Is Xpeng ready to accelerate into the global market?

On the evening of October 24th, at the fifth annual 1024 Xpeng Tech Day, despite the absence of his old partner, Wu Xinzhou, He Xiaopeng didn’t notice a difference – he looked spirited and even more confident.

Indeed, just by observing the accelerated pace of automation technology development and product implementation, He Xiaopeng has a justified reason to uphold his confidence.

After all, judging from the current progress at this 1024 Tech Day, with long-term investments and strategy in autonomous technology, Xpeng’s stride towards realization has been quicker. It has even targeted overseas expansion.

For instance, in regard to the city NGP, Xpeng has finalized the goal to reach 50 cities by the end of the year, and full coverage of major national cities’ roads for the next year; For AI driver assistance, it aims to achieve nationwide use regardless of the city, with availability after a one-time learning process; Furthermore, the Pro version and Max version vehicles will directly integrate the XNGP architecture… Moreover, Xpeng plans to cover Europe with NGP by next year.

These achievements provide enough support for Xpeng’s leading position in China’s autonomous driving sector and boost He Xiaopeng’s confidence.

In He Xiaopeng’s perspective, he’s pinpointed the ‘inflection point’ for autonomous driving at 2025. This indicates that by that time, all Xpeng models will include autonomous city driving as standard and will cover a scope of 150K – concurrently, there will be a significant cost decrease without compromising on this standard configuration.

No doubt, this is an ambitious goal but He Xiaopeng asserts that he has full confidence in achieving it.

In fact, He Xiaopeng’s confidence in Xpeng’s autonomous driving technology stems largely from his personal involvement in the business. For instance, he has done a substantial amount of planning for autonomous driving operations, and worked tirelessly on lobbying for autonomous driving related business. He even modified the XBrain architecture diagram on the day of the 1024 Tech Day – he believes that this deep involvement in frontline operations is something many other auto entrepreneurs have not achieved.

Thus, even when facing competition from other automobile companies on the autonomous driving track, He remains complacent. He understands that achieving success in autonomous driving isn’t an overnight process and isn’t solely reliant on financial investment. In addition, he holds the belief that to excel in autonomous driving, apart from having the right people and enough time, one also needs a bit of luck.

Of course, He Xiaopeng’s confidence at this stage doesn’t just stem from autonomous driving technology, but also from his new understanding of Xpeng’s overall business.

In reality, ever since last year’s crisis triggered by Xpeng’s G9, He Xiaopeng began forcing himself to undertake various operational tasks, immersing himself in frontline work, and became more decisive in his management style. For example, after taking control of the automotive technology team at the end of last year, he decisively dissolved the entire battery cell team once he discovered issues; moreover, this year he’s dismissed 12 executives.In an effort to supplement his understanding, He Xiaopeng, the founder and CEO of Xpeng, has been constantly learning from the industry, such as engaging in intensive dialogues with suppliers in terms of power and supply chain initiatives. For instance, He Xiaopeng even arranged a meeting with suppliers on the afternoon following the 1024 Tech Day to discuss issues related to establishing a factory for peripheral components of battery cells, as well as logistics and supply.

Furthermore, from a foundational perspective, He Xiaopeng is becoming increasingly aware of the importance of product positioning and organizational process systems. For instance, in the field of autonomous driving, he is most grateful to Wu Xinzhou for helping him build a solid team. In terms of vehicle styling, he believes that the underlying issue lies in the optimization of the process. He emphasizes that “system capabilities always exceed individual capabilities.”

Indeed, from these changes, He Xiaopeng increasingly resembles a qualified CEO of an automotive company.

In fact, the personal changes He Xiaopeng has undergone, coupled with a series of organizational adjustments he has made at Xpeng, are starting to see returns at the market level. For instance, from January to September this year, Xpeng’s monthly sales have been climbing steadily, exceeding 15,000 units in September. Additionally, only 15 days after its launch, the Xpeng G9 2024 model garnered over 15,000 major orders, of which the Max model accounted for as much as 80%…

Yet, He Xiaopeng remains indifferent to these achievements.

