Bayerische Motoren Werke Aktiengesellschaft (BMW) Earnings Call Transcript & Summary

Dear analysts and investors, thank you so much for joining us today to our BMW Group Digital Day. Most of you know me, my name is Birgit Böhm. And I'm very happy to have so many of you in our stream today from around the world. It is my pleasure that I have with me today 2 experts to discuss the BMW strategy on digitalization and connectivity from different angles. I would like to start the discussion with Dr. Christoph Grote, Senior Vice President, Digital Car of the BMW Group, who will focus on the engineering and development side. In the second roundtable, I will be joined by Dr. Peter Henrich, Senior Vice President, Product Management, BMW Connected Company customer and sales, who will give us insights on the customer journey and the competitive advantages for the end customer in the marketplace. Christoph, thanks for being with us here today for the next 50 minutes or so.

It's my pleasure.

I would like to start with asking you a few questions, but then of course, we will have time to also answer questions from the analysts and investors on the line.

Go ahead, I guess.

So let me start with the first one, Christoph. BMW presented the all-new iDrive with the eighth-generation operating system something like 3 months ago. And once again, it sets a standard in digitalization and connectivity. Where do you see the main differentiating potential for -- from this technology for us?

I guess every OM has to take a firm stance on what he wants to focus on, where does he want to truly set differentiating features. For BMW, BMWs have always been very functional cars that really serve our drivers that are an extension of what they want to do. The first thing springs to mind is certainly the whole digital experience when it touches the human, that is HMI. The Human Machine Interface from the BMW side always have to bridge 2 things. It's got to be immersive, it's got to be functional and it's got to be natural. And that's true before you even approach the car, for example, on the BMW -- My BMW app, while you drive and after you've actually left the car. So that's one field, which is certainly a core field where we do invest a lot of software engineering. The HMI has got to be easy, intuitive and also safe. And that's not trivial because sometimes what looks beautiful is less than optimal in terms of safety. What we try to do, Birgit, is always to have a safe HMI that's artful, that's mindful and at the same time, it's very functional. We do bear a Bauhaus gene here in our products. Another aspect that's -- that matters a lot to us in terms of differentiation is we don't see the product as a static product that we ship and then remains as it is when it was leaving the factory. With the introduction of Remote Software update, 3 years ago, that's been part of a strategy where we say our products live in the customers' hands. So when you buy something as a customer, you can rely on this product gradually improving while it's in your hands. You can also rely on the product not losing value, but in a way, gaining value. And that's from a rational point of view, but also from an emotional point of view, certainly very important. I guess there's 2 more things that I'd like to touch upon briefly. One of them is always cutting-edge connectivity. Looking at the iX, we've got a double 5G capability. That's certainly one of the first in the world. I think if you look at the detailed feature set, then there's no other car that has this dual active, dual SIM 5G capability. To finish up in terms of differentiation, we have -- we are very careful about data. For us, I think digital privacy is unseparately intertwined with our brand, and we mirror that in the technology of our products. We're very careful about that.

Thank you. Different area. There are very different approaches for OEMs when it comes to the operating system. And why do you think that a relatively small player like us is successful in developing the softwares ourselves?

That's an interesting question because I'd always turn this around. I think anybody who thinks because he's got 10 million vehicles, he can do it, while somebody who's doing 2.5 million can't do it. I think that's pretty dangerous because if you can't do it with 2.5 million, you won't be able to do it with 10 million either. What we do is basically on the -- there's a few key ingredients, how to master this. One of them is certainly that we have got a very long-standing track record when it comes to software. We are not just discovered software a couple of years ago. We've been building our software basically for the past 15 years. If you look at the key component where we said there's little synergy to other OEMs, and there's a strong differentiating potential, for example, the Human Machine Interface, then that's something that we build ourselves. And we've been doing this, as I said, for the past 15 years. That's certainly one ingredient. And I think really the product becomes a better product if you have the full in-house capabilities to code it, but also to have the functional and psychological know-how in one room, that is.

But then we do work with other tech partners. So we -- I know we work with Apple, with Tencent and others of that kind. So how is that collaboration working?

I think that's the flip side. I try to be very polarized in terms of what we do and what we do with others. If you take, for example, the Tencent example, having access to the Chinese consumer electronics ecosystem is really important, and we can't go it all alone. And that's why we chose to implement the MINI program platform, which is a smartphone-based platform we codeveloped with Tencent, basically the port of this platform into the vehicle. It took us 11 months from the initial talks to actually deployment. And so that now a BMW customer has full access to the full-blown MINI app platform and ecosystem. So that's another example. We do the same thing also with players like Apple. So the rule is basically, there are areas which we think are important to the customers that do not scale for BMW alone. And as I said, I think they won't scale for somebody who was 10 million either. There we go into a codevelopment with the leading tech player. And from there on, it's fine if others use it, too.

Okay. Great. Last but not least, I would like to touch a little bit on the topic of the transformation. We will launch the new technology flagship, BMW iX, of course. And your employees in the digital car units are already working on the upcoming upgrades of the Operating System 8 while already in parallel are working on the experience of die NEUE KLASSE. So my question really is, how are we ensuring the execution of this successful transformation?

I think there's a few key success factors that are important to observe. I think what does not work is basically to have one big bang after another. I think what we believe in on the BMW side that it's a long journey, it's a continuous journey. And for example, coming back to software, what is key that a lot of software that's been developed for generation X, like, for example, the iX now, will be carried forward into the next generation, but also in between generations. If you look, for example, at what software update did in 2019, then a lot of extensions have been built after SOP. For example, the whole navigation migration from onboard navigation to cloud-based navigation, that's something that was developed in a continuous stream and deployed via Remote Software update.

Okay. Great. Well, thank you so much for these contributions. Before we open up for the Q&A session, I would like to remind all of you of some of the technical instructions. [Operator Instructions] We will answer your questions following the sequence in which they come in. And I was also told to tell you that please be aware there is a lag, a delay between the streaming and the telephone line. Okay. We will now start with a Q&A session right after a short movie. [Presentation]

Okay. Operator, please give us our instructions for the telephone line.

Yes, of course. [Operator Instructions] And the first question is from Patrick Hummel, UBS.

