlying value is; Share his latest insights into the current stage of market competition, the intelligent future of automobiles and artificial intelligence, and how ideals will think and do in these trends.

This is not so much the expected annual report season communication, but more like a business and business class. But without paying the business school tuition, you can hear the elaboration of "Teacher Li Xiang". This aligns with his understanding of "leader," whose most important ability is to "get people to understand you enough."

The following is the collation that Li wants to share, and "Later" has been partially deleted:

Smart electric vehicles are one of the few big opportunities that involve energy + information at the same time

The automotive industry has developed for more than 100 years, and we have seen the high-speed expansion of mature enterprises, Toyota, Honda, Mercedes-Benz, BMW, which is the stage of enterprises from 10 - 100.

We see how successful these companies are, but when we really start our own business, we don't know how to build a car company from 0 to 1 and then from 1 to 10, because most of us have not yet been born when those companies go through these stages.

The last 7 and a half years have been a complete validation process for me, and we have achieved more than $10 billion in revenue with an ideal ONE.

So here I want to share with you, as a smart electric vehicle startup, from 0 to 1 what we think, what judgments have we made, and why behind it? And when we look at the scale of an enterprise to hundreds of billions of revenues, from 1~10, what challenges will we face and what problems will we solve?

First of all, why did we start this company? To start a new business, it must depend on the core trend. The most important factors influencing the development of mankind over the past few thousand years have been energy and information.

From the beginning of human beings to drill wood for fire, to the use of fossil energy, the use of electric energy, and then to the application of renewable energy in recent decades, this is the process of human civilization progress.

Another process of progress in human civilization is the development of information, that is, information technology. Humans are different from other animals in that they have language and can communicate; Then there is printing, computers, and now the rapid development of artificial intelligence.

Why are smart electric vehicles getting attention? Because it is a rare industry that involves both energy and information technology. This is an important opportunity.

Energy: Ideal to solve the problem with extended range plus high-voltage fast charging

The first thing I talk about is energy, how to understand energy? As a superpower, energy must be guided by the government. So our core is to look at the needs of users and governments, as well as the challenges and problems behind them. If we have the ability to solve this problem, we can gain market acceptance.

From the user's point of view, energy demand is three levels: the first is the convenience of energy acquisition, we often see users give bipolar evaluations of electric vehicles, those with charging piles will say that electric vehicles are really fragrant, and those without charging piles regret buying electric vehicles. Why polar evaluation, because users feel different convenience.

The second is the cost of energy acquisition. One is whether the product I bought is more expensive, here is the price of the battery. The other is the cost of use, the cost of electric vehicles is very low, when the economy is no longer 10 points of high-speed growth, down to 5 points, more and more people will choose electric vehicles, because the money saved every year can buy an iPhone. As economic growth slows, savings are key.

The third is comfort and environmental protection. Environmental protection here is different from the government's carbon neutrality and carbon emissions. The point of environmental protection that users think is to drive a car to pick up children from school, turn on the air conditioner in the summer, the smell of fuel vehicles everywhere, this is pollution, if it is an electric car, it will be good to turn on the air conditioner. Our user has a characteristic, when the user has both a fuel car and an electric car at home, his driver will be more willing to use our car most of the time, because the driver has a lot of waiting time and needs to turn on the air conditioner in place. The comfort and smell of air conditioning in fuel vehicles and electric vehicles are completely different. The better these needs are met, the higher the penetration rate of electric vehicles. In the final analysis, the core purpose of electric vehicles is to gradually replace the current 200 million or so passenger cars.

The government's needs are also three aspects: the first is that the core technology is independent and controllable. This is critical when competing with great powers, and many times it's not just that you are obedient. Japan is also obedient, but Japan has been in decline since the Plaza Accord. Because any boss has the motivation to curb the second boss, it has nothing to do with how you express your determination and attitude.

The second is the security of energy supply. We produce more than 60% of the world's lithium batteries, but our lithium reserves are less than 10%, and we have more than 20% of the world's oil, but our oil reserves are about two points and less than three points, which is the security of energy supply.

The government's third demand is that "gold and silver mountains are not as good as green water and green mountains", that is, carbon peaking and carbon neutrality. This makes China's entire industrial chain, whether it is electric vehicles, motors, electronic controls, third-generation semiconductors, new battery technology, and photovoltaics, has reached a leading position in the world. Because the government has such a vision.

As a smart electric vehicle company, we now face two challenges, the first is the difficulty of charging, and the second is the high cost of batteries.

Batteries are composed of precious metals and commodities. The better an electric car sells, the more expensive the battery will be. In January and February of this year, just like a friend on Weibo said that although (Tesla) cut prices did not make themselves victorious, they could hit opponents. Two months have passed, and almost every company has generally achieved only 4% or 5% of its full-year goals. At this time, the price of lithium carbonate began to decline significantly, from 550,000 tons to 390,000 tons now.

If we believe that electric vehicles, including energy storage, will develop rapidly, the cost of lithium carbonate will definitely not be reduced to the original 450,000 yuan, and 20~300,000 tons will be stable for a long time. In this way, the cost of batteries still cannot be greatly reduced, and even the cost of batteries today is much more expensive than in 2018.

Our solution has three paths:

The first is power exchange. Power exchange is close to the experience of refueling, and many of the problems that users are worried about can be solved. On the other hand, battery replacement can also not sell the battery to consumers, so the cost of buying an electric car is even lower than that of a fuel vehicle.

However, power exchange requires the construction of exchange stations, and the risk of large fluctuations in battery prices is required. We didn't have much money at the time. We were wondering if we could put it directly on the product side without relying on the server side, so we made a range extension.

At that time, the range extension was different from today's extended range, which was generally 50~60 kilometers of endurance. How do we combine user experience and travel efficiency? One of the best choices we made was to make the battery bigger.

All extended-range electric vehicles on sale today have a pure electric range of more than 200 kilometers, so as to achieve a scenario that we especially hope to achieve: electricity in the city, high-speed oil consumption. We have sold 290,000 vehicles, but there are more than 200,000 home improvement charging piles, and the proportion of charging piles installed by our users is very high.

So range extension is one of our solutions, it can use less battery, can also control the weight of the car, and solve the charging problem of long-distance travel, in line with the positioning of our SUV.

What is the third way? The vast majority of corporate charging piles are not profitable. Because the charging pile has to do 6~7 orders a day to break even. With 10 orders, profitability will be better than gas stations.

But today's charging piles are generally 2~3 single / day, almost all of them lose money. So the common algorithm is to allocate the cost of charging piles that can only be used for an average of 5 years to 10 years, so as to increase profits and reduce losses.

The principle of the charging pile is actually the same as that of the restaurant, which is a table turning rate. You charge for an hour to fully charge, at this time, if the user comes back from eating in the middle, the two or three hours have passed, which can easily cause 2~3 orders a day. We calculated for ourselves that if we can reduce the charging time to less than 20 minutes, the user will not leave the car. If it is shortened to 10 minutes, the whole experience is basically the same as that of a fuel car.