In his view, after a year of adjustments, Xpeng has merely entered into a positive “micro-cycle.” This round of organizational restructuring will take about five years to implement in total. He believes that the current accomplishments of Xpeng can only be described as “making a big fuss out of nothing,” which hold little significance. He Xiaopeng believes that “all my logic is to accomplish big things.”

Therefore, what he is looking forward to more is: by the fourth quarter of next year, Xpeng will be able to enter into a high-speed positive cycle.

On the afternoon of October 25, He Xiaopeng received an exclusive interview at Xpeng’s headquarters in Guangzhou from Garage 42, discussing his profound insights into the autonomous driving industry and the development of Xpeng’s autonomous driving business. He conducted a comprehensive review of Xpeng’s evolution over the past year from aspects such as technological insights, organizational adjustments, and business boundaries.

The following is the main content of the conversation between Garage 42 and He Xiaopeng, edited without changing the original meaning.

Marketing in autonomous driving must maintain integrity

Garage 42: At this year’s 1024 Tech Day, you announced that the NGP of 50 cities will be opened by the end of the year. How do you determine whether a city meets the conditions for opening the NGP? Is it you who makes the final determination?

He Xiaopeng: I don’t make the decision. It should be made by the person in charge of the autonomous driving test and the project engineer in charge of city opening. They will send a team to run all levels of roads in the city, and then run on a lower level of roads. Since we don’t know the road conditions of the city, we must run it once to determine.

Garage 42: A competitor has claimed that their self-driving technology will be available nationwide by the end of the year. What’s your view on that?He Xiaopeng: Frankly, if you only aim to get the car on the road, making sounds and moving, it’s not hard. ACC and LCC can do it. However, driving well is a different story; it’s challenging.

Automotive Garage №42: When discussing intelligent driving competition, you said some are responsible for promoting, while others earnestly work, a perfect match. Xpeng emphasizes on urban NGP, but according to our survey of AITO’s new M7 users, many are willing to pay for AEB.

He Xiaopeng: Firstly, most people may have never encountered AEB. Secondly, I believe 99% of the competition’s AEB claims are false. Their promotions are not official releases from the companies; they’re all from short videos. Our team has enquired it; their AEB cannot be used on the road, there are too many instances of inadvertent braking.

Automotive Garage №42: So, you think their promotional effects are greater than their practical usage?

He Xiaopeng: Current industry discussions about AEB mostly cover longitudinal AEB. When it triggers, the speed in most cases should be within 60 kilometers per hour. Too high speeds trigger inadvertent braking which is a huge shock and unacceptable for users.

Automotive Garage №42: How will you handle AEB?

He Xiaopeng: Our current XNGP technology will later incorporate AEB as one of its aspects. For instance, we always talk about “Static AEB”. Once there are obstacles nearby, the vehicle will actively avoid them; it can’t crash even if one tries.

I believe treating customers as guinea pigs is wrong. Some companies dare to do this, but there will be backlashes. After several years in the auto industry, I’m unwilling to proceed this way. However, we’ll surely do a better job with AEB.

Automotive Garage №42: What do you mean by ‘better’?

He Xiaopeng: I initially planned to talk about our work on AEB at Tech Day 1024, but decided against it. We can talk about it later.

Automotive Garage №42: If you wanted to put significant efforts into marketing for intelligent driving to increase user awareness, how would you do it?

He Xiaopeng: That depends on our marketing team. If they can describe an ordinary thing impressively, or can oversell an extraordinary thing, it’s their capability.

But if they overstate, I surely won’t approve. I have my standards on many things. Though not very high, they definitely surpass those of our competition.

Automotive Garage №42: In this event, you mentioned 2025 would be a turning point for intelligent driving, with urban intelligent driving becoming a standard feature. What level of vehicle will this standard feature be available?