It's Patrick from UBS here. My first question is about the revenue opportunities related to the digital car. Can you talk a little bit about how to monetize the features? What is just going to be standard and expected by consumers? And what do you think you can actually monetize, be it through monthly subscriptions or so? And do you have any thoughts how much additional MINI consumers or BMW owners are going to spend on this? Are we talking about prospectively another EUR 10 per month, EUR 50 per month, EUR 100 per month or even more? And my second question relates to China. We're seeing the emergence of very strong local players, in particular, in the EV space. And when you ask these consumers why they love the Chinese EV brands, it's not just the fact that these are good electric cars, but that they also provide a great digital user experience to the extent that it even becomes a threat to the global premium brands like BMW. So I'm just wondering what your thoughts are as far as the content requirements in China are concerned. Will you have to run at an even faster pace there? Will you have to have more digital content in the vehicles sold in China than elsewhere to fight the local competition?

Patrick, if I may, your first question on revenue opportunities, would maybe be one that I would like to keep, if I may, for the discussion with Dr. Peter Henrich because he will be coming and discussing with us the digitalization and connectivity from the customer and sales side. So if you don't mind, let's park that question and ask it to him because with Christoph Grote here, I would like to cover mainly the engineering parts. But I think on the China, I think you can -- Christoph, that would be great for you to...

Thank you for the question. I mean looking at today, we indeed see already that, for example, recent successes on the Chinese sales were clearly traceable to digital performance so also on the BMW side. Now what have we done in the past and what will we do in order to make sure that competitiveness will be sustained. One thing is that if you look at what we do today, a, we've got a sizable digital workforce on the ground in China today. So we do have a local development team that develops close to customers but also close to tech players. I mentioned the 11 months that it took us to basically launch the MINI app platform with Tencent and WeChat. That's only possible if you have people on the ground that speak the language, that know the technology and that also have the network. What we put on top of that is we recently founded a software company in a joint venture in Nanjing, China, which is ramping up very quickly and where we find it easy to attract excellent talent in order to build more software capabilities on the ground in China. I guess the last point that, that's important is basically having a trustful relationship with Chinese players. Tencent is one. Ali is another one, but there's also a lot of smaller companies that we intensely cooperate with. So in a nutshell, for us, I think China remains an opportunity, and we've got the engineering operations on the ground in order to let that stay this way.

The next question is from Horst Schneider, Bank of America. Horst Schneider, Bank of America, we can't hear you at the moment. Perhaps...

I'm sorry. I was on mute. I hope you can hear me now.

Now we hear you.

Yes, loud and clear.

Perfect. Yes.

All right. Okay. The first one that I have that relates again to your comments with suppliers you also need in the future. And you have mentioned Apple and the likes, but you haven't mentioned the classic Tier 1 supplier. So I just want to learn to what extent you still need the Contis, the Bosch, the classic Tier 1 suppliers for your future software work. That's number one. The number 2, again, coming back to your operating system, I just want to understand what part of the operating system you really do yourself. So is it really the complete operating system? Or is it more the interface that you want to have under your control? And for that, what the customer is not seeing, you still use suppliers just because that is more efficient and easier for you.

All right. Well, thanks for the 2 questions, 2 excellent questions. Looking at the Tier 1s, let me quite frankly say that I wouldn't want to change with them because the need for transformation is probably a bigger one even than on the OEM side. Now why is that so? If we look at what we do on all performance compute platforms, then already today, in the past generation that's been on the ground since 2018 and even before, we retained control over the whole software side. So what we buy is not a classical silver box that was specked and then -- but fully developed by the first year on all big ECUs, like, for example, the driver assistance ECU, the infotainment ECU, those compute platforms are integrated by BMW. So the supplier, the role that will certainly remain is that of a hardware developer. We don't plan to go into that right now. And we do the software on top of that. We think that if a first tier tries to ship a product as a big bundle, we are not going to be the customer. We'd like to buy best-of-breed software components, hardware capabilities, but we do the full-blown integration ourselves. So that role has to change on the first tier side and those that manage to change will be our partners in the future. Those that stick to a bundle, probably not.

So with that, you probably almost answered some of the -- how much do we produce ourselves, right?

Some of it. I think the second question also refers to software specifically. I've seen a lot of messages in the past where people stated they would do kind of 60% of the software in the car. I can't follow, to be honest, because if you look at, really, a total number of lines of code or binary sizes, then I think the truth is -- to take one example, the Linux platform. We use a commercial Linux distribution. And I have no plans whatsoever to become a Linux distro. It's like in -- for example, in telecom networks. You have professional distros that turn open source software into proven distributions. We use them. What we do is we integrate those distributions with a specific tool chain to basically have a perfect developer productivity and also with a lot of on-top components that build the framework so that we have a platform. But the -- what we try to follow is we have control, we have continuity so that we don't throw away platforms from generation to generation. But whenever we can buy something as a software component that we integrate into a platform that we control, we buy it. And that works well. It's been working well. We extend the scheme. And also in the next generation, we'll see the same pattern.

That's great. Just to follow up on the first question that I asked. But then you say, okay, the suppliers need to change and you take the best of breed. The classic Tier 1 suppliers are still able to offer the best of breed also when it just comes to software.

I think there's a few examples where a supplier decided to unbundle what they have to offer. Why they said, for example, I offer one particular application where I think I'll be competitive in the future. Let's take, for example, computer vision. That as a software component is fine. Having said that, I think your point remains a sensitive one because the squeeze doesn't just come from our side, integrating downwards. The squeeze also comes from the bottom where more and more silicon suppliers that we entertain direct relationships to extend their value added into the software layer. That's why the -- they get squeezed in the middle a little bit and focus is certainly essential.

The next question is from George Galliers, Goldman Sachs.

The first question I had was just with respect to scale. You mentioned that you did not see scale as a big differentiator with -- as it pertains to software development. When we think about BMW's investments in software today, do you think the total investments in software will need to ramp materially over the next 3 to 5 years? Or do you think the annual rate today will remain relatively constant over the course of the coming years? The second question I had was with regards to the 4 differentiators. You mentioned the HMI, dynamic product, connectivity and data privacy. As Patrick mentioned, obviously, there are a lot of new entrants in the auto market. You claim to have very strong software and digital capabilities. When you consider these differentiators, which do you think present the biggest hurdle or barriers to entry for the new market entrants?