That's what we're doing – fast charging. It has two cores: one is that a high-voltage platform must be used, which can achieve such a charging speed. The high-voltage platform can also bring the second benefit, when we use silicon carbide with the high-voltage platform, and then with better packaging technology, the battery cost can also be greatly reduced, and the efficiency will be significantly improved For example, the best selling medium and large SUV at present, it is about 100 kWh, to achieve 600 kilometers of endurance. We use an 800 V high-voltage platform with a better drag coefficient and silicon carbide and overall efficiency optimization, which can probably achieve the same range with 80 kWh. The vehicle will also be lighter, and the overall cost can be reduced by 3~40,000 yuan. We ourselves are now making silicon carbide modules and making our own three-in-one motors, which are closely related to the overall efficiency.

After doing these things well, we can reduce the cost of parts for a car by 3~40,000 yuan compared to today's mainstream 400 V platform in the same size and drive form. At that time, it can be seen that our 800 V platform electric vehicles can achieve the same price as the extended range, which may not be the same as everyone imagines. Today everyone thinks that 800 V silicon carbide is more expensive, in fact, it can save a lot of costs. 4C is not taken into account, it also has a certain degree of selectivity, we can charge 400 kilometers in 10 minutes with 4C batteries, and 2C batteries can also charge 400 kilometers in 20 minutes.

We ourselves have always believed that if we want to sell electric vehicle charging piles, it is a product itself for us, not a service. If we provide a 4C car, but do not have a 4C perfect charging network, it is equivalent to selling a 4G mobile phone, but you still have a 2G network. So we don't have to worry about the speed and determination of us to lay charging piles, and it's not as expensive as everyone thinks.

Therefore, extended range and high-voltage fast charging are the two core routes that we will still adhere to at the same time until 2030. Whether it is extended range or high voltage, the core purpose is to solve the charging problem and give users two different choices.

So thinking about energy really helped us stand up from the 0~1 stage in the past and allowed us to obtain considerable income.

In the past few months, Li Auto has been the top three of all new energy vehicles in China by turnover, and above us are BYD and Tesla. So our revenue scale and growth rate can support us to go all out to do a good job in electricity. Our R&D investment in the above, including self-developed self-made parts and deep supply chain, may be deeper than everyone thinks.

Smart: No one will buy a high-end car in 2024 without advanced assisted driving

In Ideal Auto, there are four AI teams working on four areas, including cockpit multimodality, which was established last year; There are also intelligent driving, the transformation of the factory and the transformation of the retail end.

In the factory, we can have all hundreds of inspectors replaced by machines this year, and we are doing the core AI algorithms for this ourselves. We have been solving quality problems with algorithms. From a quantitative point of view, if we take money to measure the cost of quality allocated to each car, we are actually significantly better than luxury brands such as Mercedes-Benz, BMW, and Audi. Behind this, the algorithm system is very critical, including our decomposition of various accidents. Our insurance prices are among the lowest in our class, even lower than petrol vehicles. Electric vehicle insurance is generally more expensive than fuel vehicles, and the ideal ONE, which costs more than 300,000 yuan, is renewed in the second year, and commercial insurance is 3,300 yuan. This is all closely related to the algorithm behind us.

The intelligence we want to do is also related to Autohome. Autohome has succeeded in almost all its products, but there is one product that has failed badly, and that is auto e-commerce. Not only did we fail, but all of our peers failed.

Why is this? We later made a particularly important summary: more than 90% of the entire business chain of automobile e-commerce occurs offline, offline warehousing, inventory, logistics, delivery and other links, we have not done any transformation, our cost is higher than traditional car dealers, we just spent 5000~15000 yuan more CPS costs online. Therefore, in the efficiency and business chain measured by money, we have not created any value except for spending a part of the extra money.

The same problem also appears in new retail and community group buying. Cruelly speaking, this is also a problem for all ride-hailing companies. You can see that including Uber and Didi, its market value today is far less than the market value of financing at that time.

The core is that "software 1.0" cannot solve the problems of the physical world. Software 1.0 is that humans make their own rules, change them themselves, and then use their own programs (whether it is the Internet or apps) after programming, which mainly runs in the digital world. We can use Uber very quickly in the U.S. to hail a car, but the cost of everything that happens between point A and point B doesn't go down, even higher than a taxi, because the empty drive rate is higher. Then there will be a problem for these platform companies, if he wants to get a high-growth share, he loses money; If he wants to make money, his only way is to exploit drivers, because the United States (platform commissions) receive 30%, China receives 25%, and rental companies only 20%.

E-commerce is different. E-commerce has changed the situation of placing goods in expensive shopping malls to sell, goods can be sold when they are moved to a warehouse, the overall cost is reduced by 35%~40%, this 35% is taken away by consumers, things are cheaper; The platform took a part, so it fed businesses like Ali and Amazon; A new generation of merchants is also profiting on it.

So I was thinking about the question, is there any way to change the physical world? I found only AI.

The core of AI is learning, not programming, not logic, AI is actually imitation and learning from humans. We have divided AI into two categories, corresponding to the two ways people work and learn in the physical world.

One is behavioral learning, with the cerebellum as the main and the brain as the supplement. We use our brains to think and train ourselves, but the cerebellum can process it after successful training. Like driving a car, and doing some testing on the production line. Many times it does not need to think completely through the brain, but relies on quick reactions. The other is the recently more popular ChatGPT, which actually includes the "ideal classmates" in the car, NIO Nomi these voice assistants. It works in the form of cognitive learning, with the brain as the main and the cerebellum as the supplement. So when you give a command, if it's complicated, it has to go back to the cloud to deal with it.

What are the benefits of ChatGPT? It's difficult, but you're not afraid of him making mistakes. If ChatGPT is talking serious nonsense at you, you don't care too much. So that everyone can use it and continue to train it. But cerebellum-based tasks, such as automatic driving, once mistaken, may be a traffic accident, a human injury.

From this perspective, I will tell you about the stage of autonomous driving. Why are we only starting all in autonomous driving at this stage?

Let's first look at how people drive themselves, how do they work through behavioral learning in the physical world?

The first is three-dimensional perception, that is, human beings obtain and process information about the entire three-dimensional world through two eyes, two ears and the brain. The imaging we see today is not the same as that seen by other animals because our sensors and our processing results are not the same. The difference between organisms and organisms is the difference in information acquisition ability and information processing ability, not who has a larger physique and stronger reproductive ability. Human beings have the strongest information acquisition and processing capabilities on the earth, so they are the hegemons of the earth.

The second step is that we can judge decisions based on what we see and our own state. If there is a car in front of me and it is an action car, I can apply the brakes to slow down, and if there is an accident car or a stone in front, I will judge what I am going to do.

