Nebius Group N.V. (NBIS) Earnings Call Transcript & Summary

Good afternoon, and welcome to the Deutsche Bank Auto Conference presentation from the Yandex self-driving car team. My name is Lloyd Walmsley. I'm the Internet analyst here at Deutsche Bank. And we're excited to have Artem Fokin, Preston Carey, and Katya Zhukova from the Yandex.Taxi business. As just a quick intro, Emmanuel and I had the pleasure of hosting this team at CES earlier this year and getting to ride in one of the self-driving cars from Yandex out in Las Vegas, and it was quite an impressive ride. So we're excited to have them present here. Unfortunately, no one is able to do rides, of course. But I want to thank the team for at least before COVID trying to get that in place. The team is going to go through some slides, and then we're going to take some questions from us and then from the audience. So with that, I'll turn it over to Artem from Yandex.

Hello, everyone. Thanks for being with us today. I'll just walk you through the short slide deck here just to give you an impression of who we are, where we are right now and what we are doing. And after that, as just Lloyd said, we will be happy to answer your questions. So if you just jump to Slide 2. It's a quick overview of our projects. So this self-driving project started in Yandex, actually quite late to the autonomous driving game. It happens very early in 2017. But we were happy to enjoy quite successful ride so far, mostly because we had the benefit to capitalize on the technologies, which were developed within Yandex Group since its foundation in 1917. Right now, Yandex is not only the biggest search engine in our part of the world, we also have very successful ride-hailing, food delivery and e-commerce businesses, which contributes tremendously to the value chain, which can benefit from autonomous rides in any vehicle possible. So as I said, 3 years slightly more into the game. We have a fleet of 120 cars, which spread across 3 countries, that's Russia, Israel and United States, Michigan. Plus we also planned to launch a small operation in South Korea that was slowed down because of COVID-19 situation, but we will resume it as soon as the limitations are off. So -- and in terms of technology progress, our fleet has driven over 3 million miles, mostly during the last year of our development when we began scaling our fleet. And that progress was possible because of, as I said, a lot of technologies from Yandex Group, we were able to capitalize on. Also, Yandex is a very attractive employer in Russia. So every software developer considers a position with Yandex being a dream job. That's why it was possible for us to attract a lot of talented engineers to work on our technology. Now if we just jump into the Slide #3. Here, you can see it. We just introduced the generation 4 of our self-driving vehicle. What's important about that is that we build it not alone, we did it together with a partner from automotive industry. We have an ongoing technological partnership with Hyundai Mobis. That's the part of Hyundai Motor Group. And the help of Hyundai engineers, we managed to successfully integrate our system into new Hyundai Sonata 2020. What's also important, the progress we achieved so far, we managed to do it at a fraction of the cost, if you compare it to other self-driving companies. So as of end of Q1 this year, our total spend on that technology was just $45 million which is, by orders of magnitude, lower than other companies working in this space which have similar results were spending. So what's also important, we're not just driven 3.4 million miles on actual roads. So it's not the simulated miles. That's actually miles which were driven in very challenging environments such as, for example, Moscow City Center, which not only enjoys a very dense traffic, but also we had to cope with the local weather conditions, which have very harsh winters and very distinctive seasons during fall and spring. Also, that technology has been tested, as I mentioned, in Israel, with a different traffic situation, different products such as 2-wheel micromobility vehicles and so on. And also we continued our testing in other locations such as we demoed our technology, as Lloyd mentioned earlier, in Las Vegas in January this year during CES where we were giving rides to many people who are interested in our technology. What's also important, these rides were done in full autonomous mode. There were no safety driver behind the wheel. So the car navigated Las Vegas traffic taking right turns, left turns and protected left turns and so on without anybody sitting behind the wheel. That gives you a different perspective if you are in the kind of computer drives itself. Also important thing to mention here, I think, is that we actually have been working robotaxi service since August 2018, that's a service which we are running in the town of Innopolis in Central Russia, where citizens of this town can hail a robotaxi through standard Yandex.Taxi app. The car will pick you up and drop you off within the city limits. And again, there will be no one behind the wheel. So it gives you or give the rider the complete autonomous driving experience. What's also great about technology we've been developing so far is that it's scalable. So we managed to adjust it with very minor modification to smaller form factor, and I'm talking about the picture you can see in Slide #4 right now. So we take our technology stack, we adjusted it to the smaller form factor and created a device we call Yandex.Rover delivery bot, which is very welcomed by companies who have to utilize delivery people for the last mile logistics. And our Yandex.Rover does exactly that. It navigates pedestrian areas. It navigates sidewalk and bringing packages or groceries or food from various sellers to the end customers. We actually started commercially deploying those rovers in May this year. And we plan to scale that business big time. We have a huge interest from other parts of Yandex Group, and we have a lot of interest from outside companies exactly for that. And now I will probably jump to Slide #5. We understand that software is the major part of self-driving technology. But software need to work with various hardware elements, such as sensors, and some of those sensors are very critical for our success, such as LIDARs, for example, which are not only the most important sensor for the vehicle to navigate itself, but also it's quite expensive. If you look at the total hardware cost we need to install on the car to make it a self-driving car, LIDARs account for approximately 40% of that cost. So by developing our own LIDARs and also cameras, we not only can drive the cost of the hardware we need to install or cut down, but we also can get more information from the sensors if we develop them ourselves, than from the sensors that we have to source from the open market. And Slide #6. Why are we doing this? And the economy or monetization principles behind our technology are very simple. If you try to compare the costs for self-driving car versus the regular car, to compare apples-to-apples, we actually need to make sure that those cars have similar features. And while self-driving car can drive itself, for regular car to be operated at least 16 or 18 hours per day, you need at least 2 drivers. So if you are to compare the costs you need to spend on the human-driven car to operate it within the fleet versus the cost you need to spend self-driving cars, you need to account for the drivers salaries. And just very simple ballpark calculation gives you an understanding that the costs you have to spend on the human-driven car and drivers is significantly higher versus the cost you need to spend on a car and a hardware you require to modify it to make it a self-driving car. So to summarize what I just said. We have been developing technology for not that long, if you compare it to other players, but we managed to achieve strong results. We believe that based on the information we can get from public sources, that our technology is in the top level if you compare it to other companies, because we -- as far as we know, like shown, the second highest number of autonomous miles driven so far. We have all the major pieces of value chain, which is required to bring the technology to commercial deployment in-house. We spent a fraction of the cost to get here if you compare it to other players. And we have a capability -- proven capability to scale that technology to smaller form factors or to a larger form factors, such as trucking or public transportation. That's probably it from my side. And now we will be happy to take questions.