He Xiaopeng: For companies like us, it will be a 100% standard feature. We aim to achieve this at 150,000 levels, if it can be done at this level, it will be even easier at higher price points.

Automotive Garage №42: You previously announced cooperation with DiDi, talked about ‘accelerating the application and popularization of assisted driving at the 150,000 level’. So, will this hot model at the 150,000 level also support urban assisted driving?Xpeng He: We progress one step at a time, and we are sure to support urban assisted driving in the future, which is expected by 2025. My hope is that the sooner, the better.

42 Garage: At that time, to what extent do you think Smart Drive will reduce costs?

Xpeng He: The reduction will be substantial, and we must ensure that cars at the 150,000 level are equipped as standard and profitable. It’s quite challenging, but I believe there is 100% feasibility, and it doesn’t sacrifice the other configurations that this level should have.

Intelligent driving is not a quick success

42 Garage: At the previous four Xpeng Auto Technology Days, you were paired with Wu Xinzhou. He’s not here this time. Is there anything different for you in a Technology Day without his participation?

Xpeng He: There’s no change.

I think Xinzhou is great, but the great thing about Xinzhou is that he has built a good team. This is what’s reassuring and thankful to Xinzhou. The power of the system is always superior to the individual. Previously when Xinzhou was here, I personally made most of the planning for intelligent driving business in terms of direction and inputs.

Now I have a call with Xinzhou every one or two weeks, mainly to discuss some cooperative businesses, such as the next generation Smart Drive computing power.

42 Garage: How do you plan the intelligent driving business?

Xpeng He: I will start from the perspective of technology or the product, look at what capabilities it should achieve in the next three to five years, and counter push what resources and technological foundation we need; Then, to see which of these resources and the technological foundation is long-term and which is short-term…based on this system, I will then discuss it with the team.

In fact, in the intelligent driving technology system of Xpeng, some things are done exceptionally well, but some are not so good, and the money spent is quite considerable. But once successful, these investments look trivial in hindsight. If not successful, even being frugal in spending won’t help.

42 Garage: From an industry perspective, the actual progress and external promotion of intelligent driving by many manufacturers show inconsistency, but you guys are relatively consistent. How did you do it?

Xpeng He: First, it’s a matter of decision-makers. Many technological plans of competitors in the past were not made by the boss, possibly by vice-presidents; but the boss would change the plan due to market competition and other reasons after the decision, hence the delays. But for Xpeng Motors, I personally did the business technical planning for intelligent driving, so I kept the original pace.

Secondly, I personally went through the many details of Xpeng Motors Smart Drive business, so I understand technology and engineering, etc. For example, last year, I ran to every ministry for a Smart Drive landing project and found out that I was the only one running. When I couldn’t go on, I talked to my competitors and found out that they didn’t even realize it; they didn’t do research, so they certainly didn’t know what problems would arise later. So, me entering the field and participating in some key nodes really helped.

Thirdly, I want to thank our team. As it’s said, everyone will strive to do it afterwards. Hence, it’s a matter of leadership, and also a matter of the team.Garage 42: Many competitors are now focusing more on autonomous driving, and everyone is eyeing you, trying to catch up, or even overtake.

He Xpeng: Everyone is starting to pay attention to autonomous driving, which is a very good thing. But I want to emphasize a particular point: autonomous driving, must not be rushed.

There are companies that claim to quickly possess certain capabilities within three or five months. There are two possibilities here. One is that it has accumulated a lot of relevant capabilities in the past; the other possibility is that it has used the abilities of others but claimed them as its own. This so-called ability, frankly, is all about fast food and shortcuts.

If it is an internet company, they might benefit from a first-mover advantage. But the manufacturing industry is not like this. They may be able to afford the initial development stage, but they must change their approach later, because it is very difficult to continuously function in this manner — hence, in the field of autonomous driving, catching up requires long-term accumulation, hard work, and effort.

Garage 42: But some competitors think they have a lot of money and can invest heavily in it.

He Xpeng: We also invest billions every year.

Garage 42: If a competitor wants to catch up within a year, how much money should they invest?

He Xpeng: I dare not say. In autonomous driving, the first is to find the right people, and the second is to accumulate over a long enough period of time; these two cannot be bypassed. Then add a little bit of luck.

Money is only necessary to ensure the right people and the right amount of time; it is a necessary condition, but it is not a sufficient condition for victory. It also needs a combination of process, system, structure, and luck.

Garage 42: What do you mean by ‘luck’ here?

He Xpeng: ‘Luck’ is a bit of metaphysics.

Let me give an example, though it might be inappropriate. Suppose a company hires 5 experts; one of them has to distract his attention due to a family member’s illness. Although he has 100 points of capability, he can only exert 60 points and fails to communicate with the team, delay the progress… It might just be a small thing like this that causes the company to miss a huge opportunity. So, luck does play a part.

Garage 42: By 2025, how many players do you think will be involved in autonomous driving?

He Xpeng: I don’t know, it depends on everyone’s luck. Some have a higher probability, some have moderate, and some low – but it’s definitely not about money.

Garage 42: Then what is it?

He Xpeng: We’ve proven many times that for a large company to do well, it may be harder than for a small or medium-sized company. Why? Because in the early stages of a enterprise, they tend to repeatedly disrupt themselves with high efficiency and low cost. The larger the company, the less willing they are to disrupt themselves. Moreover, the cost difference is too great, it’s on a different order of magnitude.

Business boundaries are about becoming number one

Garage 42: The 1024 Tech Day focused on autonomous driving, but comparatively, the cockpit part was barely addressed and many features were not elaborated. Why?He Xiaopeng: In fact, the most crucial function of the cockpit part yesterday was the SR intelligent driving environment simulation display. This feature essentially serves in facilitating human-driver co-piloting. Because humans will continue to drive for many years, and the intelligent driving system must also participate in driving, it needs to reassure and relax human drivers. Therefore, this co-driving function is very important and will have long-term value for many years.

Garage 42: In your opinion, how valuable is AI- ‘P’, backed by the big model?

He Xiaopeng: AI- ‘P’ will undoubtedly be smarter and more delightful to people. It can execute ongoing dialogues, drawings, and entertaining poetry writing, but these capabilities will not become the car’s core competitiveness, it may even increase the cost of using the car.

Let me give you an example, calling AI-‘P’ to pen a poem might cost a dime, which is expensive; the old ‘P’ might only need a milli or 0.1 milli. Therefore, we should be interested in the actual customer value, not gimmicks.

Garage 42: You think that features like letting ‘P’ write poetry are more like gimmicks.

He Xiaopeng: I think it’s more of a minor itch, not a major pain. In a car, the experiences between those who drive and those who don’t are vastly different. To the drivers, what use does a poem recited by ‘P’ bring? Therefore, we must focus on perfecting the most crucial aspects.

Take, for example, the 3D human-computer interaction function we announced last year, which we have accomplished, but it doesn’t have a noticeable effect. I believe it’s a feature that seems valuable but does not add value in actual use. So I’ve reduced its importance.

Garage 42: Externally, Xpeng’s understanding and layout of many aspects of the cockpit are very advanced, including full-scenario voice; and the development of the cockpit seems to have reached a node. Do you think the cockpit will have some new imaginative space?

He Xiaopeng: We’ll share more on this later. This is quite a challenging question. I’m also curious about what others think. What we’re divulging now is what we’ve decided to go ahead with.

Garage 42: Currently, in the field of cockpit, the industry is discussing ‘Car-Human Connectivity’. How do you consider the interconnectivity between mobile phones and cars?

He Xiaopeng: We are doing two things in this field. First, we aim at going global, and we can’t ignore the Apple and Google ecosystems around them; so we are doing something about it.

Second, how should current phones connect with current cars? How should future phones connect with future cars? On this issue, our thought is UWB. Except for the MONA project, which I’m not sure about yet, all our car models will support UWB from next year, regardless of the level. I think UWB is the future trend, its main cost is an increase of 100 dollars per car. Under this trend, we are bound to make some changes on the automotive side.

Garage 42: Will you make mobile phones?He Xiaopeng: I believe making mobile phones is cool, but it’s not what Xpeng should do.

Garage No.42: How would you judge whether something should or should not be done?

He Xiaopeng: My core logic always revolves around the end goal. Firstly, I look at the direction; secondly, I consider if it’s something within our capability. I am adept at determining long-term strategies, and Feng Ying is excellent at breaking down a strategy into detailed sub-strategies, methodologies, and plans within a 12-month timeframe. Therefore, we make a great team.

Garage No.42: Why do you think making flying cars is something that should be done?

He Xiaopeng: When I decided to make flying cars, almost no one agreed with me. However, back in 2014, I believed that flying cars were the future, with an estimated landing time in 2030. From a different perspective, without having laid out the plans a decade ago, making it happen would have been impossible.

Garage No.42: What is the fundamental logic behind your belief that flying cars can be realized?

He Xiaopeng: It’s a combination of a few factors. First, it’s technically feasible. Next, it’s policy feasible. Of course, there are quite some difficulties policy-wise, but if it’s viable in overseas markets, China’s market just needs a breakthrough to follow suit. Moreover, creating a brand new track without disrupting anyone’s cheese would make it relatively easier. This is also part of my consideration.

Garage No.42: The lines between businesses are getting increasingly blurred for both car companies and smartphone manufacturers, and Xpeng is also working on flying cars and robotics. So, what would be the boundary for Xpeng’s business?

He Xiaopeng: The core logic in defining our business boundary is to go all out to be the leading player. To achieve success, one must make alterations and adjustments in their business operations. Yet, if one succeeds, the lines of business they’re involved in are likely worth pursuing; if not, they might not have been worth the endeavor.

Fundamentally, no matter which business one is in, everyone is in it to win.

The price war will continue, AI will become core competence

Garage No.42: On Tech Day, you emphasized that Xpeng should be a global company and Chinese enterprises should go global and establish R&D units overseas. So, was becoming global a part of your mindset when you initially invested in Xpeng?

He Xiaopeng: Indeed. I have experience with global business, so Xpeng has tried various methods in going global. However, in the end, I decided to apply a consistent approach to technological innovation that is clear, stable, sustainable, and uniform, especially uniformity, which is extremely important; inconsistency would pose a significant challenge to managing boundaries.

Garage No.42: You’ve declared plans for NGP to launch in Europe next year. You’ve already entered several countries and regions, why is NGP prioritizing the European market?

He Xiaopeng: We began deliveries in the European market at the end of last month. Upon entering the European market, it will be slightly easier to approach other markets.

In terms of NGP coverage, we will focus on Europe next year, and eventually expand to other countries. Overseas, there are no approval restrictions on high-precision maps, and they are cheaper. The key is to ensure the technology is highly efficient and secure and that it outperforms our competitors’ products in a way viable to us42 Garage: A few years ago, you always said that the product should not be narrowly targeted, and your choice was to make the best-selling models at every level. But in the past year, I feel you’ve started to talk about positioning. What kind of transformation have you undergone?

He Xpeng: The main concern is Feng Ying. We are now talking about a concept called category innovation, which means finding an innovation point within the large track. This innovation point may initially seem small in scale, but it’s easy to launch into, then expand upon this basis, ultimately reaping a considerable dividend on this large track. The logic here is the same, but the entry path might be more exhilarating for us.

42 Garage: So, what is the latest product positioning for Xpeng?

He Xpeng: Our logic is consistent, that is, to use intelligent driving as the core force driving our product matrix development, including at the level of 150,000, we will have a brand new one.

