Latest / Autonomy Insiders / Grab's Autonomous Vehicle Playbook for Singapore and Southeast Asia | