Latest / Autonomy Insiders / European Autonomous Driving: Inside Niulinx and the Robosharing Model

Transcript

European Autonomous Driving: Inside Niulinx and the Robosharing Model · Autonomy Insiders

• Daniel Abreu Marques: Welcome to Autonomy Insiders. I'm your host, Daniel. And the US has Waymo and Tesla, China has Baidu Pony AI and WeRide and the EU until now has had no continental champion in autonomous driving. And my guest today wants to change that. Luca Foresti is the CEO and founder of Niulinx a spin-off from the Politecnico di Milano that just raised 38 million euros, the largest autonomous driving round Italy to date. Today with Luca Foresti. luca foresti: So So. Daniel Abreu Marques: Welcome to Autonomy Insiders, the show where global industry leaders in autonomous driving unpack their real-world insights. And before this, Luca spent more than a decade building a healthcare company that served millions of Italians. He trained in physics and financial mathematics, and had he has called Niulinx as the hardest challenge of his career. So what this conversation really interesting is the model. ⁓ Niulinx not chasing a copy of the American robotaxi, So they have a different idea and a different word for it. ⁓ it's robo sharing. And we will get into exactly what that means today. So Welcome Luca. I'm so happy to have you on the show today. luca foresti: Thank you, Daniel, for inviting me. Daniel Abreu Marques: So let us start with where this came from, because it did not start as a company. So Niulinx came out of the AIDA research group ⁓ at the Politecnico di Milano under Professor Savaresi. And for people who don't know the backstory, what was the research that made think there was a company here, not just a lab project? luca foresti: The basic fact that when you work on autonomous driving, when you to go from a research project to something that goes in the streets, servicing customers, you move amount of money, an amount of people and complexities that a research group in a university cannot manage. So the only way for going forward, for pushing forward the project and the idea was to build up a startup, a spin-off. we are the biggest spin-off done in Italy so far in terms of number of people. The day after the closing of the investment, I signed the first 40 contracts. Now we are at 50 engineers. And we are aiming at 100 engineers in 12 months, 200 engineers in 24 months. So trying to scale up as fast as possible talent in quantity and quality that can allow to work on this business that as you know perfectly, it's a very complex stuff. You need a lot of good... ⁓ minds working on it and yeah this was the only way to do it. Daniel Abreu Marques: Yeah, that's super interesting. And ⁓ as I mentioned in the intro, you spent more than thirteen years building Centro Medico Sant Agostino in healthcare. So ⁓ why walk from that to take on this autonomous driving challenge? So a field where the giants have spent billions less. luca foresti: Okay, I sold the Santa Agostino twice, once the first time to a private equity ⁓ then I remained as a CEO and we sold it a second time to the main Italian insurance. ⁓ then after a few months, I'm an entrepreneur, I'm not a manager. And understood I couldn't work in a big company because, know, culturally, ⁓ there is misfit. between me and big companies like that. And therefore in February 2024, I decided to exit from Sant'Agostino and then I did one year traveling around the world, going to China, India, United States and several other countries, trying to understand what was going on in the world, particularly in the business world, but not only in the business world. Then I came back and I built up from scratch a startup studio doing a couple of investments in AI based company, one healthcare and one in physical AI building, let's say, twins of physical stuff, ⁓ rooms or or stuff that. And then basically by chance, met the Savaresi And I was following, since many years, the autonomous driving industry, trying to study and understand what was going on, but I never worked on that. And I couldn't really develop a company from scratch on my own. But I met Professor Savaresi and I had possibility to understand all the work they did and the level where they arrived and all the challenges. ⁓ technical, economical and so on and the fact that I would enjoy it. therefore we decided ⁓ have me as a CEO and main shareholder of the company. And we worked on building up the deck, all stuff you need for asking money to investors. And then we spent ⁓ several months ⁓ finding out if market was ready to help us to build such. venue ⁓ succeeded. This is the story Daniel Abreu Marques: you mentioned that you have hired quite a lot of and ⁓ what I also read in an article is that you have a team of twelve co founders and ⁓ your average age is ⁓ something like twenty seven, So also quite young, ⁓ can you ⁓ talk a bit about how that came through? So like the twelve co founders? luca foresti: Yeah, very simple. I arrived in a team that since several years was building up all the technology. So if you want to build up a culture in which, let's say, engineers feel attached to the