You can think of it as, our previous strengths as core capabilities will continue to advance, but our previous weaknesses will be dramatically extended, such as cost control and platforming capabilities. These can both reduce research and development costs and improve product quality, while increasing the flexibility of the supply chain.

42 Garage: You now emphasize cost reduction very much, including mentioning three aspects of cost reduction in activities yesterday. Is this related to this year’s fierce price war?

He Xpeng: Firstly, no matter the price, we need to control our costs well first; this is a fundamental aspect, while it also improves the brand premium ability to gain appropriate profits. Secondly, economic problems will persist. This ongoing situation does not necessarily represent bad news. I think it may have a positive side, but we need to brace ourselves for significant risks, maintain high-profit margins and store enough food and fodder reserves.

What I mean is, as an entrepreneur, it’s unwise to only consider things from a long-term perspective; we must also stabilize the present. The storm we faced last year was indeed the result of problems accumulated from 2020 to 2021. So we need to make up for our weaknesses in manufacturing and organization to meet the current changes. In these changes, we see great prospects.

42 Garage: Why do you have a great opportunity?

He Xpeng: Because many of our competitors, I feel, have a negative competitive power. This isn’t a pleasant thing to say but their competitiveness is relatively low. At present, some car companies may seem to be doing quite well, such as having high sales in some niche markets, but they are short-term or dependent on some external advantages. However, I believe what is really important is comprehensive ability, which is long-term.

42 Garage: So, do you think sales are not the most important?

He Xpeng: Sales with value are the most important.

42 Garage: What is sales with value?

He Xpeng: For example, in the Chinese market, we look at the insurance volume of To C, which is the most valuable. As for the insurance volume of To B, I think it needs some discount. If you make electric cars, the value coefficient is slightly higher; if you make gasoline cars, the value coefficient is lower, as the decline of gasoline cars is inevitable.

If the brand is strong in the overseas market and the country is good, the value coefficient will be slightly higher; if the brand is not your own, the value coefficient will be very low, close to 0.1. If it’s an independent brand and the country is a bit weak, there’s no choice but to wage a price war. This competition logic is exactly the same as the competition logic of the Internet.42 Garage: How long do you think this domestic price war will last?

Xpeng: I think it will last a long time, until 2025 or 2026. There’s a key reason for this: there’s still profit in gasoline cars.

42 Garage: So, it will continue until there’s no profit in gasoline cars?

Xpeng: Currently, the biggest strategy for gasoline cars isn’t to improve product capacity such as aesthetics, performance, etc., but rather price reduction. For example, after Honda’s recent price reduction, it still effects sales a bit. I believe there’s still a 15% room to drop.

42 Garage: Xpeng announced collaboration with Didi this year, some believe this is to increase sales under high competitive pressure. Do you agree with this?

Xpeng: Regarding the collaboration with Didi, after talking with Cheng Wei, we agree that Xpeng is better at manufacturing cars, and Didi is better at operations. So, we just exchange our abilities. Didi gives us their car manufacturing business, and we can do it better and cheaper. In the end, its To C sales will be magnified on existing foundation, and its To B business can be done well on this foundation.

For us, it’s a win-win. As Didi is also our shareholder, they are motivated to increase sales. We have high expectations for its sales in both the To B and To C sectors next year when this car is released.

42 Garage: So, Xpeng also hopes to gain some sales in the To B area.

Xpeng: I’d rather see it from another angle. For us, the car’s To B sales are mainly to ensure that we manage to launch in scale at the 150,000 price point. At this price point, volume is important. If a certain scale is reached, costs can be reduced, and compete in the To C market – it’s a cycle.

In the ten-thousand-dollar price range, the hardest part isn’t making a good car, but how to scale up a car. Once reaching a certain scale, cost disbursement can be completed in a year or two. At that point, when competing with cars in the same price range in the market, they’ll find out that they lack competitiveness.