All right. That's 2 broad questions. So first, let me talk about the software investments. What we've done in the past and what we'll continue in the future that will increase our investments into software. If you look at the number of people or the money we spend, then it's clear that we have a very continuous growth over the time. Now that continuity, I believe, is important because if you grow too quickly -- I've never led a software organization that was particularly successful when it tried to grow too quickly. So we have a very continuous growth. If you look at the amount of software in the car, let me give you a figure. The binary image of an X5 2018 generation was about 1 gigabyte. Now the binary image of an iX is already 1.8. So the reality of the software is that it does grow in terms of how much code is in the car. And therefore, what we'll always do is we'll reconsider what we do ourselves. When there's a commercial package, we buy it. It's a dynamic game. In terms of scale, I think what I'd like to be more precise upon, we don't try to seek our scale over the volume of cars. We try to seek our scale, for example, through synergies with, for example, the network equipment industry. A Linux distribution that's good for telecom grade networks is good for automotive, too. That's quite another scale than just looking for another 1 million cars. So that's the attempt to address your question number one.

The second one, I think, was which is the -- is there a biggest hurdle for entrants on the differentiating factors, right?

Absolutely. I think one thing that we shouldn't forget that function -- software is not just code. We are -- the organization doesn't consist of code monkeys. The -- what the -- I think the interesting thing, the interesting capability comes when you've got in one brain, the ability to code in an excellent way to do it at scale, that is with a lot of people together because a car is a big -- it's a huge software project. It's not a small one. It's not -- it doesn't depend on the individual genius. It depends your ability -- on your ability to actually have thousands of people in all geographies work together and combine that with functional knowledge. Take HMI, again, as one example, but I've got numerous others. Our software engineers always have 2 skills. One of them is a particular software technology skill and the other one is the -- is a functional skill. For example, HMI. HMI is psychology. HMI is experience, what works in reality, what doesn't work in reality. And it's not that easy to build an organization that actually has excellent people that have both capabilities in one brain.

The next question is from Arndt Ellinghorst, Bernstein.

Yes. Christoph, I think what investors and analysts are struggling with at the moment is on the one hand, we're being told software is the most important thing that will define the future of carmakers in the future. And at the same time, we don't really understand what's differentiating different players from each other. And we don't really understand how it's driving more consumers towards BMW versus Mercedes or Audi or others. Can you help us understand what these real differentiating factors are? Or whether this is more like a hygienic technology almost, something that has to work in a car rather than something that's really defining differentiation, especially in the premium space? And we obviously all see at the moment, if it doesn't work or if there are problems -- like with some of the VW EV launches that it's becoming a problem. But I think it's hard for us to understand how you will really differentiate in the future. And then I have one follow-up question.

All right. So thanks for the excellent question. Let me indeed start with the one necessary hurdle you've got to pass. Not everybody, I believe, will master the challenge to build these huge software projects and actually maintain that over time. But is that a differentiating factor? No. It will only select a few who kind of fail to deliver. Now on the core differentiators, I think -- I'll give you a few examples where you can really see that software is the enabler in order to actually build excellent products. So let me give you examples. I won't touch upon the HMI because I did that before. I'll give you other examples. If you look at the driver assistance system, for example, that's been hitting the road in 2018 and compare it to the next generation that we're building now, then again, you see that it's not so much the feature count. If you superficially look at driver assistance systems, you see a lot of similarities. And I understand your question, what's the difference. But if you actually drive that and you see how much craftsmanship have -- has gone into it, then you start to see differences. And those differences are built on a technological foundation, which is very different between various OEMs. Touching upon that example, now what do we do? We always have generations of driver assistance systems. And we always intensely gather data on the usage and on the surrounding of our cars on generation X. We use that data. We process that data in order to really hone the functionality, to extend that functionality in the next generation. That, for example, is a capability where -- like in the past, if you think about, well, every cars have steering. Why should I buy a BMW? Then I would have said, well, just hop in and try. And it's the same with a lot of digital functionality. So I think it's a misconception that digital is there or not there. I think that's a wrong interpretation of digital. Digital requires craftsmanship, long-term software assets, long-term data assets, and we deploy those in classical fields of differentiation. Like everything that touches the human, that really matters to us. Design. In the past, hardware design has clearly been a key success factor for BMW. And the same holds for the digital side, and it's a misconception to think that you can easily copy that. There's a lot of secret sauce in that, too. So in that sense, I guess, I could give you a long string of examples where you see that something being digital doesn't mean it's easily copyable. It's just the clay of the future to craft differentiated products with.

Cool. And the other question I had was, if there was a headline popping up on my Bloomberg screen, BMW and Apple are into a deeper collaboration, whether that's vehicle manufacturing or BMW taking some of the software ecosystem from Apple in-house, a deep collaboration, your stock would be up, say, 20% and outperform and people would applaud it in the capital market. Can you talk, from your perspective -- we might get this totally wrong, right? But from your perspective, how would you look at a much deeper and broader technological collaboration with one of the big tech players? What are the pros and cons, from your perspective?

I mean that's a broad topic. The -- we have a long tradition to collaborate with Apple on technology. Like, for example, the new smart access, I've always been convinced that carrying a big key in your pocket and having a smartphone in the same pocket is not a good idea. And that's both radio technology as well as digital technologies, on the phone and on the car side, are to the advantage of the consumer. So we started -- initiated talks with Apple a long time ago in order to build a new system. And we did it in a codevelopment way. And as you know, it will be launched with the iX. And from there, it will spread into the industry. Now I think that's an important point. A lot of digital things have to scale. And the idea of building a proprietary technology for Apple for a digital car key, I think it just doesn't scale. And therefore, I very much value the position of being a very frequent lead developer with those companies, whether it's in any way realistic or advantageous for both sides to enter into anything exclusive would be something that would need to explain to me.

The next question is from Stephen Reitman, Societe Generale.

My question is about the software development as well. You mentioned -- you gave an example that the iX has 80% more lines of code than a 2018 X5. I'm just wondering, with operating system made, how much carryover code is there? To what extent, with the new architecture in the vehicle, you have to design code from the ground up rather than try to do things accretively additively, which, again, has been mentioned by one of the previous questions, I think led to some problems with the launches of some of the car companies? How much is it completely new, particularly as you move to -- from -- towards a centralized domain systems, with some ECUs having dedicated chips and the like?