The third step is to perform through the cerebellum and nervous system: how I should hit the steering wheel to avoid it, how I should apply the brakes, what kind of force to generate, what speed to steer, which is also related to my own running speed.

The fourth step is results and feedback, which is training. If the whole experience is smooth and safe, I will gradually strengthen it into my instincts. But if I accidentally rear-end and you believe me, it will be difficult to rear-end within three years, and you will adjust your behavior next time, but step on the brakes in advance whenever you encounter this situation.

In fact, learning to walk and learn volleyball when we were children was a similar process. Once you experience this kind of problem, feedback, brain training, the back cerebellum can deal with it. So when we're driving very skillfully, suddenly we take a phone call, call for half an hour, an hour, and your brain doesn't actually do anything about driving, but you're still driving safely.

In contrast to this process, the first stage that autonomous driving has gone through is as follows:

First of all, it can't do three-dimensional perception. At that time, the entire sensor and computing power were not strong, the computing power was Mobileye, and the perception part was only a forward-looking 2 million pixel camera. It can only recognize two-dimensional images in the video, and the AI used is only two-dimensional. Two-dimensional can also be ranging, that is, the wider the picture, the closer it is to me, and the narrower it is to me. However, there will be various jitter during on-board driving, resulting in inaccurate identification. Therefore, we used a relatively low-cost method, which is to put a millimeter-wave radar to measure the distance between each picture. That's the assisted driving of 99% of cars running on Chinese roads to this day. Even if it is already on the Orin high-computing platform today, it still works like this. We call this the 2.5-dimensional way of working. It is more like a bat, working by dynamically sequencing various obstacles and targets.

In the second part, that is, judgment decision-making, all judgment decisions delivered in mass production on the road today have nothing to do with AI, they are all implemented by rules, relying on a lot of trial and human programming. So whether the experience is good or not today depends on who wrote more rules and tested more scenarios. What is within the rules can be solved, but what is outside the rules cannot be solved. When there is responsibility, all car companies say that it is the responsibility of consumers, because I am only assisting driving.

The third part, the whole aspect of executive control, is also completely regular. Therefore, when you test the car, you will find that different vehicles also encounter a traffic jam, and the car that follows closely will make people motion sick, because it brakes more sharply and starts faster, but it can prevent other cars from jamming. Sitting in a more comfortable car accelerates more steadily, but it is easy to be plugged by other cars. It depends on how you adjust to a balance point.

And the fourth part, the part that feeds back growth. This is divided into two types: one is that I collect more videos, and I will have a calibration team to label this data. Tesla's calibration team is in India, and its self-driving layoffs in the past year are this part; China calibrates a large number of outsourced in Guizhou. The marked (system) is recognized, knows that it is a traffic pile, is a car, not marked, can not be recognized, it may directly hit it. So at this stage, there is often a joke about artificial intelligence: there is as much intelligence as there are people. Another part of feedback training is to adjust the rules, to adjust the decision-making and implementation rules more comfortable and safe, the way is to give more people to more scenarios.

So this previous stage, essentially has nothing to do with automatic driving, it is assisted driving, no matter how many "+" after L2, it is L2.

After that, there was an essential change, starting with Tesla's FSD, including the fact that they recruited AK from OpenAI (Andrej Karparthy, who joined Tesla in June 2017 and left last July). He made a big change after he went, because he did 3D visual perception with Li Feifei at Stanford. At the same time, the efficiency of the FSD chip is very high, it is a dedicated BPU (Tesla more often calls the FSD chip NPU, which means neural network chip; Horizon calls its chip BPU, which means brain-like chip), and although it is only 144 TOPS, two FSDs can basically be about the same as the OrinX performance of two 508 TOPS. At the same time, Tesla also upgraded at that time, installing a 360-degree camera.

At this time, it began to use large models, using Transformer (artificial intelligence model proposed by Google in 2017) to do BEV (overlook view) 3D, which became a three-dimensional perspective. And the core of the three-dimensional perspective is pixels, which is the process of pixel occupation and pixel movement in the entire three-dimensional world. This brought about a fundamental change: the car began to see the world in a similar way to a person.

With three-dimensional perception, the next judgment decisions are different. The whole judgment decision is divided into two layers, one is white-box, that is, programmed, for compliance with traffic rules and laws. The other layer is that the judgment decision, execution and control of the car will become one, which we call end-to-end. Cars can record the entire process of what humans see, what judgments they make, what executions they make, and then learn. It becomes a learning process, which is the essential change. Whether you use a GPU or a dedicated BPU, you have such a closed-loop capability.

Next, the whole feedback level is different. It will turn the whole process into structured data, which is then transmitted to the supercomputing platform for training. And the whole training is a black box process, because it is a large model. Although there are also algorithms, some algorithms are white boxes, I think it is 1.0, 1.5, and we think it is 2.0 when it is completely black box.

This also brings a change, that is, the car side learning, collect enough samples and put them in the cloud for training, then the cloud can do automatic labeling and classification, and there is no need for so many people. In fact, the number of algorithm teams of Tesla and OpenAI is very small, which is brought about by the automation of large models.

Since it is a black box, the whole training is not about right and wrong, but about the improvement of training quality. Programming is looking at right and wrong, but when machines start learning, it's looking at quality changes. Therefore, after the supercomputing terminal training, it will not be directly deployed to the car, but will return to the car to do top-level verification, and verify with a large number of human samples. When the quality of the verification algorithm is significantly improved, it will be deployed to the vehicle, which is safer.

This is the whole process, and this is where the essence of autonomous driving changes: as long as there are enough end-to-end training samples, the entire value chain, safety will get better and better. Therefore, the next competition of car companies in autonomous driving is actually three parts:

The first is to reduce the cost of the vehicle-side computing platform and sensors, so as to achieve standard equipment as much as possible. Just put a camera, you can't train anything. Indeed, Tesla's cost today is very low, its entire cost is only $1,500, and the cost of our double Orin is basically more than $4,000, which is a huge gap.

The second is who has more end-to-end closed-loop data. It's useless that you only get a portion of the data, you have to do a complete end-to-end training. Moreover, the training in different countries and different scenarios is different, and it cannot be simply moved over and used directly. That is, you can sell enough cars full of these sensors and computing platforms.

The third is the training of large models with higher costs and greater challenges. The United States has restricted the development of Chinese training chips since last year, and also restricted NVIDIA's sale of high-bandwidth training chips to China.

In fact, today, whether it is using a GPU on the car side, or using a GPU in the cloud (general computing GPU, which refers to a GPU without graphics capabilities and mainly does parallel computing), it is not the best way to perform for large models, generally only about 1/5, equivalent to another 80% of the cost is wasted, not exaggerated at all. And like Tesla to do its own AI training chip "D1", after its entire system is built, it can probably achieve the cost of NVIDIA A100 1/6.