Okay. Great. Well, I will go ahead and lead with some questions to begin with, but we also have the capability for the audience to queue up to ask questions as well. So please go ahead and do that through the system, and I can ask on your behalf. The first question I was going to ask was, how have you guys been doing your development work during the pandemic?

Well, fortunately enough, we didn't have to hold road testing during the pandemics because those restrictions were not applied in our major testing ground, which is Moscow. So our road tests continued, and which is great because road tests are essential for technology development. Our drivers were supervising that, taking all the safety with masks, sanitation and so on. But yes, so it was fine. The biggest challenge we got from COVID-19 is that Detroit Auto Show was canceled. So we don't have a chance to demonstrate new technology right now as we initially planned. That's why we have to do just presentation, but we would be happy to demonstrate that live as soon as it's technically possible. And the second, we have to delay our -- delay the start of our South Korean operations.

Okay. Okay. Great. And then, I guess, yes, you mentioned you're continuing to test on the road, which is essential. What are some of the differences in your mind between simulation and public roads?

Well, that's a very good question. So simulation is important part of development. But with the simulation, you can test only certain bits of technology because no simulator is accurate enough to represent real world. So we in preference showed you with simulation, it's basically one software testing another software. And the task to create totally accurate representation of the real world is probably even more complicated than a task of creating self-driving car. That's why simulation is important for certain tasks, but to be able to understand how your -- the level of technology or how good you are, you actually need to do continuous testing on public roads, discovering more and more edge cases, fixing those cases, implementing changes, retesting it again and doing it over and over and over.

Okay. Okay. Can you give us a sense of how your technology is differentiated from others such as Waymo and GM Cruise?

Well, it's a good question, again, Lloyd. Thank you. But the challenge here is that you -- it's very hard to answer that because like most companies who are developing autonomous driving technology are quite secretive about what they have. And the only actual possibility to compare one technology to the other is like for ride in one car then to ride in the car provided by the other developer and compare like your impression from each ride. But that said, you can also look at some other things, for example, notable milestones and by notable milestones, I mean, does the company have an actual service deployed somewhere? Is those service being run with a safety driver behind the wheel? Or it's a driverless operation? How many autonomous miles have been driven by that company on the roads? Because that's important indicators because if you are capable to drive many miles, that means your technology is good enough to drive that mile. Because you can invest billions and billions of dollars but if your technology isn't good enough, you just won't be able to drive, your cars would be stuck and so on. So these are the numbers you can look at. And by the number of miles driven, as far as we know, we are the second -- like we've driven the second most miles on the roads in the world. We also have deployed a public -- service for the public. That's what I already mentioned. And that's it. And by looking at all those milestones, I'd put us in the top 3 bracket among all those companies who are developing such technology. But again, it's very hard to measure because there isn't a single widely recognized benchmark that you can apply to one technology or the other and say that this one is better by the other by certain number of percent.