project, it's a good idea to have them ⁓ you ⁓ co-founders. ⁓ Of course, there are small co-founders and we gave to each ⁓ stock options with a certain formula, certain ⁓ in which now here everyone ⁓ participating to the, ⁓ of the project. This is typical in Silicon Valley, as you maybe know. It's not so typical in Europe. ⁓ but I think it's a good choice in something like that that is a cutting edge in the technology. So you need to attract ⁓ people, super fantastic engineers. And was one way Daniel Abreu Marques: So maybe now let's it ⁓ a bit more concrete. So in in one in one plain sentence, what does Niulix sell and to whom? luca foresti: At present, Niulinx is working for going as fast as possible to homologation. At present, in Europe, homologated L4. And so it's an open issue. ⁓ How fast you go in that direction? And for which kind of service are you So our ⁓ aim, goal, I'll to explain why it makes sense, Niulinx is working for going as fast as possible to homologation. At present, in Europe, nobody homologated So our ⁓ aim, our goal, is to homologate a service that we called Sharing. is to homologate a service that we called Robo Sharing. In very few sentences, the idea is you have an app you call the car to you, the car arrives autonomously at a maximum speed of 30 km per hour, stops in front of you, you jump on board at the driving seat and you drive the car at the speed that you like, that you desire. You arrive where you want to go, and without parking it, you get out of the car and the car goes back in the fleet for another customer. This is the type of service we are building up and we would like to deliver. Why ⁓ approach sense in Europe? There are several reasons and are not so clear at the beginning and many people don't really grasp the idea behind it. So first point, ⁓ Risk. If you go at 30 km per hour maximum speed, the car can be stopped in 3 meters. And the possible accidents tend to be much less ⁓ for the people. Second reason. If you at 30 km per hour and you have a false positive, it's pretty easy to stop the car as fast as possible. in the right side of the streets to have people remotely taking care of what the car should do from that moment on. If go to 50, 60, 70 kilometers per hour, it's much more difficult to do that with a frequency. Third reason, in this way, you don't need ⁓ a license because not a taxi. And in a country like Italy, but Europe is very similar, the European countries, in a country in which you have 25,000 licenses for taxis and 45 millions private cars, our aim is to talk to the 45 millions and not to the 25,000. And for doing that, you need to not have a constraints in numbers because of taxi licenses. in a country in which you have 25,000 licenses for taxis 45 millions private cars, our aim is to talk to the 45 millions and not to the 25,000. So all these reasons make what we are doing sensible choice. And other is money. If you want to have ⁓ a robotaxi, complete robotaxi, you need billions in investments because you need to tackle the edge cases. In the robot sharing ⁓ We can keep a certain number of edge cases false positive because we manage it with people. Delivering the service because there are no people on board in that moment. So this makes sense because you exchange billions in capex with comma something in opex. That's the idea. Daniel Abreu Marques: Makes sense. And do you also then envision this as a first step? So first start with robo sharing and later on develop into something more like real ⁓ robo taxi like service or do you want to stay with that? luca foresti: Correct. Correct. The goal, the final goal, because of course, is a service with higher quality perceived and real quality, is the robotaxi approach. But it will take time, it will take many kilometres, days ⁓ the streets. And it a huge difference in terms also of financing to do that. delivering the service that is paid developing in ⁓ &D mode like for example Waymo did since 2004. But they had tens billions of investments available. At in if you go to whatever investor you may imagine and you say, give me one billion. developing such a technology is fairly difficult to obtain it. So you have to be pragmatic, a very pragmatic approach adapted to what you have here. Daniel Abreu Marques: Makes sense. So basically you can already earn some money while further developing the software and everything and then at the end have ⁓ also a robotaxi service. Makes sense. luca foresti: Correct. And also let me add one point. I believe talent density. So if you place together under the same roof, 100, 200, engineers, vertically ⁓ able develop in autonomous driving, ⁓ you obtain something not necessarily push you in the direction of robotaxi. but allow you to deliver many different use cases with autonomous driving because you have the right people together working with the right technology. Daniel Abreu Marques: what do you in-house buy? So probably the software is then from you, but also like remote fleet management and something like this. So so what's make versus buy? ⁓ luca foresti: we buy hardware, so we buy sensors, buy actuators and we buy computing power on board. We may buy, it depends ⁓ on price, ⁓ the ⁓ 3D At present in the market the 3D mapping is very expensive, so we are also thinking about doing by ourselves the 3D mapping. All the software stack is in-house produced, perception, localization, the decision making where to go, the remote supervisor, software stack internally developed. Yeah, the technology is internally developed. We don't develop the car, so we get ⁓ a one approved car and on top of it we place our hardware, the hardware that we buy and we place on it, like Waymo approach, let's say. It's very similar to the Waymo approach in terms of technology. Daniel Abreu Marques: Yeah. And and what is the vehicle you're retrofitting then? So is a specific brand? luca foresti: At present we are working with 500 electronic from Stellantis and we have an agreement with them for working on it. We are also watching very very very carefully an L7 ⁓ Reason is, the L7 is a vehicle for two people at a maximum speed of 80 km per hour, electrical ones. And it makes sense to watch in Europe to such a vehicle because the average number of people in the cars here is 1.3. And Waymo as well published data about 90 % of the trips with one or two people on board. So if you think about that and you do all the math and you have in mind, and I have it very clearly, that in the long run the price per kilometer will make the difference in this game, then it doesn't make too much sense to have big cars, very expensive ones, five people, seven people cars. It's much better to start with a two people car and to have them going through European cities that have small roads, a lot of... ⁓ ⁓ regulations in terms of accessibility in the city centers of cars and so on. So that's the idea. So Stellantis is one platform we are working on at present. We are watching and working and analyzing L7 cars. We may have other kind of platforms in the future, but here the game at present is to baptize ⁓ one platform to get homologated on that platform and then it's much easier to work on other platforms because the software stack is available. The hardware stack specifically for L4, you already did it for one platform. It's not so difficult. It has some difficulties, but it's not so difficult to adapt it in other platforms. So that's the idea. no, the business model we have in mind is a kind of franchising business model. So we develop the cars with all the hardware and software stack, and we are the franchisor, and we give the cars to local ⁓ operators that ⁓ buy the cars and operate it. ⁓ deliver the car, we deliver the app for managing the car around, ⁓ we deliver the central remote station for all the franchisees. It doesn't make sense to have one remote station for each single franchisee. ⁓ so this is what we provide them. So our business model is to sell the cars. ⁓ that we created and to sell a service for managing certain services of the cars for the franchisee. Daniel Abreu Marques: Okay, makes sense. Also staying asset light. you also described Niulinx a regulatory native. So what does that mean? luca foresti: It means this idea, if you think about Tesla or think about Waymo and so on, they worked a lot on the technology. So they started with this question, which is the technology that most likely can work for doing autonomous driving? And they worked for years on that. When they have a technology that in their opinion is working good enough, they go to the authorization authority and say okay I have this car I have this technology please authorize me. This is methodology that can work in the United States but is a methodology that may have a ton of problems in Europe. Why? Because basically Europe is interpreting the authorization for this ⁓ kind of technology a thing ⁓ near to the authorization of normal cars. In fact, if you think about it, the talk a lot about L2, L3, ⁓ L4, if they are prosecution, you something in which ⁓ you do small changes and you arrive there. It's not like that. It's a totally different technology. Basically, in my opinion, we should talk not about L2, L3, ⁓ L4, but we should talk about technologies in which you have the driver and technology in which you don't need the driver. That's the difference, the big difference. therefore, coming to the answer to your question, our idea is let's take ⁓ the regulations, all the rules, all the certification needed and so on. And let's build up the technology in a way that natively is homologable are able to homologate it without having problems in terms of that sensor doesn't have the particular certification you need. That piece of doesn't allow you to ⁓ go through certain that ⁓ will be required by German or ⁓ Ministry of Transportation they provide you with the homologation. If you do that, of course you limit yourself a lot in terms of the possibilities, but these limitations allow you to go probably faster, to go directly to the homologation path. Maybe you have a technology that is not as good as the Waymo technology or the Tesla technology, but you arrive first in homologating, in delivering a service. And then you can get better and better and better when you are in the market. Daniel Abreu Marques: Makes sense and also one topic