I think this may be the part that companies determined to do artificial intelligence in the future must solve and invest in research and development, and this is the result we see. So in addition to our real product, platform and algorithm behind it, the real competition will be at the bottom in the future.

In the AI era, this includes the AI operating system, which requires extremely low latency because it runs in the physical world, and if it is delayed by 0.5 seconds and braked 0.5 seconds late, the car may be destroyed. At this time, the efficiency of the real-time operating system of the whole chain, and the operation of large model algorithms on such an operating system, becomes very critical. It also includes an on-board BPU. BPU is a typical cerebellum-based, brain-supplemented way of working. Then to the training end, it also includes cloud training with TPU, which is the super central brain in the future. We actually see the root of the entire development of artificial intelligence.

How do we do it? We are more careful, but in fact, we are making in-depth choices in this area for a long time, and after determining that this route is feasible, we will do it all in.

So our own real investment in automatic driving officially began after Tesla ran the general logic. Including why multimodal is done in the L series. The logic is the same, our in-car algorithm has gradually abandoned the previous scheme, and in-car training has also begun to use large models, only large models can achieve the 2.0 artificial intelligence I really want.

There is also a key point in this piece, which is one of our predictions: when is the real intelligent era of electric vehicles?

Today everyone is very painful, maybe you are optimistic about a lot of companies, they have developed well in this area, invested very early, but it is not reflected in sales. At this time, there must be a precise value point that allows the user experience, and the outbreak will begin.

For example, smartphones, I think it met three conditions at the same time to develop rapidly. The first is the 3G network. The second was the launch of the App Store in July 2008. Before that, if I installed software for my parents, either I went to the computer to sync with the iPhone, or went to the computer city to spend 200 yuan to install a pile, and from that moment on, my parents could directly click on the phone to download the file. Third, after the release of the iPhone 4 in 2010, OTA (remote update) function began to be provided. It used to be that the phone update was to be downloaded to my iPhone like a download part and then updated, and now anyone can directly click to update the entire phone system.

So after all three conditions are met, from 2010, mobile phones have really entered the intelligent era from the touch screen era. Before 2010, many touchscreen phones sold very well, tens of millions of units a year, but this has changed since then. Almost the same year that OTA was available, Xiaomi started making phones, and it seized the opportunity for Android to have these capabilities.

Back to smart electric vehicles, when are smart electric vehicles really smart?

I make my own prediction that this point in time, especially for mid-to-high-end cars, will appear in 2024. When the city NOA based on the big model, based on BEV technology, is truly covered by all cities, it becomes completely different.

Today we have done a relatively good job, our assisted driving usage rate in high-speed scenarios has exceeded 50%, but the overall assisted driving usage rate is less than 14%, because high-speed scenarios are limited, and it is not yet possible to use (assisted driving) in urban areas. I think a technology can no longer be separated if the daily mileage usage rate of users is stable at more than 60%.

What are the results of a city NOA? In fact, it does not need to become fully autonomous, but it means that when you are stuck in traffic every day, he will help you solve it, 60% or even more than 80% is handed over to it to drive, and safety is no problem.

This is a huge change, like if you buy a building with twenty or thirty floors, it is the difference between whether there is an elevator or not. Therefore, in the future in the high-end car market, if urban NOA cannot be provided, consumers are the difference between buying and not buying.

One phenomenon this year is that enterprises on all Orin computing platforms will basically deliver city NOAs in the fourth quarter. I myself think that by the end of this year, most of the leading companies will be able to reach the level of Tesla by the end of 2011, and by 2024 can reach the level of Tesla in North America at the beginning of 2023.

Therefore, at this point in time in 2024, it will first enter the intelligent era from the middle and high end, an intelligent era based on software 2.0. Otherwise, we can only sell electric cars forever, and it has nothing to do with intelligence. Smart is not to install an Android, this is not difficult, today many people do not understand these things. In the future, we will not provide products based on the original way, city by city calibration, we will provide products based on large models, covering all (regions).

Business cross: brand is the guide, product is the center, high gross profit guarantees high research and development

In the first two questions, I answered why we started this business. Next, I think the most important thing to be a 0 to 1 business is actually the brand. Because brands have to answer questions about both internal and external coins. We need to answer inwardly first who I am, where I am going, and how I am going. On the other hand, we must be clear to consumers: who we are and what kind of value we provide.

I think the most important thing about the brand concept, which Qin Zhi (former CEO of Autohome) brought us to do after joining Autohome in 2007, and he brought a huge help.

It determines the efficiency of your business, and it determines whether everyone believes in it. So what is the ideal mission? In fact, from the moment of registration, our company name is "Car and Home", we hope to change the two most important spaces through artificial intelligence and renewable energy technology, one is the car and the other is the home, so our mission is to "create a mobile home, create a happy home".

The second is the corporate vision, which is what kind of enterprise do we hope to grow into in the medium and long term? We hope to become the world's leading AI company by 2030. We have to have capabilities like Tesla and Apple to get a foothold.

The third is corporate values. It's the ability we have that truly belongs to us and continues to help us succeed. You have to go through a certain verification cycle before you can extract what you have, which is a better stage for the establishment of corporate values.

When Qin Zhi joined Autohome in 2007, Autohome had already achieved the first in the industry, the second in the forum, and the third in the information. We wrote on the blackboard why we could do this, and we wrote a blackboard. The first article identified at that time was to "put the interests of the consumer first"; The second is "do the right thing, not the easy thing"; The third rule is "get 60 points first, then do 100 points".

When we first started doing ideals, many people said why do family? Why make a six-seater? Behind this is actually the core value of the ideal car - to exceed the needs of users.

Because the automotive industry is so long, it is easy to become obsolete just to meet demand. What is Beyond User Needs? In 2016, if we do a user survey and ask if we want to extend the range, users will say that this is redundant. Ask if you want four screens, and the user will say that he needs more buttons. Q: Do you want six seats? There were no six-seater SUVs then, and he said why don't I buy five- and seven-seaters, and pay more for one less seat? Until the user drove the six-seat SUV, the elderly and children found that getting in and out of the car was so convenient. Therefore, our code of conduct is to always put user value first, not demand, but value.

The second is "do the right thing", why do many things have to think first, because this cycle is too long, so in the ideal, you have to have deep insight and think before acting.

The third is to "solve all problems collaboratively". We must not only cooperate well with ourselves, but also with all our partners, and the product must cooperate well with technology, manufacturing, and service, which is effective, which is our own core value.

The other aspect is that we are consumer-oriented, who are we and what kind of value do we provide?