Well, thank you, Steve, for that excellent question because I think it touches on software reality that is frequently ignored. If I look at what fraction of the 2018 code made it into the 2021 generation, that's about 80%. So we always build code that even if we change hardware partitioning, we aim for code to be carried into the future, sometimes changed, sometimes -- most of the times extended. Now that is a philosophy which is important for a number of reasons. One of them is, basically, you can waste a huge software workforce in recreating what you've had before. I don't believe in that. If you look at the iOS for example, that's made in the same way. It's not reinvented from scratch every year. Every now and then, maybe every 2 or 3 generations, we see that we completely need to rearchitect some parts of the software because requirements have changed too much or because we need to render it more efficient. That's cleanups that are necessary. But we always aim for an evolutionary software development here. Now that's also important for another reason. I'm a strong believer in basically the idea that customers want the car as a perfectly managed device where software is kept always fresh. Now that only works if you keep the software workforce working on one stream of software, which is continuous rather than basically chasing them from one big bang to another. With that continuous software stream, we are able not just to deploy software to 2.5 million cars by the end of the year, but also to have software to deploy, which is the other side of the metal that you need to fulfill.

And the next question is from Henning Cosman, HSBC.

My first one was just on your double 5G capability, the dual SIM that you talked about in your intro remarks. I just wanted to ask what that is and why that's important. Why did you mention that specifically as a key differentiator? The second question -- I hope that's not too basic, but I found it very interesting when you talked about the software components and what you may be buying. I think computer vision was one element that you mentioned specifically. Can you just talk a little bit more, if you don't mind, about sort of base layer, middle layer, application layer? I think in the first part of this roundtable, we talked a lot about HMI and you're doing mostly the things that directly touch the human. I'm sure, in some ways, you go deeper. If you could -- it's slightly related to Horst's question in the sense of what can you still buy and where would you buy things that are also lower down in the middle layer or base layer. I think that's still a bit of a mystery to all of us, where exactly you think it's important enough for you to do it yourself.

All right. Maybe let me start with number 2 because it's maybe the more complicated question. So going from bottom up to the top, let me touch on a few areas. We buy hardware today, essentially from first tiers, and silicon suppliers are second tier. In the next generation, a lot of that will be changed where we have direct relationships with silicon suppliers. Now with that hardware comes usually the first bottom layer, basically the board support package. We don't build board support packages, but we buy them from those people that design the silicon that has to support that BSP. Now on top of that, operating system. Operating systems, we don't build operating systems. I just mentioned the Linux example. We use commercial Linux distros and use that Linux distribution. We do influence the distribution through requirements, and we integrate the distribution into a software package that also supports automotive-specific extensions of that Linux that native Linux does not have. On top of that, communication frameworks in the car, security frameworks in the car, frameworks for Remote Software update, diagnostics. Also, for example, health of is ECUs, health of pieces of silicon, health of sensors, general system health capabilities. Those are all shared, platform-type capabilities that are not part of a Linux distribution, but -- which we build ourselves. Now going up from that, application layers. We are polarized. Let me just give you one example, navigation. In navigation, we buy a very commercial package that we did not specify. It's the product as it is by a software vendor that today, I don't want to mention yet. And we supplement that with everything that's not on his road map, we build that ourselves and keep that as a long-term asset. Now we do this for navigation because the car is about mobility, charging is very closely intertwined with customer needs and needs the machine has and therefore, we do it in this very polarized way. The top layer, to be sketchy -- HMI we do 100% ourselves. Now I left a few areas open, but I just tried to give you a flavor that we try to be very black and white. Either we buy something or we build it. If we buy it, we want to be nonproprietary, but scale throughout the industry. That's the philosophy I'm driving through the organization, not just today but for the past years, to be honest. Now 5G...

That's really helpful. Okay. Can I just ask where autonomous driving would fit in? What layer? Is that on the same layer with the communication and security and diagnostics? Or do you see that on the same layer with navigation? Or is it both?

No, I think being a software-centric automotive company, we don't take this as a monolithic block. Now there's parts that we purchase because I personally think it just does not scale if we try to build it ourselves. The -- one example is computer vision. So turning images into basically objects is something we buy, today from Mobileye as well. There's other areas, for example, the object fusion is something we build ourselves. So if you tear it apart in, say, something like 30 components, you'll find a very conscious decision on what we buy and what we build. It's not one monolith. But we think we have a software strategy on this where basically, we buy where we think there's a global market through which we scale and other stuff we consider as long-term assets. And we've been building in the past, we will be building it in the future. Now on 5G, I mentioned dual SIM, dual active. So in a nutshell, we use 5G for data communication and for voice communication. The difficulty that a lot of customers find today already is that when they have their phone in the car, the car is shielded against electromagnetic radiation also on the windscreen. So it's harder and harder to actually let the radiation get in and outside the car. And that's why beyond the standard data modem that the car is using very heavily, we have a second modem whereby the access data, the SIM data from the phone gets transferred to the car so that effectively, you use the second radio on the car side in order to your -- to do your personal calls.

[Operator Instructions] And the next question is from José Asumendi, JPMorgan.

It's Jose from JPMorgan. I just have a couple of questions. You mentioned data. And I was wondering with the data you have been gathering now for the past years, are you effectively able to use this data? And how are you using it? And if you could speak a little bit about what your cloud suppliers in Europe, China and North America. That will be the first question. The second question, if you can speak a little bit more about the software. I mean when you carry out the software upgrades, specifically with iX, are there any changes with your software upgrades? Are there any applications within the car that with your now 5G connectivity, you are able to reach quicker? That you're able to change, to update that you were not able to do before? Are you able to reach to some applications within the car, either [ thrust module ], ADAS or interior that you were not able to update before and now you will be able to update?

All right. Thanks for the excellent questions. I think the data question is actually one which is sometimes underemphasized when people talk a lot about software, but too little about data. So thank you for that question. There's one thing that I think in public communications we don't talk that extensively about, and that's the following. What we've been putting into the cars for the past 2 generations is an infrastructure that would allow you -- basically, for each individual vehicle identification number, we are able to deploy specific code in a local compute environment that preprocesses and selects data to be transmitted into the cloud, be processed there from the whole fleet in order to generate services and answer very specific questions. I'll give you 2 use cases. One use case is in driver assistance, we want to know in what situations does the function not work because somebody has been putting his foot on the brake while the system was engaged. We can evaluate those situations. We can transfer a lot of the raw data through that infrastructure and understand what went wrong, to put it that way, or what can be improved. That's one example. Quality issues. Sometimes you are in a situation where you know you have very few cars that have a problem, but you don't know which. So from a commercial or entrepreneurial point of view, it's really important to be able to localize the cars in order to be very selective on which cars you recall, for example. Here, we've got an infrastructure whereby we can run local code, we can analyze, for example, precursors of a problem. And therefore, in a recent incident, for example, it cuts the number of cars we had to look at in the workshop on a hardware issue by a factor of 100. Those are just 2 use cases, but I think it's a flexible infrastructure that actually, I believe, very, very few OEMs have in place. Now on the cloud...