that always comes up ⁓ in terms of is ⁓ the ⁓ argumentation about okay, End End AI versus a modular stack. where does Niulinx sit in the spectrum A? luca foresti: I think you know it perfectly. In the long run, let's say in 20 years, the technology that will remain is the end-to-end AI. The problem is that nobody out there can show that the end-to-end AI will be ready without false negatives in a quantity that allows the homologation in one year, five years, ten years. Nobody really knows about it. And difference is this. ⁓ for this reason, we follow the more traditional approach at present because we want to arrive as fast as possible to homologation. But of course, in the perception part, we use a lot of AI and in other parts, we may use a lot of AI we can compare what we do with, for example, Alpamayo results from NVIDIA and to see the differences. There are other possible stacks we can compare. And we can also use other end-to-end AI as a software stack for comparing if you have a problem or not. No, you can have the checking of the results from your software. from the other software, when you see differences, you can decide, okay, let's stop the car and let's see what is going on. So it's not a monolithic approach in which you have or end-to-end AI or the more analytical approach is a very pragmatic approach. We know that in the long run, we as everyone out there should go to the end-to-end AI approach. But... most likely you will need some kind of a federation of data, some kind of work of several companies for having enough data in a way that you lower the number of false negative as much as possible. This is another problem. As far as I've understood so far, it's basically impossible to calculate the number of false positive of your system. Because if you don't have accidents in one million kilometers. How many false negatives you have? No, you don't know. It's in mathematics is a particular branch of mathematics when you don't have enough data and you have to forecast what's the probability. This is an issue. This is a problem. And this is an issue because as you know, if a person make an accident in the car, it's normal. If a system, if a software makes one single accident and you have, I don't know, one tenth of the accidents compared to the people is a huge problem. So we have to face it and we have to work towards, I don't say zero accidents, but let's say very, very few accidents. speaker-0: ⁓ speaker-1: me what is important for me is to say and to communicate that Europe has a lot talent and without companies that build up cool things, important technologies and so on, that talent will go elsewhere. ⁓ very important to have companies like the one we are building, but I hope many others will come, it's the only way to create something that keeps talent here and let that talent build up technology that is European and able to service ⁓ European people. This is a very specific technology because we are in front of a national critical infrastructure. So something where a car can kill a person in the street, can stop the traffic and can patrol with the centimetric precision our cities. And this makes a huge difference compared to other technologies. Therefore, each country, United States, China, and I hope Europe as well, will pay a lot of attention on who is the one having data and deciding what this software will do in our streets. So that's one key point. speaker-0: Yeah, absolutely. And we we're also seeing a lot of Chinese players, you know, WeRide Pony or Baidu pushing into Europe right now. ⁓ Really real competition here. So but do you think you have an then an edge on your home turf, so to speak? speaker-1: I'm a physicist. ⁓ believe in ⁓ what I see ⁓ I believe in hard work. I think that the edge in companies ⁓ typically not ⁓ the ⁓ you have ⁓ that moment, but in your ability to get better and better day after day for years. Then you earn an edge. in the market. That is a mix of technology, good you are with customers, how good you are in building nice stuff from a design viewpoint and so on. And all together that is the edge. It's very difficult in this industry to say that you have a particular edge because of a certain technology. It's a complex mix of technologies, many of which are, let's call it standard technology, mean, a LiDAR. is a fairly complex stuff but is today a standard technology. A video camera, it's a standard technology. An actuator is a standard technology. But try to put together cameras, actuators, computing, software and so on, then you don't have a standard stuff. You have something where you need a lot of money, a lot of people for bringing all together and to letting it work in a way that you don't have failures. So that's the idea. speaker-0: Yeah, absolutely makes sense. And in terms of geographical footprint, I think you will start in Italy based on your background. But what countries are you focused on? speaker-1: No, no, no, no, no, no, no, no, no, What is the important thing opinion? So we are a European company and the beginning we will open ⁓ Italy and in Germany but our aim is in the shortest time frame possible to have four to five testing sites