Our core positioning is the preferred luxury electric vehicle brand for families. There are actually two key elements here: luxury, that is, like BBA, we only do more than 200,000 user groups; What does family mean? In fact, 79% of the more than 200,000 user groups are family users, just like Mercedes-Benz for luxury, BMW for driving, Volvo for safety, we are not the pioneers of the electric vehicle industry, so we must have a clear image of ourselves to do this.

Of course, we are also good at doing family, because at the beginning, these partners, each with an average of three children, so we know this 6-seater scene and multi-person travel scene very well.

The brand concept should be sorted out very clearly from the outside to the inside, and it should be talked about every day. It keeps all the resources invested, all the energy together, and it lets people know how to make trade-offs. Trade-offs are important in this industry.

After the brand concept is there, whether from 0 to 1 or from 1 to 10, we actually take this picture to diagnose, as an intelligent electric vehicle company, we must face every aspect of ability improvement, challenges and competition. Here is a model of the cross:

The first is the brand I just talked about. Of the more than 300 emerging automakers, only 1% of the companies are qualified in terms of brand, and most of them do not even have basic common sense. Today I told you about home, and the next day I told you that I accelerated in three seconds, all the logic is out of order, and it will affect his products and all aspects of his operation. Basically, all brands are folded on the brand first. You can see that the company that has problems, the internal management is a mess, because everyone has a ghost, which is the biggest problem.

The second point is the product. The brand is our head, and the product is the body. As an automotive product, I think it is roughly divided into three levels, the first level is that it must be safe. I can safely tell everyone that the L7, L8, L9 are the safest cars in the world. Because my family is in the car, and all my friends around me buy this car, how can I make it unsafe. And for home users, what they care about is not the safety of one person, but the safety of the family. For example, the A-pillar, using high-strength steel and using hot-stamped steel, in fact, the cost is hundreds of dollars, but many companies have to save these hundreds of dollars. My entire anti-collision structure, front and rear anti-collision structure, when others dismantled the car and evaluated, found that there were two anti-collision beams under the A-pillar wheel fending, and there were anti-collision beams at the tail. Even if these anti-collision beams use the best aluminum structure - aluminum is the most effective in anti-collision - the cost is less than 1,000 yuan, but some companies have to put foam plastic. Because in the past, cost competition was too fierce.

In addition, all of our front seat belts are explosive pretension, which can hold you firmly no matter what the situation. All our cars are full of airbags, including the head airbags and side airbags in the second row, and the overall cost is 2,000 yuan, but most companies are unwilling to spend this money. Once there is a problem with safety, it is a super big problem, safety is the bottom line, safety is never any budget, just do it directly, not the left side meets the crash test requirements, the right side does not do, the left side and the right side are the same, because the right side sits my wife, of course she is important.

What is the second layer of the product? is the value of the product. Space, comfort. In fact, it is divided into two parts, one is the software part, and the other is the hardware part.

The core of the software is that when I can do the central controller, intelligent driving controller, behind the entire software are all self-developed, in fact, the software is not a problem, the software is what needs the user can meet, and even we have now let users be their own product managers, that is, we provide "task masters". Apple's shortcut can only call the APP software function, and the car is hardware and software that can be called together. So at this time, how do you measure self-research, self-research depends on how many functions can be mobilized in the task master, because these function providers are not open to you, only your own self-research can mobilize all functions, and combine these functions together. Normal suppliers will never open white boxes to you.

But the hardest part is the hardware part, the hardware of the car is equivalent to 52 poker, you have to remove one function for every additional function, not that I want to add cost.

Many people say why the L7 is not made into a linkage seat, like the Mercedes-Benz S and BMW 7 Series, that is, when I adjust the back of the seat, the cushion automatically extends forward. Because it is another skeleton method, when you want to ensure the safety of the structure, and then use the linkage method, whether it is the BMW 7 Series, Audi A8 or Mercedes-Benz S, when I adjust the seat back, the cushion is adjusted synchronously, and when I put away the seat cushion, the seat back is also adjusted at the same time, which causes the second row of seats in these cars to not be put down. This makes it very difficult for you as a family car to transport something. This is a trade-off: Are you a service boss or a home user? This trade-off is easy to make. If you want to accelerate a car for more than 3 seconds, it may sacrifice some other parts, such as the energy consumption of medium and low speeds, and the motor size will have a corresponding impact.

At this time, everyone found that the core of the hardware is actually a trade-off. What are the trade-offs? Around your brand, around your users, not for the existence of the R&D team, it must be a trade-off for the value of your users and your user base.

At the same time, we must establish a complete verification system. In the ideal car, I said that it is impossible to change a thing, it is completely different from what you think, to go through a very rigorous process, to have a demand analysis, to what extent the demand analysis will be written, how many cycles will be written for the development of this thing, how all competitors do it, what value will be generated after the actual line, how to calculate this value. A piece of software and a piece of hardware must be fully analyzed. Therefore, many traditional car manufacturers will understand after coming to us to make products that the products of traditional car manufacturers really cannot surpass us, because our workload and system are much more complete than traditional car factories. From requirements analysis to requirements verification, to parts SOR (Specification Of Requirements), to the subsequent parts verification, vehicle verification, our whole process is very complete, no one can bypass it. If you do not do a complete requirements analysis, this thing is impossible to establish a project, and research and development will never be done. In the process of doing (product analysis), research and development, including supply chain, products must be intervened in advance. If we want to make a new feature, we have to go to the supplier, go to the R&D, and get all the key details.

This is the key to how to make a good product: you must know how to make trade-offs and establish a complete verification system.

The third layer of the product is your brand's "special sense of yearning".

We want to "be a mobile home", and we can't be a dragonfly on a "mobile home". If the car is to be like a home, it must have the largest and most comfortable space, and more screens. Everyone holds a screen at home, why not in the car? On the other hand, if you want to do autonomous driving, you have to go on LiDAR, double Orin, and make NOA standard in all cars. This is the core, don't dragonfly a little water, must do overinvestment, overallocation.

These are the three dimensions of the entire product: safety, value and brand aspirations. Trust me, consumers can fully feel these things, not fake at all. It's just that you can't look at non-user groups, and your real consumers can understand and identify with these things very well. That's how we deliver a good enough product to our users.

After the product, the third is the gross margin.

Is gross margin so important? Or is it related to the automotive industry. If you want to become a car company that stays on the table, you basically have to have hundreds of billions in revenue. If you want to become a world-leading enterprise, you must have trillions in revenue. How to operate such a scale of revenue becomes critical. You can't rely on financing forever, you can still rely on financing when you have billions of income, tens of billions of income, how can you rely on financing when you are big? We can also clearly see from the history of the development of the automobile industry that when there is an economic crisis and cars are the most difficult to sell, it is often when they cannot be financed at all. So even like GM and Chrysler, after the 2008 economic crisis and car sales fell by 40%, they all went bankrupt. In addition to the decline in income, bankruptcy is that it uses too much leverage, so Delphi was later taken out, because it turned out that he bore all the leverage himself, and the 40% decline was almost all borne by himself.