42 Garage: The styling of Xpeng’s previous car models has always been a controversy and some believe it’s a weak point that has reduced the competitiveness of Xpeng Motors. What’s your take on this?

Xpeng: We’re working on improving this. For instance, implementing some product review reforms, or introducing some exceptional designers. But that’s not enough.

After some reflection, our previous styling issues were also a problem with the project management process of the car model, being given little time to finish. This can’t yield a good outcome – even if we bring in the big guns. So, what’s more essential, is to establish a good process system.

42 Garage: The public might hope to see a unity in Xpeng’s styling design, like having a unified visual design language.

Xpeng: We’ll establish our own style. If by unity you mean being formulaic, I won’t agree. I think being formulaic is meaningless.Garage 42: In this 1024 Tech Day, you emphasized “AI defines automotive”, so does AI define products or the industry?

He Xiaopeng: I believe for the ordinary customers, it’s mainly about the product. Because changes in product promote the industry, not vice versa.

Garage 42: As AI defines car products, if automakers lack such product-defining ability, they may not keep up with the pace of times. So do you think AI is an indispensable ability for car companies in the future?

He Xiaopeng: AI will become a basic capability of automakers, and indeed a core basic one.

For instance, in future when you’re buying a car, it’s like how you’re buying an iPhone or Android phone today, would you purchase it if it can’t install any apps? If AI reaches such a tipping point, then this capability will be increasingly important.

Garage 42: So when can we reach this tipping point? Will it be in 2025?

He Xiaopeng: By 2025, we should reach the tipping point in terms of awareness, though not entirely in terms of abilities. But once we reach this point, the presence of AI will be indispensable. Then, automakers should either do it by themselves or find partners. I believe the likelihood of achieving this through partnerships is dwindling. Buying software is easy, as is buying services, but purchasing AI as a combination of software and services is very challenging.

In terms of this tipping point, I’m uncertain if there’s a definitive timeline. But I’m sure it’s approaching faster than we think.

Organizational Structure Will Continue to Evolve over the Next Five Years

Garage 42: Primarily, in the Xpeng system, you managed the smart team you’re familiar with and the automotive technology wasn’t your responsibility, but now all tech teams are under your aegis. How do you manage unfamiliar tech business in the team?

He Xiaopeng: Coming from a tech background, I believe there’s a lot of commonality amongst different techs. Let’s consider power technology, initially it felt challenging, but later I found a direction to solve the problem, such as how to compete with gasoline cars, demystifying prolonged winter anxiety, dealing with spontaneous combustion, controlling the costs, and managing the system.

The same goes for other areas. I set a direction for the team with a timeline of 6 months, 12 months, or 18 months, and then they offer me different options.

Garage 42: In this process, how do you identify problems within the team?

He Xiaopeng: The previous issue was, they thought they achieved 60 points, but reported 40 points, and then told their superiors they scored 80 points. Hence, earlier this year, I got rid of half of the management in the power department who claimed they were doing great, but on investigating, I found otherwise. They were resistant to change and disagreed with my decision, so I offer them a chance, the second time – it’s an immediate replacement.From another perspective, it involves raising one’s own standards, using the most basic logic to think about matters, and then identifying the right positioning. It’s also about discerning what needs to be done well and what should not be done. For instance, at the end of last year, I laid off our entire battery cell team. Why? Because they secretly built a battery cell factory and I decided to sell or lease it. They asked who would bear the losses of Xpeng, to which I replied, I would.

42 Motor Garage: So, bad management of the battery cell was a misstep in your past technical management?

He Xiaopeng: Yes. Last year, I discovered that a single model had eight types of battery cells. The development cost for one type of battery cell was 35 million, with an additional management cost of 35 million. You can imagine how much money was involved. All of this money was company expenses which included R&D costs and management costs. After taking direct control, I mandated the use of only two types of battery cells for all models, significantly reducing costs and improving efficiency.