Yes. Okay. And then your self-driving car business sits within Yandex.Taxi, which is partially owned by Uber given the transaction you guys did to kind of merge your entities over in Russia. So the question is, is there an opportunity to increase your ownership stake within the self-driving car unit, perhaps through buying out Uber since they seem to be focused on their own path to profitability in their own core businesses? And is that something that might be interesting to you guys?

Well, it's a difficult question. I'm not sure I'm in a position to answer that right now. Obviously, we are looking at the options to make sure that all branches within Yandex Group can benefit from that technology. We are also open to provide -- give that technology not only to Yandex entities, but to the other companies. So for example, if we are -- not if, but when we are to deploy in a country where Yandex.Taxi is not present, we would be happy to connect our self-driving cars fleets to other demand generation platform. But to answer about like increasing stake, I'm not the right person to answer this question.

Yes. Okay. Fair enough. And I guess is there an opportunity that you could provide your vehicle to someone like Uber in the U.S.?

I'm sorry, I missed the last part. Can you please say it again?

Yes. I was just wondering, is there an opportunity for you to provide your car -- you kind of alluded to providing your vehicles in other markets where Yandex.Taxi isn't present. Is there -- would that include potentially providing your vehicle to Uber in the U.S.?

Well, as I said, our fleet would benefit from any demand generation network in a country Yandex is not actually present. So we are hoping to connect to other demand generation ride-hailing platforms such as Uber or Lyft maybe or other platforms in other countries. That's why we develop in technology and scalability in mind. That's why we're testing it not only in Russia, but we are testing it in Israel, in the U.S. and potentially other countries coming soon as well because we want to make sure that our technology can work everywhere in the world. And if we're developing that technology with idea to scale out, that means that we cannot stick only to the possibility to integrate to Yandex.Taxi, of course, we need to be able to connect our fleet to other company, to other platforms.

Yes, yes. Okay. Okay. And how quickly has regulation been developing in different parts of the globe? And when do you think we can see self-driving technology at a mass scale?

Well, regulation is a tricky part. So obviously, what we can see is that many countries taking step into developing certain self-driving regulations. We're happy to work with the authorities in Israel, helping them to define how the regulations need to be formulated to make sure that the companies like ours can deploy fleets in Israel. We're also working with the authorities in Russia to make sure that commercial operations of self-driving fleets happen sooner than later. We're working with the respective ministries of transport, internal affairs, trade and economic development to provide such road map. And so far, we're quite happy with the fact that our opinion is being heard.

Yes. Okay, okay. And I guess do you feel like there needs to be significant upgrades in terms of either road infrastructure or 5G implementation to scale deployment?

Well, actually, we don't think so. Because when we are developing our robo driver, we are developing robo driver for current infrastructure and for current roads, wherever we go. So things like 5G are definitely not essential for deployment of self-driving cars. And with the old infrastructure we have right now is good enough for that. Obviously, when infrastructure develops, we would be glad to benefit from that. But again, technology works with what we have. And talking about infrastructure, when I hear when somebody is saying like, "Yes, we need a special infrastructure for self-driving car." it's very -- it's a very strange thought because if you look, for example, at aviation, the infrastructure for airplanes was developed after the flights started operation -- operating commercially and the understanding how that infrastructure should look was formed. The same should happen with self-driving cars. So first of all, we need to start operating the fleets. Then we need to have like majority -- not majority, but significant number of cars -- of self-driving cars out there on the roads, and then by understanding how they operate and what infrastructure changes might make them more efficient, those changes should happen, not vice versa.

Yes. Okay. Okay. And then you mentioned in your presentation, you guys are developing your own LIDARs and cameras. I guess how much of an advantage do you think that will be for you all? And then as we think about the potential monetization of that, would you ever sell them to third parties?

Well, we're developing -- right now, we're developing them for our own use. Because we will benefit from the cost of those sensors if we develop them ourselves. And second, we're developing with our specific application in mind. That means, we can implement some of the features we cannot get from the sensors which are available on the market. So yes, so far, the plan is to develop them for our own use, but that might change going forward in the future. So we're not saying no to possibility to sell it to different parties sometime later on. But so far, we're focusing on our own needs.

And I get -- I think you pointed out in your deck that 40% of the cost can be these LIDARs of a self-driving car. Are you -- how much lower are you kind of bringing the cost down, if you're willing to say, versus what kind of market cost would be if you bought them from a third party?