And then yes, that's -- that must be our last question. [ Continue on ] the cloud, please. Go ahead.

On cloud suppliers, we -- in general, we are moving to public cloud. We've got still some things on-premise. We use the usual suspects like Azure and AWS. Cost control is an important one, of course, here. We're not having one strategic partner actually for that reason. Now on software upgrades, I think it's important to emphasize, there's been some confusion around sometimes what part of the car can you actually upgrade. In a nutshell, you can upgrade the complete car. On a -- so you can upgrade all the powerful ECUs. You can also upgrade virtually all the smaller ECUs. To be very precise, there's, for example, ECUs that I don't even dare to call ECUs. For example, a tiny piece of electronics that controls whether you can unlock the fuel flap. Now for example, that is very stable. It doesn't change. It's very stupid. That's one of the, I think, 6 ECUs in the whole car -- calling them ECUs, that we can't update today. But basically, all the rest, anything that's significant, we've been able to update since 2018.

Well, thank you. Thank you so much for those really great questions. And Christoph, thank you very much for joining us here today.

It was my pleasure.

With this, we conclude the first part of our BMW Digital Day. And now it's time for a little break before we continue the program with Dr. Peter Henrich, and we'll be back at 3:00 p.m. Center European Time. For those of you who have joined by phone or on the live stream, you can just stay online. [Break]

Ladies and gentlemen, welcome back. After the inspiring discussion with Dr. Christoph Grote on the digital car, let's shift gears to the connected car and digital sales. And I'm very pleased to have with me today here, Dr. Peter Henrich. He's Senior Vice President, Product Management at BMW for Connected Company, Customer & Sales, and he's joining us for a second roundtable. Peter, thank you for being with us here today.

Pleasure to be here.

So let's start with my first question. We have stated our aim to deliver the industry's best premium customer experience. Now, especially in the field of sales and marketing, of course, the greater use of digital technologies I think offers an opportunity to really create a leadership with regards to customer interaction. Can you maybe just elaborate on how BMW creates this premium digital customer journey?

First of all, it's important for me to show that we really take a holistic approach to what we call the customer journey. It's from the first contact with somebody who's not even a customer yet, but maybe prospect, over this whole process towards a transaction, and then during the life cycle and the usage of the vehicle. And to be open, that's quite a bit of a change also for us as an organization that we have been driving now for a couple of years now. In the past, there was a separation of disciplines. There were the sales and marketing people, and the CRM was one key task. And then there were the car people and the interaction with the car was one. But for the customer, it has to be seamless, one connection with us as a brand. And so we have created a framework that really conceptually describes this customer journey with all the different possible use cases, not every customer is the same, and we definitely look at the different experiences that we can create. And digital technologies bring us a whole new range of possibilities of interaction from website over apps, over in-vehicle conversation. And yes, bringing this out of one touch and feel, let's call it that way, is a key to success. And if you look from a business perspective further on that, it's not only about the experience for the customer, it's also for us as a company to get to know this customer as good as possible to also make customized offers in terms of vehicles, options and services.

Now have you seen somewhat of an acceleration recently just because -- of course, with the pandemic, we're all living a little bit more in the virtual processes. Have you seen an acceleration of utilizing and accepting the digital interfaces?

Definitely. Especially, if we look at the first part of this journey that's before the sale, where we as a company that has a retail level that still is a key enabler for our business. There, all of a sudden, dealerships were closed and a lot of the consultation that normally takes place in the dealership had to be shifted towards the virtual place. And together with also the dealers globally, we did see a lot of acceleration. We implemented programs with remote consultation, things, to be honest, that have been present already in other industries in the past, but now we were forced into that experience and made very positive experience that it's helping for customers and sales force in a lot of occasions. Still, we have a very emotional product that also lives from touching and feeling and ideally also test-driving it. But again, not every customer is the same, and you can do a whole lot more than maybe some people thought in the company digitally and now we do.

Great. One other major topic is -- when it comes to software is the creation of revenue streams. And actually, in our earlier session, Patrick Hummel from UBS, asked also the question on revenue opportunities. So I'm just wondering, how does BMW approach this opportunity? And which services and functions on demand do we offer? And how do we think this unlocks earnings potential for us?

Of course, there is a lot of buzz around this topic, and for us, it's not new. We have started connecting our vehicles 20 years ago. So I would say, we have quite a bit of experience. We have gone through a journey. We have 14 million connected cars out there, and this is the basic enabler to have this connectivity in the vehicle and the possibility to interact. If you look at the business side, over the last years, digital content, digital options has grown in importance. Look at everything around navigation and real-time traffic information, look at the driver assistance systems, all this is digitally software-based in combination with hardware, of course. But there are a whole lot of items in our option list that were not there 20 years ago. So our revenue creation based on digital content is already pretty strong, and it's definitely clear that this is a field for also further expansion. Today, the vast majority is made with the new car sales. But we have already started also some years ago selling via our Connected Drive store digital content after the purchase. So in the life cycle. And we have learned already a lot. We have also solid double-digit million revenue stream per year already today on those offers. And we are clearly dedicated to further grow that. And yes, is it on specific functions or is it in a subscription model? And that was also, I think, a question earlier. Those are all possibilities that we are currently investigating, and in the end, it has to fit to the customers. They have to accept it as the right price-value relation. But I'm very confident, and we have a lot of ideas of what we can create there in the future.

Okay. Great. When we just talked to Christoph, we heard that the BMW iX sets really a new standard when it comes to digital. And it brings the new operating system to the market, which is, of course, the most powerful technology stack that we've ever built. And what makes it so special for the customer now?

What makes it so special? That was also, I think, an earlier question, what is really the differentiating factor? It's not pure feature-counting. BMW brings a soul to functionalities. That's our aim now. And we have, on the one hand, of course, functionality, and it has to work on top-quality standards. But there is also this aesthetics and this emotional experience. So we have -- it's a bit of a marketing wording, but I believe it describes it very well. If we say, well, it has to be tech magic and human-centric. And the combination out of both to create something that is a character and that is feeling and looking differently from other competitors' offers, that is what makes the BMW. And actually, that is not so much different in its basic philosophy from the past where digital experiences were still a bit less important. The way the BMW feels, looks, drives, is different from other strong competitors that also have the ability to drive and decent looks, et cetera. But creating something that makes you feel at home, that makes you smile when you use it, that makes you proud when you own it, that is our aim. The standards that are hygiene factors are rising. So of course, it's a tough competition out there on the levels of standards, and we are, of course, perfectly committed to meet the minimum standards or the standards of expectations, let's put it that way. And then we do the add-on, and that is the art of our work, yes.