So what is gross margin? Gross profit is all sales price minus cost of sales, cost of sales includes the cost of consumables for the car, manufacturing costs and corresponding allocations, as well as the transportation costs of the car, software costs, and taxes. The personnel expenses in the store are called sales expenses, which are in addition to the gross profit margin, and the part of our gross profit margin is to cope with the sales expenses. So the gross profit margin is equal to the sales revenue - the cost of sales, and the rest is the money that the company can invest, which is the part that improves its own capabilities.

Going back to the perspective of the car, as a smart electric vehicle company, we believe that the healthy gross margin threshold is 20 points. R&D expenses are basically more than 10 points, and sales management expenses also need 7-8 points if they are done very well. At the same time, you also have to bear certain risks, as well as capital investment, build factories. So we think 20 pips is healthy.

At present, Tesla is more than 20 points, we are stable more than 20 points, in fact, BYD is also more than 20 points, because BYD is not a direct way, if it includes dealer fees, the gross profit margin of the car is also more than 20 points, which is relatively healthy. Otherwise, how do you invest in R&D? You can't say that the capital market is good to invest in research and development, the capital market is not good to shrink a little, cars are not like this, cars you make any plan should be a cycle of more than 5 years.

Of course, the higher the gross margin, the more difficult it is to operate the business, the lower the gross profit margin, the easier, and the negative gross profit is equivalent to sending money, which is the easiest.

We judge that 20 points is the minimum gross margin that a business can develop in the long run. So you can see that Tesla has been losing money for so many years, but from Model S to Model 3, its gross margin has been stable at more than 20%. I can spend a lot of money, but I always insist that the gross profit margin is stable at more than 20%, only one year, Tesla's gross profit margin dropped to 18%, it should be when the Model 3 is difficult to produce in 2018, and all the remaining models are above 20 points. Tesla has cut prices on a large scale this year, but at the annual report meeting, he said that it is still necessary to maintain a gross margin of 20%. In this way, the enterprise can develop healthily, and when making any long-term investment, it can not be instigated, and can it firmly invest.

After having gross profit, how is the money spent? I think the first important thing is to invest in R&D.

We divide the research into three layers. The most basic research and development is product research and development, just like all traditional automobile factories, I use suppliers to make products.

The second layer is platform research and development. After the new L series, we made one of the most important decisions in 2019 to go all out on platforming. The traditional platformization just means how to share my chassis, and our platformization is more extreme, basically according to the iPhone way: we divide the entire platform into 4, including an extended-range electric platform, a high-voltage pure electric platform, an intelligent driving platform, an intelligent cockpit platform, and an auxiliary small platform, such as an electronic and electrical architecture platform. They're completely versatile, which is why we OTA is efficient, because we only need to match one car, we don't need to match four or five cars. This avoids the fact that in the early days of smartphones, if you give Nokia Apple's iOS operating system, it won't work, because just matching thousands of different screens and different processors is basically crazy.

However, the investment in platform is particularly high. Who was doing platformization in the past? Not a traditional car manufacturer, but a super-large supplier. Almost every car manufacturer in Europe buys a supplier's PHEV system, as does autonomous driving. And the extended-range electric system we want, no supplier can make it, we can only do it ourselves. Including range extender, all controls behind it, and intelligent driving.

What is the result of handing over to the supplier is obvious to all, not to mention that it is impossible to do large model training later. The supplier's supporting L2 assisted driving may cost $5-1 billion to invest, and if you want to make a larger model, the investment needs $1-2 billion. Generally, he will sell you 1~200 million and continue to provide you with parts, which is its business model, and the advantage is that 20-30 companies can be found to share the cost. But if we do it ourselves, it starts at 100 million US dollars, and if we want to do a complete urban NOA in automatic driving, the investment will cost about 2 billion US dollars, including the training later. So in the early days, whether you can invest in such a platform actually becomes very critical in your overall gross margin and financial structure. Are you determined to keep investing?

But after the scale comes up, the efficiency will be different, your update speed will be faster, the development efficiency will be higher, the supply chain management will become easier, the user experience will become better, the quality will become better, and the cost will be better reduced. To be honest, L7, L8, L9, except for the shell and seats, are the same, just like if you buy iPhone 14, iPhone 14 Plus, iPhone 14 Pro and iPhone 14 Max, the things behind are the same. Platform research and development basically accounts for our cost of 7~8 points, which is much higher than the research and development expenses of traditional automobile manufacturers.

The third layer is system research and development. Further down how the cost of the platform falls, explore from a deeper level, including whether we use someone else's and usable system, or develop the operating system ourselves, and the reasoning chip behind it. If I make my own inference chip, can I do the same cost as Tesla, because the algorithm is in my own hands, including the entire training platform and training chip behind it.

In the long run, only after achieving these three levels of research and development, you can become an enterprise at the level of Apple, Huawei, and Tesla, and the competition will have an advantage. For each higher dimension, there is a strategic level of one dimension higher. We often say that if Samsung does not have a huge problem in China, if Huawei is not restricted by the US government, I think Apple, Samsung, and Huawei that have these three layers of capabilities at the same time can have a combined share of more than 90% in each market, which is the possibility of research and development.

With R&D, it's important to have our ability to deliver.

The whole chain of the car is too long, long chain, long cycle, manufacturing cost is extremely high. So if a product is sent wrong, the company's valuation will drop by 80%, and everyone is playing the game under the rules. To deliver products, it is not just to build and develop them, but also to improve a lot of capabilities.

The first is commercial capability. This includes all sales and service links. Many people say that marketing is not important, of course marketing is important. You thousands of people have spent billions of dollars on research and development in the past 3 years, making a product day and night. Marketing determines whether you pass 100% of your value to consumers or 30% of your value to consumers. So this money is not money for marketing, but money for R&D as a whole. Most companies (delivering value) give a 2-3 discount, casually talk about it, the most basic products are not understood. This is a very serious problem.

This also includes the entire user service link. When we sold one product, we used central management, and now when we sell multiple products, we start to have "governors". We will go further than traditional automakers, not to the region, but to the provincial level, where the "governor" can allocate all the key resources. How to effectively manage it with direct management, the challenge is different, because direct management will face three major challenges: first, cash flow, which used to press on dealers; The second is the speed of opening stores; The third is how to avoid the big pot of rice and build the incentive model.

In February, all of our staff supporting the storefront have achieved a per capita sales capacity of close to 7 cars, and if the management is removed, the per capita is actually more than 10 vehicles, which is basically more than 3 times that of the new forces, so that we have the opportunity to let our employees do the work of three people and earn money for two people, so we think it is more pullable, so our people are now the most difficult to dig.

Including the charging network just mentioned, it must be built. To what extent? A high-speed line must be built all the outlets in a province. We look at the necessity of strategy in the whole structure, not where to go.