in different European countries and as soon as we have the homologation to open up the entire Europe. So we are a European company and at the beginning we will open in Italy and in Germany the first two offices, but our aim is in the shortest time frame possible to have four to five testing sites in different European countries and as soon as we have the homologation to open up in the entire Europe. One of the problems when you compare Europe to the United States is that in the States this question, where do you work at the beginning in Silicon Valley doesn't make any sense. mean the answer is United States, always. In Europe this seems to be an incredible answer because France is different than Germany, is different than Italy. So one of the challenges for us is exactly to be able to go everywhere in Europe as fast as possible and to be a European company. speaker-0: And what is currently the regulation in Italy So can you already operate there speaker-1: the entire Europe for doing this business you have to go through three steps. The first is authorization for doing testing sites ⁓ and is a pretty easy step. It's something where ⁓ in different countries you have different regulations. This kind of authorization in Italy is likely different than in Germany or France, but It's you can do. It's not science fiction. Then the hard part ⁓ is the homologation. And the ⁓ you have to do it one country, but when you obtain it, you can go everywhere in Europe. It's one homologation for the entire Europe. It's not over because you need the national traffic law that allow it. And at present in Europe, there are only three countries that modified the national traffic law for allowing it, that are Germany, France and Croatia. The hope is that in the coming 12 to 24 months, I would not say all. of the European countries, but the main European countries will change the national traffic law allowing have ⁓ driving cars over there. So this is the situation in Europe. Of course, Italian law for the experimentation is slightly different than the Spaniards law or the German law and so speaker-0: And you mentioned that you're retrofitting cars currently. but do think long term you will also partner with a car maker? speaker-1: I mean, in the long run, autonomous driving cars are built for autonomous driving fleets. For many different reasons, it doesn't make sense to think that in 10 years time, 15 years time, the cars that are in the fleet are not built for that. Eventually, also without, all the stuff you need for driving them, as some of the first companies like Zoox or others are working on. The problem is in the short run. So at the beginning, when you have a startup working on autonomous driving, does it make any sense to build up from scratch a new car? How much does it cost? How difficult is it? How long does it take? What kind of and what of people you need for doing it? ⁓ It's the main bottleneck for your business. At the beginning is much better to retrofit it. ⁓ Or you have billions. Then if you have billions or even you have a company that is building cars like BYD or Tesla, then it's a different story. Then you have the expertise, the money and the possibility to start from scratch. Fine. I respect that way. ⁓ most likely if I would be CEO of one of that company, I would seriously think about that. But in the short run, it doesn't make any sense for us. As I told you, we have to be pragmatic, we have to be fast and we have to watch very carefully the bottlenecks and to leave all the nice to have things aside for a while. Then we back and we get better and better and better and at a certain point we may decide that the fourth generation, the fifth generation of our cars are built from scratch. ⁓ for autonomous driving fleets. speaker-0: Makes sense and and maybe this question I will ask now you cannot answer, but I will ask anyways. how much does a normal standard retrofit from for the Niulinx car cost approximately or no concrete numbers or ballpark? Because sometimes or often you hear like, Okay, retrofit is fast, but it's more expensive and so on and you also need the unit economics to work. speaker-1: very interesting question. Maybe I will not provide you with the answer you want, but let me discuss with you what's the problem in retrofitting. If you take a European carmaker and you try to retrofit the car, the reality is that the European carmaker doesn't own its own car. It's a puzzle different ⁓ of technologies coming from hundreds of companies around. And when you try to work on that, it's enough to talk to the Mercedes, BMW, Stellantis, but you should talk to the Bosch, to the ⁓ ZF so on. This makes very expensive long term and whatever. So in this sense, whoever tells you that retrofitting cars is expensive and ⁓ and super time expensive is true, it's true absolutely. But there are not only big OEMs, there are also smaller OEMs that own completely their car. When you jump in the L7 kind of car, the regulation is different, it's much simpler to work on that. And so you may think about retrofitting something where you place your hands into it till the last cable, till the last pieces of hardware. So