Great. When you mentioned earlier the on-demand functions that you can book, that also means that you have to put in some hardware at the time of delivery. Now as a finance person, I need to ask you, how do we make that work in terms of a business case of putting in hardware, you don't know if the function on demand is going to be demanded end to end?

Yes. Yes. Well, more from an internal perspective now, not so much to the outside customer, but from an internal perspective and to see the community that's listening here as part of that internal perspective, we separate into digital services. So that's basically content that is being displayed. On the display, you need there also, of course, a hardware computing power. But that's basically content, be it entertainment, be it music, be it news, et cetera, also valuable things. But there, you are based on the basic digital hardware set. So that's what we call internal digital services. Then we have the second category, function on demand. That is where we use hardware-based functionalities that are booked in the later life cycle, but also available already at the new car sale, normally at least, yes. What are examples for that, there is the famous seat heating, but there is also this big field of driver assistance systems. And the question is, how do we make a business case? Of course, we are not naive and just put everything in the car that's possible. We very closely look at what is an expected take rates already with the new car sales -- and we have a pretty good system now to also estimate the success of innovations -- and then we see what is the rest potential. And there, we -- you have to take a certain risk and say, okay, it makes sense to put it into the rest of the vehicles because we believe with offering maybe a test period or a good explanation later while the customer uses the car, then we can create additional demand for this feature and then we make it available. And of course, the ideal and the thinking that we are now putting into the organization is, where can we make intelligent use of existing hardware for additional functionality? I'll give you an example. We have -- it's a small product, but it's pretty successful. We call it the Drive Recorder, and that's a system that uses the cameras that you have in your car when you have a surround view and driver assistance systems, and then we say, okay, please have a look, and you can book it for a month or for 3 months or even longer. And you can record from the existing cameras a little film, and it's always recording in the background. And if you have, for example, an accident, and if you activate it then, you have the last 20 seconds and the next 20 seconds. So those are things where we are now experimenting, how it's being perceived and also how our development teams are being even more creative of what new functionalities can you create out of existing hardware.

Right. Great. Well, thank you so much. Thank you very much for those details regarding our customer -- digital customer journey. We'll now proceed with our Q&A session. So operator, if you would please just quickly remind us of the proceedings.

[Operator Instructions] And the first question is from Dorothee Cresswell with Exane.

I wondered whether you could talk a little bit about what proportion of sales you think is going to happen online in the coming years? Maybe you have a target proportion. How does that mean or what does that mean for the dealership model? And how can they participate in the digital revenue streams? And then just coming back to that question around functions on demand, is there a benefit from lower complexity if you equip cars in a similar manner? Obviously, not everyone maybe books the functions on demand, but actually, that still makes it viable.

Yes. Thank you. To the first part on digital sales, the development towards digital sale is a continuous one. We already now do quite a bit of the sales process online, and I talked earlier about the consultation. And of course, we are also going towards this final transaction on the -- via digital channel. We are having already some pilots, and we have said, let's go towards 25% in the -- within 5 years. But that does still include the dealer body because also a digital transaction can work perfectly in a wholesale model. So that is definitely something where we're clearly defining what is the role for the dealer, what is changing in this more digital process, but you still need fulfillment. You also don't want to push a customer necessarily only into one exclusive channel. Maybe on the way, he or she decides to go to a dealer for another test drive, et cetera. And we want to be open for different approaches because, again, not every customer is the same. But very clear and we have seen that over the last 1.5 years. A lot of customers are ready to do a very big proportion or even the whole transaction journey digitally, and we will support that with our process.

I think the second question from Dorothee was regard to the function on demand, and how that works also with the reduction of complexity that works.

Complexity is a huge topic, and we are currently rebalancing, I'd say, the way we are defining the different offers. We are, as of today, especially the build-to-order markets, in a situation where we have, from my personal perspective, too many smaller items and it's for real experts, very nice to configure the vehicle very individually with a lot of different items. But we are very clearly going into a way where we bundle more, where we make it easier to configure, which is also an enabler for a digital sales process. But -- and that's also a clear statement from my side. It will not be a one-fits-all BMW as the individualization is an asset for a lot of customers. And I'm talking through the range, especially if you look at colors and trims and different tastes. We want to offer a good range of different choices. And also on the functionality side, there are different preferences and also willingness to pay. But again, it will be a much easier offer structure than we see it today in the European markets. In some markets, if look at U.S., look at China, we are already to a much larger degree in this build-to-stock markets working with different profiles, which are changed once in a while, but we have already very good examples of reduced complexity. And sorry, to come back to the first part of that question. That is indeed an enabler also for more function on demand. If you have more bundles, it's also easier to design the features that are in the life cycle time sellable.

The next question is from Kai Mueller, Barclays.

Super interesting presentation. When you think about the revenue opportunity again, can you give us a little bit of a sense how much of that is sort of digitally enabled? I remember you talked about this 25%. I wasn't sure if that was referring to that. And how much does require hardware to be installed in order to really get this revenue opportunity going? And then the second point, as you mentioned, of course, you have since 2018 already quite a lot of technology on board with those vehicles. As far as I understand, initially, often these services are coming for free with the vehicle when it's at purchase. When at renewal stages, can you remind us, is it after 3 years? So when do you renew it? And how has been the uptake afterwards when people actually have to pay extra for these services?

First of all, to clarify. So it's good that you mentioned it. The 25% I mentioned earlier are the share of sales of new vehicles that are done on a digital channel. So it has nothing to do with a function on demand, but it's the new car sales process. If you then look at the second part, so the function on demand, we are currently in a model where we are selling certain services with the new car, that the majority of the parts are valid for 3 years. And by the way, they are not for free. We charge them. So either they are included in a bigger package or they are individual items, but mainly, we have already clustered to reduce complexity. On the renewal share, we have to say that there is definitely a way to go because they were in the lower percentage, double digits but in the lower part. And we definitely took a closer look on what's the analysis. And basically, 2 main things. The renewal process was not how easy and seamless it should be. And also, in terms of offer, we have been rather limited. There are services that work very well. So I mentioned you an average number, but for example, real-time traffic information is something the majority of customers are also taking after 3 years again. Others, a bit less. So that is, for us, also input into how are we creating our future offers. And also this 3-year time period is something we are looking at and for certain services, we are probably also revising it. And I think you mentioned already earlier, subscription models. It's also an idea where I believe that certain data-oriented services where you have a regular exchange of data are very well suited for a subscription service because people are, in the meantime, used to that. So I could very well imagine going in that direction as well.