Commercial capacity is followed by supply capacity.

Why we climb relatively fast, the most critical thing to limit the climbing speed is the three electricity. We divide all the parts of the car into four parts, the first is the traditional rearview mirror, bumper, the automotive industry originally has a production capacity of more than 20 million passenger cars a year, this part of the supplier to electric vehicles, to fuel vehicles is no difference, do not need us to do. The second is new electronics, chips, such as domain controllers, China used to have a very good foundry system, and China purchases about 40% of the world's chips, this system manufacturing efficiency is very high, to a car to produce 200,000, 300,000 sets of domain controllers without any difficulty, Foxconn can do it for you, the solution is not complicated. Everyone talks about the lack of chips, but there has never been a lack of chips and computing platforms from Qualcomm, NVIDIA and Horizon, and the lack of traditional small function chips is because China is still very mature in this regard, and China is the world's strongest electronics foundry.

The third type of parts is more troublesome, that is, related to the three electricity. For example, our L series a car is to use three motors, two drives, a generator, assuming that 300,000 units are sold a year, I order 900,000 to any motor factory, it is impossible to take it, which means that he has to build 10 factories, and everything is borne by him. So there is a very interesting phenomenon, that is, when these car manufacturers face these new forces, the order amount you report to him is different from the unit amount he calculates, and they have a system to analyze how much you can actually sell, and I will produce so much for you. At the end of last year, all the traditional suppliers thought we were doing about 150,000 vehicles this year, and when we proved that volume, we were transferred to 250,000 vehicles.

Therefore, like the three electric parts, we need to do it ourselves, otherwise the supplier will not be able to supply it at all. We built a range extender factory in Mianyang, and the first five-in-one motors of the L series are produced by ourselves next to Changzhou. In addition to the factory in Changzhou, we will also establish a factory in Beijing to produce pure electric vehicle products. These things are our supply capacity.

Supply capacity needs to have 24 months of forward-looking planning, otherwise it is too late, it is necessary to take land, build factories, recruit people, and trial production in advance. Including cooperation with suppliers is also like this, if you want a European manufacturer to only make us 3-40,000 sets of air suspension a year, talk to him for a year, may be able to supply up to 60,000 sets, but we need hundreds of thousands of sets a year, then we will discuss with Chinese suppliers how to build production lines, factories, to ensure that secondary and tertiary parts are in place. So the supply chain has to be in place, and we have not done well in the past year, but some advance layout has brought us a lot of help.

Another problem with the supply chain is that the self-control rate cannot be too high. All companies in the world have experienced, such as Toyota, like GM, and later Toyota took Aisin and Toyota Boshoku out for others to supply, and Delphi also took it out from GM. Many parts manufacturers are spun out of car companies. Because if you do all self-made, it means that when sales drop by 40%, you will bear all the leverage and all the costs. At this time, I am afraid of encountering an economic crisis. So we think 30% self-control is healthy. Many problems can be cooperated with suppliers, even if we develop ourselves, we are handed over to some suppliers to produce, rather than our own production.

Organization: round after round, overcoming laziness, inertia and ignorance

The third biggest challenge is organizational capacity.

Because cars are too complex, especially smart electric vehicles, we are an automobile company, an Internet company, a software company, we are also an artificial intelligence company, a commercial retail enterprise, a manufacturing enterprise. Organization becomes one of the biggest challenges. The organization determines our efficiency and effective positive feedback.

Organizations encounter three different types of problems. The problem at the bottom may be laziness, which is a personal problem. One level further up is the inertia of management. For example, I used to make a car that sold for 100,000 yuan, and the people I hired at the beginning were all 100,000 yuan cars, and then I wanted to make a car that sold for 300,000 yuan. What to do? We bought a bunch of really nice cars and let the employees drive them every day. You must not be like the original, take a ruler to measure the size of the suspension and go, you must drive every day, drive for 3 months, and then talk about how to adjust your suspension, you have to go to him to break their inertia. Further up, to the management level, the biggest challenge comes from ignorance, that is, not knowing that you don't know, because this complexity is higher.

How to solve these problems, we will face three levels of business. First, professional, that is, I can do things well and manage myself well; The second is management, that is, I can not only do good deeds, but also lead a team, whether it is a project or a department; Further up is management – how to use good money? In the management you see the cost, at the management level you see the resources, how to use the resources effectively.

The operation of the enterprise from 0~0.1 verification, from 0~1 to 1~10, keep rising, the problems faced are actually different. How do we turn laziness into initiative, how do we turn inertia into necessity, how do we turn ignorance into cognition.

Two ways, the first is to look at the process.

Processes are the products that this business operates inward. All processes around the world are designed to solve strategic problems. A good process isn't about asking you to 123456789 and telling you how to do it, it's about saying you don't skip it.

For example, we go to solve the professional process of specific business, than you have to do a user demand analysis, at this time you first have to look outward, how is this function at present, you have to analyze, find the data, what the user actually looks like, and then do planning: what the product looks like, what the function looks like, how long does research and development take, how much cost. Then you have to go to the operation, plan what value the product will generate, what kind of activity it has, how much it will be used. And the review process, how to turn other people's things into your own ability. The core of the process is that you don't skip these important things, as long as you follow this thing, you will become very professional. It always gives you a positive cycle, and the product you make will get better and better. Users are becoming more and more satisfied, and the market position is naturally proactive.

What will be the problem in the end of the enterprise, that is, the processes and tools I provide are low-dimensional, I only provide you with a professional tool, but I want to find you to manage; I only provided you with management tools, I asked you to operate, this is the business is irresponsible. Because there are no tools for corresponding dimensions, why upgrade employees.

Processes also contain management tools. For example, after I have a business plan, how do I analyze what the team looks like? You need to first look at what organizational structure your peers used to do this business, and even what salary was given to each position and who was recruited. Second, you have a plan, according to the business process to design what positions to meet my business plan, what kind of people need to be recruited for each position. That's when HR's job is very easy. Because it is clear why this person appears here, what other companies look like, how much this person should pay, and what kind of specifications are recruited. It is to let everyone not skip these things, so that training, every manager has a very good, or three-dimensional ability.

One level further up to the management floor. For example, we go to learn the first-level processes of many large enterprises, such as IPD is integrated product development management, IPMS is integrated sales service management, and ISC is integrated supply chain management. They help you see, what are the results of your business? How effectively you allocate your money and people's resources can be measured.

Measurable purposes are twofold, the first is to make sure that your resources are really in place in the process of doing it, rather than applying temporarily, we are sure that to do this, you have to really give the resources, you must use them. The second is to be able to effectively review, and I initially planned to use these resources to achieve this goal, and what was the actual difference in the end. If you do well, you will strengthen, which is the same as autonomous driving. People who don't do well will definitely have to solve this problem first next time.

