Name: Driving Toward the Next Generation of Mobility
Uploaded: 2025-10-28
Duration: 13 min 54 s

Transformative technologies like autonomous vehicles, electric air taxis, circular battery solutions, and AI-driven traffic systems are reshaping transportation worldwide. Global electric vehicle sales grew more than 25% last year, and billions of dollars are being invested in smart infrastructure and logistics. The shift to smarter, cleaner, and more efficient transport is accelerating. Leaders now face a pivotal question: How can they join, shape, and scale this transformation to unlock economic opportunity, cut emissions, and drive growth? 

Driving Toward the Next Generation of Mobility

Conferences

Tony, we're so glad you're here. Thank you for joining us. My pleasure. Um, I recently took my first ride in an autonomous vehicle. It was in San Francisco, and it was thrilling. I felt like I was on a very safe amusement park ride, but it sort of left me thinking, what is the problem that the autonomous vehicle is designed to solve? For me it was just kind of fun. I, when I need to drive somewhere, I drive myself, but there are some real social purposes and social goals behind autonomous vehicles. For, from your point of view, why does the world need AV? OK, so. Uh, that question is quite straightforward. They think about why do we need washing machines? Uh, why do we need programmable switches. Before the invention of program switches, we have lots of job opportunities for human operators to plug two phone lines together, but now we can find it can done by computer or machine. Think about our civilization. It's just lots of things in can be. done by a human being or can be done by the by the animals like horses, cattles, and gradually replaced by the machine power. AI is the same thing, you know, nobody wants to drive for 12 hours, nobody wants to get stuck in the traffic. So what if a machine do it for you and they will never have any mistakes. I think that's the dream of most of the people and that's why. I found it right. That's why we worked so hard for more than 8 years to make this thing a reality. So that's the analogy is that the productivity of the time for all of these people who would otherwise be driving that's freed up for whatever they need, whether it's keeping up with their email or inventing the next AI so. And now you've been thinking about mobility. You've been involved in AI and computer engineering for a long time. We write 8 years old, but you're, you've been thinking about mobility for much longer. Is that correct? Yes, yes, you know, I was, I was majored in at that time image processing slash computer vision in 1990s and then in 2000 computer vision. I got a PhD in computer vision and then taught at the University of Missouri Columbia for more than. 10 years, and then, you know, all my career has been devoted to the development of so-called AI because for a while, you know, after 1950 or 1960s, AI become a taboo or very cold name and people like to use computer vision and then artificial neural network and then proper models, all kinds of related terms. And then after 2010, 2011, AI. Gain her popularity but all my life has been devoted to R&D in this area AI, yeah, in the kinds of headline grabbing innovations lately that we've seen with large language models, uh, does that benefit the autonomous vehicle field as well? Does that same kind of acceleration make the cars operate better? Yeah, for sure, because, uh, you know, people talk about this kind of large language model, but inherently. This is um based on now we more like, you know, the researchers more like college foundation model. So basic large, very large neural network model it can be used to model language, but you can basically can use to model all the knowledge, uh representing all the knowledge. Therefore, you know, uh, we develop our autonom driving systems leverage on a very rapid progress in large language model and the machine learning. I see. And now, so we've explained quite a bit, um. We write is now in I believe 30 cities in 11 countries. Is that correct? Uh, one of the countries that's coming to next is Saudi Arabia, in fact, right here to Riyadh, um. What you, you know, at this point you have experience working with governments in many, many places. What's, what kind of regulatory framework do you need for an autonomous vehicle company to thrive, you know, what kind of cooperation or uh or rule making. I think we really need policies and uh the most important thing is the the determination and the openness to autonomous driving from the political leaders or officials and of course this is a new thing and Uh, there's no perfect new thing, and, but now it's getting into a certain level of maturity that can be deployed in a large scale. I'm not saying we can supply a kind of transportation tool that is 100% safe, but we can provide something that maybe 10 times safer than human driver. Do you want to take it? If you want to take it, and then you have to generate a certain regulatory framework. One thing I want to say very a few good things about the Saudi government and also the UAE government. We plan to the in the in the Mid East to our big surprise, you know, the political leaders, local government officials, they are very open, very determined, very, uh, uh, take lots of responsibility to put all of all of this autonomous driving into reality and I think we write team team up with Uber will soon provide autonomous driving. Robo taxi in Riyadh and we will soon uh uh supply full driverless robo taxi in Abu Dhabi and Dubai. Today if you go to Abu Dhabi, you can already hail Uber through auto driving vehicle, but with a safety driver, but maybe next month once we get the final approval, we will have full driverless operation just like what you have experienced in San Francisco. So truly driverless, no, not even the sort of comfort level of the person in the passenger seat could be there if something goes wrong. Is there um. An infrastructure element to this. I mean, I think of the fact that Riyadh and other cities in the kingdom have a lot of new construction, you know, up to date, top of the line broadband. Uh, but you also operate in cities like Barcelona where the infrastructure is, you know, many centuries old. Uh, are there, when there's, when the infrastructure is older, are there different kinds of challenges? Yeah, but definitely the infrastructure is old, it will cost extra challenges. But here I want to make up a point is like we've recently, I would say breathtaking progress of large language model and uh neural network. Previously, we probably need to rely a lot, a lot on the infrastructure, but now with the development of the AI technology, we start to depend less or less on the infrastructure, but still a good road, very good