So just to check in again, do you think it's the purchasing process for the renewal that has been the challenge rather than the product?

Both. So that's definitely both. Some products -- and so we have a good feeling now already what products work and what don't work. But you definitely also have to work on the process. And there, a direct approach towards the customer is a key enabler, and we are now taking big steps where we have launched last year a new app, My BMW, My MINI app, where we have a much better direct contact. We are also talking about in-car contact. So we have clearly defined the enablers to bring those figures up, but we are starting from a, I'd say, pretty good base, certainly, if we look at the capabilities compared to competition. So we are well prepared to continue that increase of the revenue stream.

And the next question is from Henning Cosman, HSBC.

So I'd like to come back to that EUR 5 billion digital revenue number that you've put out. I don't want to steal Patrick's question, but I think it's what a lot of us are thinking about. So I understand that you're still sort of exploring between one-off payment solutions or is it subscription. You've also, at some point, put out the number of 1 million vehicles being enabled for updates over the air. And then by the end of the decade, it's 10 million vehicles. How did you come up with EUR 5 billion number? And can you just clarify that revenue? Can you maybe talk about contribution margins on that revenue? Should it be very, very high? Does it include packages that you sell into new cars? Or is this digital-only at the renewal state? I mean, ultimately, I'm of course asking myself even if it's 10 million vehicles, if you want to generate EUR 5 billion of revenue from that per year, it's EUR 5,000 per vehicle per year, right? So please, can you just talk a little bit more about how you came up with that number? And how you see it in context? For example, VW is only talking about a 3-digit million contribution from, I suppose, what we all understand to be relatively similar in terms of business model.

Yes. No, it's important that we clarify that because I'm not putting fantasy numbers in there, and let's see in a few years who comes closest to it, no. This definitely includes all the digital-based features that we sell already today in our vehicles. So look at an option list and you find the navigation system, you find head-up system, you find driver assistance system, and that is what we all include in our digital development. So together with Christoph, that's all in scope of digital features. So -- and I'm not so much up to shifting revenue streams with one customer from the new car sales moment to a later part in the life cycle. No, not at all. Whatever we sell and we can sell and can offer to a customer at the point of new vehicles here, we want to sell it. And that is the vast majority of that EUR 5 billion number, which is a projection for 2030. So already today, it's a significant amount, and that's definitely in our business model, staying on the majority part. I want to sell a complete vehicle with strong digital content at the point of new car sale. And on top, we are working on what I just mentioned, additional content that we can offer to the customer at a later point where he or she maybe has an experience that was not possible during a test drive and driver assistance system, for example, obviously pretty well for that because it's sometimes not easy to grasp what is the capability of Driving Assistant Professional, for example, if you do only a 1-hour test drive. But later, when you are in a daily heavy traffic situation, you will appreciate it. So that is then something on top. And maybe also once the purchase price is digested, there is a new opportunity to spend for the customer. So that's the additional stream. But again, this EUR 5 billion number is to the vast majority content that we also want to sell already with the new car sale.

And it's revenue, right? I mean, to be clear that...

It's revenue, yes.

And are you prepared to give us a number that's only for the over-the-air update type of revenue streams? Is that possible relative to, for example, Mercedes is EUR 1 billion or VW's 3-digit million number. Is it 10% of the EUR 5 billion, so EUR 500 million? Or is it too early for you quantify that?

I would prefer not going into that figure race with the competitors here today, but we definitely take it as a very serious business opportunity. And again, I'm very optimistic that with the better knowledge of the customer and the situation he or she is in, we can make those offers that are being appreciated and that also create revenue.

Okay. I think now we have Patrick. Let's see if we answered your question.

The next question is from Patrick Hummel.

It's Patrick again. I would just have one follow-up here in terms of what consumers could be willing to spend for this. I'm thinking about the monthly sort of amount. And if I'm looking for subscriptions that people pay for, two things come to my mind. There is a mobile phone contract, which depending on where you live, is probably about EUR 100 a month or a little bit more, a little bit less. Alternatively, I could look at the Spotify subscription, which I guess, depending on where you are, it's somewhere between $15, $20 or euros a month. How much do you think people are willing pay for connected services in a car? Is that amount somewhere in between those 2 amounts I was referring to? Do you think they could pay even more for that? And what's driving your thinking behind that? And the other question I have is in terms of the pricing power you will have. Most of that stuff is software-enabled. So basically, all the cost comes upfront and the incremental costs, putting it into the car for BMW is zero. And where you have platforms that are highly concentrated, be it Spotify or Apple Music, there is pricing power. We have a lot of OEMs out there and some might be not so disciplined in terms of pricing their digital services. How can you avoid ending up in a situation in which basically the monetization opportunity of all you develop is diminished by very aggressive competitors who are just trying to grab share?

Well, let's just put it this way. I'm convinced the marketplace comes up with a price value feeling, I'd say. So therefore, you have to be transparent in what you offer for the monthly subscription. And people will need to say, yes, okay, that's fair, I'm willing to pay that amount. And again, what will be or what should be the content? I mean we talked earlier about the digital services and entertainment, et cetera. But I don't see that people will pay a second time for Spotify account because they already have one. Our -- the importance is that we have a strong integration and a good user experience for that Spotify. And that is always up to date, that it's -- when there is a Spotify update, that it will still work perfectly in your vehicle. And additionally, we have a lot of ideas of what is specific content that only BMW can offer and create new aesthetic experiences in the vehicle. Let's talk about, for example, light experiences. Let's talk about sounds or art mode, et cetera. That's nothing you can buy from another third party. So -- and the mix out of this integration and access to a third-party content and updating the pure BMW content, that makes the content of a subscription. And where the price point exactly will be, that's also good in these days, you can test it. You can work with smaller markets to begin with and then scale afterwards. The clear logic of all the digital business model is scale it. If you make it too expensive and it stays too exclusive, that's -- and you mentioned the mechanisms of the software generation and the monetization, you have to go broad. So if you start with a subscription model, you will definitely aim for the majority of customers. And in function of this, we'll define the price.

[Operator Instructions] And the next question is from Stephen Reitman from Societe Generale.