again, you think about ⁓ as fast as possible, there are ways to do it in a way that you can retrofit cars. speaker-0: And maybe now let's talk about the thirty-eight million you raised. so thirty-eight million is large for Italy, but if you like you said, if you compare it like to Waymo's war chest of billions or maybe also Wayve's one billion, it's a bit smaller. So how far does this get speaker-1: Okay, first, if you take the history of Wayve let's take Wayve, okay? What was the pre-seed of Wayve? How much money? And how many people were working in Wayve in that moment? If I'm not wrong, they started up in 2017. So we are talking about nine years ago. In my opinion, it's not fair to compare pre-seed funding with 16 billion in Waymo that started in 2004, so 22 years ago. 38 million is quite some money as a pre-seed is nothing if you compare it to 16 billion in Waymo, but it allows us to show to the market, to the investors, to ourselves, that we are able to go very, very fast towards milestone that is much ⁓ to homologation ⁓ other players around. And when you are there and you go back to investors and tell ⁓ with millions, we did this part of the path. Do you believe that with another 60 million, 100 million, we can arrive much closer to homologation in a very short time frame? This is the game we are in. And I believe that is also a fair game because if you're not able to spend well 38 million, why an investor should give you 100 million? No, it makes sense. agree with this ⁓ ⁓ Of course, up to us to arrive to a certain point, up to investors to decide which appetite for risk they have. speaker-0: And you think about ⁓ like how you spend this, is it then mostly on all the engineering talent you're hiring right now or how is that capital allocated? speaker-1: The capital is allocated in engineers, it's allocated in cars, it's allocated in hardware, it's allocated in computing power, it's allocated in consulting for the homologation, ⁓ in countries like Germany where the homologation is managed by intermediaries like TÜV these are the main main part, but every time you open up a new testing site ⁓ need to place there a certain number of cars. You have to buy them you have to Put your technology on it. Then you need the drivers going around and making a ton of ⁓ Then you to download those data Then you need to analyze those data understand where the problems were get exactly the use cases that you want to get better and on the software side for getting better the day after and again and again and again. So this is a game and a business where the testing is not something that you do at the end of the game, it's something that you do every day from on till moment where we close the and hope that we'll never arrive, but we will be always under testing. I hope that at a certain point the testing will go together with servicing the customers out there. speaker-0: Makes sense. And would also be super interested in your in your read on why Europe has no autonomous driving champion yet. So is it then capital? Is it regulation? Is it ambition? So what's your take on that? speaker-1: Let me say nasty thing about that. In opinion, Europe is ⁓ of its success in car making. If you are the most successful place on earth in car making, making money, making volume, having 14 millions of workers on car making, It's very, very difficult for you to leapfrog the technology and to say, okay, now we invest for 10 years without profits for being the first in 10 years time in autonomous driving. And this is what happened in Europe for sure on the vehicles where China did it, they leapfrogged. They decided to be the first in the world in electronic vehicles and they won. Now Europe, in my opinion, it one possibility to leapfrog the electrical vehicle engineering stuff to say, okay, we lost that battle, but we may win the autonomous driving game. Particularly because autonomous driving will not be simply a winner take all economy. but there are tons of use cases for that kind of technology. therefore Europe is a different place compared to China, United States and other place on earth. And therefore Europe can develop its own use cases for going in its own cities and its way of life, its way of mobility. So the problem. So basically the thing that is happening in Europe is that OEMs don't want to develop their own technology on autonomous driving. So OEMs, big OEMs are the first asking for foreign technology. Give me Pony AI, give me WeRide, give me Waymo, give me Wayve and let's sell That is only business we able to do. build up nice cars and sell them. We don't want to invest billions and billions in a technology that will make some profit in 10 years. Too difficult. If I'm a CEO of a big OEM, I have to present in three months results, economic results. game in autonomous driving is about investing tons of money for many years. As Waymo knows, as Tesla knows, as Chinese guys knows. If Europe wanna do like that and have autonomous driving, has lost the battle from the very beginning. speaker-0: And one thing we're seeing in the US is that the US, for example, also restrict the Chinese virtual driver companies to come over