wireless connection is necessary. I'm very glad to see that Riad Abi Dubai, they all have very good infrastructure and um so I'm quite excited to Uh, fastly deployed, uh, auto driving vehicles in these cities in a very large scale. What are the um infrastructure is no longer a barrier to expansion, um, and I was actually surprised uh. To hear that it's relatively inexpensive, that the per unit cost of a driverless car is if, if you had asked me before our conversation how much it would be, I would have guessed a much higher number. But can you tell me the number you, yeah, I think currently the cost of for us to build our auto vehicle is a little bit above like $40,000 US dollars, $40,000 yeah, think about $40 40,000 dollars US dollars you can. You know, I got a Tesla Model X. It's much more expensive than, you know, in the US, much more expensive than 40,000. So, you know, leverage on the strong, very strong supply chain of electrical vehicles in China, right? You can get a lot of things, you know, very cost effective price. And so that's the price and the point I want to make is for technology, our product will just come and can give us very good experience. We should strive for better experience, for safety record, not for the low price. I got enough with this kind of lowered the price and flooding to the market. That's not good to anybody, but don't always try to win the competition by lower your price. I don't think that's a wise movement. I think Elon Musk recently said if just one of these things gets in a crash, we'll have national headlines for months. Um, I mean, there is that there's sort of the cultural issue statistically. Autonomous vehicles safety record has been very, very strong, but because there's still a novelty, it seems that if something does go wrong, the attention it attracts is outsized. If I crash my car into a fire hydrant, nobody cares except me. If it happens to an auto hominious vehicle, it's it's news. That always happens to the, to the invention to the new stuff, right? You know, if you just, just dig into the old news, old news old papers, right, like one airplane just invented by Wright Wright brothers, a crash of airplane will make to the headline anytime, even nobody got injured. So it's, it's, it's just like that. It's our human civilization history, but uh. Great things like uh as entrepreneurs, as inventors, we should be really serious enough, we should be brave enough to push forward for the new innovation because we live in the AI era, think about our offspring, our, our kids, if we can make this a reality, they will appreciate us. Is there also an environmental case to be made for uh for autonomous vehicles? Is is part of the case that eventually you just don't need as many vehicles and there's a leveling off of? Yeah, I think it's all depends. It's actually in order to, you need to have to collaborate with the municipal governments because, you know. Actually, there's one point I want to make, like how many cars can the city hold? It's not depends on the population, it depends on the road networks. OK. If you build 10 times more cars, they basically they just park in the parking lot, consume the resource. So if we have an efficient way to operate so that everybody can hail a car when they need it, that is the number of vehicles you need and actually that's very friendly to our environment, very efficient. Is there, uh, that makes me realize we rides products right now are all essentially, uh, for hire. There's robo buses, there's robo taxis, but there isn't say a robo minivan that I would buy to put in my own driveway. Is that part of the business model that you essentially. that you wouldn't be intending to produce these things for individual ownership. Oh, we won't. We actually, we, we will, we have already produced something that's very similar to FSD and we write uh produce advanced system actually currently. Uh, Tinao with Bosch, a European company, the number one tier one supplier in the automobile industry, we team up with Bosch has already supplied, uh, other system to Cherry, and also we, we supply other system to GAC and also we talk, talk with, we are in talk with several carEMs. I think people will need it, you know, if you have experience to drive these days. You know, it's just like using smartphones. Your old Nokia phone is good enough, durable enough, but once you switch to your smartphone, you'll never got back. Once you switch switch to FSD and you can enjoy the trip. So I, you know, many of my friends using Tesla, and they will never switch back. Given that that's possible, given that that adoption rate is as high as it is, what's the How widely does we need to be adopted? How how widely do AVs need to be adopted before the autonomous vehicle businesses are, are profitable? I mean we write, I know you, you're approaching it, but you're far from it. Uh, how, how big does the market penetration need to be before it's actually a profit making business? It's a very good question. It depends on the market, you know, if you, uh, you know, if you want to deploy your robo taxi in the market, where I have a rule, OK, I have a rule of thumb. So in any market. If the cost of taxi or car hailing service is like more than $1 per kilometer, that is a very good market. Today with the price of ride with the cost of ride we can already make profitable. Anything between 50 cents US dollar, I'm always talking about anything between 50 cents to $1 that's why you have to optimize a lot of things in order to make the whole business profitable. Anything below $50,000 it's very, very, very, very challenging. I see. But the cost of everything behind that, the construction of the vehicle and the and the AI stack itself, those are all steadily lowering. Yeah, that's a big deal. Do you have a, where do you think the industry will be five years from now, 2030? What is your vision 2030? 2030, I think all the major cities, um, you know, if they are open to auto driving, you will see like what you have seen in San Francisco today for all major cities will be like to be like that. And besides, I'm more excited about talk about within the 10 years frame, I think we can build the autonomous driving vehicle in L5, which is L5 level, which is autom driving vehicle will drive as good as human driver. I think that is just like the GPT moment these days, you know, GPT already can. Achieve a certain capability in the in the law set in a mathematic competition it's already parable to comparable to the top human player. I think L5 within 10 years, you will see um the driving capabilities on pair with the best human drivers. So L5 within 10 years comparable with the best human drivers. Uh, I love driving, but I really look forward to doing less of it. Tony, thank you so much for joining us. Such a pleasure. Thank you.