Well, it's kind of hard to gauge because the changes of the hardware set we're installing in car is decreasing every year. So it dropped at least twice over the last 2 years. So we dropped 2x. So the client was expensive and it was 2 years ago. And the costs continued to decline. But yes, we do believe that we can cut the cost at least like by half by implementing on our own LIDARs from the current point. But again, the market goes down with economy of scale and more players come into the market with their own sensors. Overall, the market cost for those sensors will be lower than that. And obviously, we understand that if we had to use our own LIDARs, they -- not only they need to be very efficient with our technology, but they also need to be very affordable for us, so it could make sense to install them in our cars.

Okay. Okay. And then, I guess, how do you guys think about where to partner with automakers versus doing other -- doing everything on your own more broadly?

Well, we're actually very happy with our ongoing technological partnership with Hyundai Mobis. We understand that they have expertise that we don't have. For example, like hundreds of years of automotive experience with an industry is very important. But on the other hand, we have something that most automotive guys don't have, we have an experience building complicated algorithms and artificial intelligence-based systems. So -- and the tremendous results can come if you combine those 2 experiences together. Yes, we are happy to cooperate with Hyundai Mobis. We're open to cooperation with other automotive partners because, again, our product is robo driver, and robo driver needs a car to drive. The car which is adopted to robo driver needs, because the most cars -- actually, all the cars you can buy in the market today are adapted for human driver needs. And autonomous driving-ready vehicles is slightly different. That's why we're talking to automakers to make sure that the features, which robo drivers would require, are taken into consideration in the future plans of those automakers.

Okay. And then, I guess, do you all foresee that your fleet of cars getting to tens of thousands of cars? How do you kind of see that evolving?

Well, obviously, right now, with our fleet of 120 cars, 200 by the end of this year. And like the vehicles we had in [ this year ] are going to be Hyundai Sonata. A lot of work should be done on the car after we bought it from the dealerships, right? So we need to do a lot of retrofitting by ourselves or in case with Sonata, together with Hyundai Motors engineers. We are -- we would like -- and we actually foresee that moving forward, a lot of changes which are required on which we are doing right now will be done by the factory, by the automakers, specifically for the companies which operate self-driving fleets. So moving forward, more and more automakers as we see now, would be interested in building special purpose cars targeted at self-driving fleets.

Okay. And I guess when we think about the monetization of the technology, how do you guys see this playing out? And is it kind of a licensing fee model where a ride share network may pay for all of the capital cost for the car and pay you a licensing fee? Would you guys be providing the car? How do you guys see the ultimate monetization of this technology?

Well, during initial deployment and initial scale up, obviously, we see that the fleet -- it will be our fleet, so we will be building our own fleet of self-driving cars and connecting them to third-party networks. As technology matures, as the acceptance of technology becomes broad and the safety and all the safety of technology will be proven within our own fleet and the customer satisfaction rate also really proven by our own fleet, we can think about licensing if that conversation occurs. But during initial scale out and scale up, we believe that the fleet will be ours.

Okay. And then how do you guys -- you mentioned kind of being delayed in Korea. What are the plans, I guess, once COVID issues lift? What are your plans this year in terms of new country expansion?

Well, just as we built a fleet in Israel, just as we built fleet for the Detroit Show in Michigan, we also built a small fleet in South Korea and start operating there. First, for testing and then for deployment at small scale and then scaling out to larger numbers of cars.

And I guess can you maybe talk about how much needs to be kind of -- how much customization needs to be done for your robo driver as you enter a new country? How long does it take to kind of adjust it if whether that's different road conditions or different driver behavior? Is it starting from scratch? Is it not take long? What could you give us about that?

Yes. That's a very good question, friend. Thanks for asking, Lloyd. Remember that demo you had in Las Vegas?

Yes. It was a great demo. It's a very impressive.

It was -- well, thank you. Thank you. So it took us about a month to prepare just 2 cars. So we didn't do anything in Las Vegas until December 2019. And by January, we were able to drive autonomously. So that's it. And that was just 2 cars. With more cars, that deployment time could be shortened because you drive more miles, you get better understanding and get update about drivers' behaviors, you catch more -- slightly different cases versus you used to experience, and that's it.

Yes. Okay. Well, look, I think we are about out of time here, but I wanted to thank the team for presenting and wish you guys the best of luck in your development and hope to see you all soon live in person and hopefully, in a venue where we can test some of your cars. But thank you for being here.

Sure. Thank you very much, Lloyd. A pleasure speaking here. I would be happy to give you a ride in our new gen 4 Hyundai Sonata. Yes, looking forward to that. Thank you.

All right. Thank you, everybody. Thanks for listening, and have a great rest of your conference.

Thanks, Lloyd.

All righty. Bye.

Thank you. Bye-bye.