there and Europe does not. Do you think should also kind of restrict the Chinese virtual driver companies from coming in so that ⁓ the European OEMs have to look at European alternatives as well? speaker-1: I mean, the answer in my opinion is pretty clear. If everyone else is building this technology, is restricting only to their own companies the access to the market, if you open up the market, it's fairly impossible that you develop that technology in a competitive way compared to a country with 1.4 billion people. or another country with 350 million people but with 65 thousand dollars per person as GDP. I mean, Europe has to, let's say, grow up, has to understand exactly what Europe should be. Do we want to be one of the many satellite countries? It's a possibility. Then we buy as much as you can. We as citizens are happy as consumers for a while using technology coming from abroad. We as producers start to have unemployment rate going up, start to have zero technology built here, start to have salaries going down because we buy ⁓ the high value technology from everywhere else. So that's the decision we should take as Europeans. I'm always for suffering in the short run for being happy in the long run. speaker-0: Yeah, absolutely makes sense. And one related question so kind of a prediction. So by twenty thirty, what do you think does autonomous mobility in Europe actually look like? And also where is Niulinx in this picture? speaker-1: You will have several test sites running with tens of thousands or hundreds of thousands of cars autonomously driving around. You will have two, three companies managing fleets or developing the technology for providers ⁓ ⁓ of fleets locally. will be in the ramp up. So we will be in a phase where of course the big majority of cars are driven by people, but you will have a small important amount of cars that autonomously will go around in Europe going up very fast. Let's say with growth of percentage points month. ⁓ or more than that. And this will go on year after year after year. Of course, you will have new cars, new technologies. The LIDARs will cost less less and less and less. The cost per kilometer will go down. At a certain point, you will touch the cost per kilometer of the cars ⁓ by the people. And then you will go below it. And you know. In this way, a certain point, more more Europeans will start thinking, I don't have to buy a second car or a third car. I will buy the first car because, you know, I want to be free, my family, I need space. I want to go on with everyone, I want to have my own car. Fine. But the second and the third car, I don't it. I have a fleet out there. I call them and they arrive. I don't have to park in the city centers. can read, study, talk to the people without paying attention to the streets. Fantastic. So this will be something that more and more people in Europe, as everywhere in the world, will start thinking about that. speaker-0: And maybe one more question because ⁓ just came up my mind. is this maybe exactly the point that the OEMs don't want to have this ⁓ technology on their own, or less because you said OEMs are selling cars. And if you think this technology will reduce car ownership and you don't have one car, two car, three cars, ⁓ so they have an incentive ⁓ to have it. Is it is maybe a point? speaker-1: This is a very interesting point because it's a misunderstanding. And let me go through the math why it's a misunderstanding. First, concept, the number of kilometers that you will have with autonomous driving fleet will go up around 30 % compared to the present one. Why? Because at present the car moves only when there is the driver on board. with autonomous driving fleets is happening is that the car moves also alone going from one customer to the other when you do the simulation the math you arrive at around 30 percent more ⁓ step in the concept how many kilometers a car will do before finishing his life maybe going on you will have more robust cars but more or less this number will not change too much therefore the number of cars sold most likely will go up not down but the number of cars that are going in the streets that ⁓ available will be much less so the turnover of cars ⁓ will be much higher. Instead of having one car surviving 20 years as of today, you will have one car surviving four years, five years, and then you have to change it and buy a new one. The total number of cars sold will be more or less the same, a little bit more. With a difference, you will not sell the cars to private people, you will sell the cars to business that are managing fleets. That's the point. So OEMs of course have to change their business, have to change the type of cars that they produce, but at the end you may have one OEM specialized in autonomous driving fleets and tons of cars. speaker-0: interesting. thank you much for your time, Luca. We're the end and I appreciate you took the time and it was very insightful conversation. thank you so much for coming on and I'm really looking forward to hearing more news from Niulinx what you're doing your way to type approval. So yeah, thanks again, Luca. speaker-1: Thank you, Daniel.