Therefore, the core issue is how to balance safety, battery life, and other aspects while ensuring diligent execution of tasks. Just now, I was discussing logistics and supply issues of a peripheral component of the battery cell with a supplier.

42 Motor Garage: It seems like your contact with suppliers has greatly increased.

He Xiaopeng: Yes, this was necessary. Many technologies can actually build your own understanding, or you can consult different suppliers. Essentially, it involves going out and learning. This allows you to step back and observe the industry, then apply what you’ve learned. Thus, I’m confident in my management approach towards different tech teams as I’ve overlooked products and technologies for many years.

42 Motor Garage: Can you still write code now?

He Xiaopeng: I might not be able to write algorithms as I’m not a particularly good programmer. But I do have a solid understanding of the objectives of an algorithm and what makes a good one. I oversaw testing management for many years and frequently checked others’ work.

42 Motor Garage: After deciding to restructure the organization last year, you took several actions, such as laying off the battery cell team as we discussed, and firing 12 executives during your last press conference. It appears that you have established a more assertive feedback and decision-making mechanism.

He Xiaopeng: Yes. Now the decision-making mechanism is straightforward—I make the decisions. Previously, we used to make collective decisions; I later realized, there’s no need for collective decisions, I can decide. If there are objections, we can discuss; if after the discussion there’s an agreement, then we firmly implement; if there’s no agreement, we implement my decision. Anyone who doesn’t adhere to this is let go.

42 Motor Garage: So, you think that last year’s G9 controversy was due to collective decision-making, not management issues caused by individual personality?

He Xiaopeng: Yes. Over-balancing is not right. In 2019, the company had just over 1,000 employees, but by 2022, the company grew to 20,000, now there are over than 10,000. With this rate of growth, the company’s management and operational capacities were not fully established. Too many people were using experiences from the past to handle current situations, something that is insufficient.The competition is changing, technology is innovating, the team has grown more than tenfold, the management difficulty has increased dozens of times. If you don’t adapt, the company will collapse. Therefore, unconventional methods must be applied to overcome unprecedented challenges.

Garage 42: Last time at the G6 launch event, you mentioned that organization structure adjustment was only partially completed. Up until now, are you satisfied with the state of organization restructuring?

Xpeng: Impossible. Even the phased adjustments are yet to be completed. It’s a process of large, medium, small, and constant micro-cycles. Building a sound system in an enterprise could take about five years in the best circumstances. Under normal situations, it might require seven to eight years.

Garage 42: So you’re looking at five years for this round of adjustment?

Xpeng: Yes. Starting from this year, the following five years will be a constant state of adjustment, but it will get more comfortable. The future adjustments will not be major shifts, but micro-shifts.

Garage 42: You mentioned last year that you do not like dealing with the details of business operations. What about now?

Xpeng: There is no other choice. If I don’t do it, who will? We are hiring in all areas, including additions to the smart-driving team.

Garage 42: How would you assess the current state of Xpeng Motors?

Xpeng: At the end of last year, and beginning of this year, my judgement was: If we do not adjust the organization, the length of the company’s survival is just a question. In the second half of this year, we are getting into micro-cycles, by Q4 next year, we expect to be in a high-speed positive cycle. With our organizational changes and various alliances, I have more confidence.

Garage 42: So you believe Xpeng has already entered a positive cycle.

Xpeng: It is a small and weak positive cycle now. Morale is rising, sales are climbing, gross profit margin is gradually improving, but it’s not strong enough overall. We’re just continuously climbing small steps.

So what does a ‘high-speed positive cycle’ mean? It implies that the company’s sales have reached a considerable scale and are still moving forward at a high speed. Only then it has value. Everything else is a sideshow. All of my logic is aimed at doing big things, sideshows are pointless.

This article is a translation by AI of a Chinese report from 42HOW. If you have any questions about it, please email