So we found that all the most advanced processes in the world, of course, this process includes the process tools, IT systems, and methodologies that we are talking about, and it is exactly the same. It's all front-end land to help me solve every level of the problem, solve the problem of laziness, solve the problem of inertia.

Whether it is just mentioned, the tactical layer, we use our own product demand analysis, including like we use Toyota's TBP (Toyota Business Practise), that is, Toyota Work Method problem demand analysis, and then the next layer is LBP (Ideal Work Method), and then talent structure planning, up to the first level of processes like IPD, IPMS, it all principles are the same, but it can solve problems at different levels and different professional fields.

So at this time, I will tell the team: the difference between living things and living things is the difference in information acquisition and processing ability, and the biggest difference between organizations and organizations is the difference in the use of tools. You have 3D tools and you can easily crush businesses that are just using 2D tools.

Everyone knows that process change and organizational change are not easy, but only after you sort out this thing, will you find its necessity, only by doing this, can we solve the problem, can we improve the efficiency of the supply chain, can we motivate the sales team, and can turn each store manager into a management talent, not a professional talent.

At this time, you are facing different businesses and different levels, and you know how to adapt. For example, we originally from the 0~1 stage, we are actually the most important thing is management, so when we look at enterprises with hundreds of billions of revenues larger than me, we are looking at the three-dimensional world from a two-dimensional enterprise and seeing that they are complex and rigid, but when we upgrade ourselves to the three-dimensional world, we find that the original people are both simple and rich.

So the first step in doing everything is to look outward at the industry, what your opponents are doing, and know yourself to improve your own cognition. The second stage is to set goals, do planning, not only business planning, but also look at shortcomings and gaps, and then effectively allocate people and money. The third step is what is done execution and collaboration, these resources to really use, and then who does it and how it can be done. When you do the first two steps very well, coordination is inevitable, no need to hold a whip to urge, because no one does not want to do an efficient thing. The goals, resources are clear. Then the fourth link, all these good processes and tools are the same, will definitely do a review. Just precipitate the ability to do well to your team, to your process, to the company's product, to ensure that the next thing is based on this, not random.

One level after the process is to the person.

The process is equivalent to a company's product, which is constantly upgraded to solve different problems as the company grows. There is also the problem of people, and to solve people is to solve the problem of the mind.

Starting from Autohome, Qin Zhi introduced us to "7 Habits of Highly Effective People". Ideally, we wanted to go deeper, so we partnered with Covey (author of the book) of "7 Habits" to do a big course for ourselves.

Many people have read this book for a long time, it is actually corresponding to the 3 levels of "professional", "management" and "management" to train ourselves, it is a very effective methodological system. The first three of these habits are telling you how to become an adult. It's how we make sure we're in a positive state, know what we want, how to build goals, and manage our time effectively. We first manage ourselves and manage our relationships with ourselves to become adults. Don't look outside to find the reason when something goes wrong, look inward first. In this way, all people are willing to work with you.

The second link is how we become a competent manager: it is our obligation to understand others. Whether you bring a project or a team, you fully understand that everyone on the team is obligatory, and you can't say why he is different from me? Everyone is valuable because they are different, and if it is the same, you can do it yourself.

Going up the next level, helping us solve our goals at the operational level, it is more complex. I have talked about every training, every wave of staff training, once a week I will tell you. Then every new P8 and above P8, I will talk to him, one-on-one. Because the world is too complicated, everyone must know why behind it.

Back to the third link, how to become a leader is how to get more people to understand you. Don't complain about why others don't understand me.

During the Liberation War, the Kuomintang talked about the Three People's Principles. Nobody knows. What did Chairman Mao say? Fight the local tyrants and divide the fields. All the peasants knew what they were doing, so more people joined the Communist Party and won the liberation victory. For example, Deng Xiaoping also said a lot of very popular words, called "the computer starts with the doll." You make it clear that everyone benefits.

Why Amazon Bezos and Buffett wrote a letter to shareholders. In fact, Amazon's letter to shareholders is about two things, one is why Amazon is the company with the best user experience in the world. Second, e-commerce companies should have a value called operating cash flow, not measured by its earnings, so Amazon has obtained hundreds of times the price-earnings ratio, if Amazon is measured by traditional PE, then its resources are completely different. Why Buffett keeps writing letters to shareholders is also two things, one is value investing, and the other is how he chooses a good company. Speaking for decades.

Including last night, why did Musk send the third phase of the grand plan, he hopes that more people understand Tesla and participate in Tesla. When Jobs opened the iPhone conference, he said: Why do we build an iPhone, it says that the iPhone is an MP3, a player, a browser, a computer, everyone can understand it when they hear it, even children can understand it, I think this is a skill.

And what China's auto companies, including many mobile phone companies, will say. I have the highest score. Therefore, many companies have not learned the essence of Tesla and Apple, such as its gross profit margin and channels. It often appears, there is no life of apple, and the disease of apple.

So when we are adults and have professional processes and tools, we are sure to solve the problem of laziness. When we become a qualified manager who can understand the differences of others, we will definitely be able to manage well and solve the problem of inertia. When we become a leader, let more people support us, understand us, and create success together, we can use various business tools to solve all kinds of ignorant problems. Then continuously improve the team's awareness and become the team's capabilities. Say this is the philosophy of our core organization.

Or take a picture (business cross chart) to end today's sharing with you. With such a framework, we can not only diagnose ourselves, but also see the strengths of our opponents, the mistakes we have made, the advantages we learn, and the mistakes we avoid. See the gap between us and advanced enterprises in terms of brand, product, gross margin, manufacturing, platform research and development, system, entire delivery and commercial capabilities, our supply chain capabilities, and Tesla, and BYD.

One of my big feelings in the past three startups is that when we challenge hundreds of billions of dollars in revenue today, we will go through 5 stages.

The first stage was that I always felt that I was different from others, never looked out, and hid in a well. The second stage is to look outward, but do not understand. These corporate neuroses, engaged in such a complex process, a group of people here to engage in operations, analysis. The third stage is to feel that it is still not possible, pain, many problems can not be solved, find some high-ranking people, consultants, let him show you to understand.

Then in the fourth stage, it will enter a I understand and do it after I see it, which is a problem, which is not something you can do by recruiting a few employees from a good company. Because people who can plan, build and operate roads are not the same as those who only drive on the road. So often we find a group of people who drive to build roads for us, or we don't plan anything ourselves, and we come up and build roads.

By the fifth stage, we will have a clear idea of which consulting firms and even which teams the company has used at some point in the past. And who the company used at this stage to plan, build and operate roads. We recruit him and slowly build the road ourselves, and then the team will drive and transport the value on it, and the efficiency will change.

So it's very painful, each round is to do a stupid thing, more than once, and then as long as you are alive, do another upgrade, and so on all the way. This is a sharing with you today. Thank you all!