Again, a question about what people are prepared to pay. And as an owner of a 2017 BMW, I had a -- when I got the car, it had a 3-year subscription to Apple CarPlay. So I was thinking, we had a big subscription. When it comes to an end, it's a bit quite expensive to renew it. But then BMW decided to actually just to abolish the subscription and to just give it as a -- continue it as a free service. So could you -- was that -- what was the thinking behind that? I mean was that already in reaction to customer opposition to paying for what was a very useful service already? Or how does that inform your thinking about what you can charge with future services?

Well, that was actually a good example, a learning example for us on what works with the subscription or with a renewal and what doesn't. And today, we are not just giving it away, the Apple CarPlay or the Android Auto, but we have it included, but as a life cycle service. Because here, we did have indeed a customer feedback to say, well, I have paid for my Apple phone and service and I paid for the BMW. And I've once paid that they both go together, why should I pay here for, again? There is nothing new. And therefore, indeed, we switched the pricing model here. And the learning is, well, there has to be a continuous development of keeping something fresh or adding something, but the customer has to really understand it. That's also my argument always when people say, well, are you overpricing? In the marketplace, there is no long-term overpricing because customers need to accept and buy it to make it a success, and we will act accordingly.

And you mentioned also some potential uses of sort of like a subscription model. You mentioned, for example, head-up display. Now that's clearly a hardware issue there. So do you envisage actually bundling such features in the car and then just verify activation or is it still going to be more of a you choose it when you first order the car?

Yes. Then I was maybe unclear. The head-up display, I mentioned, as part of the digital area in the vehicle. So -- and that's also part of the EUR 5 billion, but that is nothing that is suitable for a function on demand. Because here, you would put the camera into the vehicle, and this -- I would doubt that it makes sense to have it there -- visibly there, and then switch it on and off. Yes, that's probably not a good example for a function on demand.

The next question is from Tim Rokossa, Deutsche Bank.

It's Tim from Deutsche Bank. Obviously, you are aware that the main number that people will take away from this -- to our conversation will be the EUR 5 billion. So I don't think you want us to leave thinking that you want to generate EUR 5 billion with radios and navigation systems and maybe head-up displays. I think it would be really quite crucial for you to at least put some flesh behind this and say, is it proper digital revenues in the sense of really stuff that you make a lot of margin on? And we both -- we all know that margins on parts became quite good, but what people obviously get excited about is the 70%, 80%, 90% plus margin that you make on a software product, so even more than that once you have it fully rolled out. Is that going to be a sizable number of that EUR 5 billion? Or are we talking about maybe a few hundred millions or so to begin with? And then as a second question, when we talk about the digital sales channel, it's obviously probably fair to say that, that is more profitable than selling over a stationary dealer. When you already sell so many cars online today, can you give us maybe a feeling for how profitability would look like for your digital versus stationary? Obviously, you don't have to put too many numbers behind it, but just in comparison.

Well, to the first part, once again, in our definition, digital is not only software. Digital has hardware components as well. And in the navigation system, there is computing power and their software, and we sell both combined as a product to our customers, and the same driver assistance systems. And actually, those are a pretty important part of that revenue figure. I mean just look at current pricing list. We are -- for the top system Driving Assistant Professional, depending on the vehicle because the capabilities are still a bit different, but we are easily talking about EUR 3,000 to EUR 4,000 per vehicle, and this with a pretty good and continuously increasing take rate. So I do want to tell you that the EUR 5 billion are digital features, functions, but not only software. So that is, I think, important to really avoid misunderstanding.

Anything more on the digital sales channel?

Digital sales channel, obviously, if you can take complexity, structures, workforce out of the sales system that is reducing costs. But don't forget, we have a global footprint, and we also believe the look and feel and touch and feel of vehicles is, for a long time, playing an important role. And that's also what makes a difference for BMW and a lot of customers will just test it, experience it driving or at least sitting inside. So I'm not seeing a complete reduction of the sales force that we have physically in the world. But of course, further digitalization is creating efficiencies. And that's also something we work on together with the dealer body because it's a joint operation.

Yes. Thank you, Peter.

I think my first question might have not come across properly. I think that's exactly the problem, right? So people want to take away that you make EUR 5 billion with digital services. But actually, what you just told us is you make EUR 5 billion, of which a good chunk will just be hardware. Do you want people to leave this event thinking that the EUR 5 billion is actually just something you generate with navigation systems and radios? Or are you happy to tell us that within the EUR 5 billion, there is also a decent chunk of what the capital market would understand as digital revenues in the sense of subscription models and on-demand services?

It sure is a decent chunk, yes.

So feasible. It's not just EUR 100 million or so?

It's not just EUR 100 million.

Okay. Good.

I mean, again, a figure that I mentioned already earlier, we are already in a good 2-digit million today on pure store revenues and that's a lot to come in terms of content and process.

Since we only have a few minutes left, and I have one last question that I would like to ask you. We talked a lot about the functionalities. Do you see any regional differences in terms of that? I mean, we have -- you mentioned the global footprint and our customers do have different preferences, different tastes. So in terms of our digital offer, do we make regional differentiation here?

What we do see in the industry, and I think it was a point in the discussion with Christoph earlier as well, China is there at the forefront. And the Chinese customers are very experienced and very demanding in terms of digital experience. So they also are more open for and also more demanding for new content. But I don't think it's extreme fundamental difference, but it's a matter of timing. I'm sure that a lot of this advanced digital orientation that we have in China already today will be seen in the rest of the world, and we see already a part of it, of course. So everything that we develop for China, we also look into how can we make use in the rest of the world. Of course, if you look at specific content providers, Tencent I think was mentioned earlier or the other specific Chinese entertainment features, they, of course, remain local and we have to increase further -- and we're already doing pretty well -- our capabilities of really addressing those regional market requirements and being able to link with the local players. And that's not only in China, even though there, we have the biggest development team also, but also in other parts of the world. We are increasing our local links to the relevant providers to our customers.

Great. So thank you very much. I think that was a great discussion. A lot of good questions, tough questions, but really good and lively discussions. So Peter, thank you very much for joining us here today. Thank you to all of you for joining us and asking those questions. I hope you've enjoyed our BMW Digital Day and the 2 deep dive discussions we had with Christoph Grote and Peter Henrich. So you'll find an on-demand version of the stream here of today's Digital Day soon on our BMW Investor Relations website. And I'm looking forward to seeing you either virtually or hopefully also in-person in our next event. Until then, take care and stay safe. Goodbye.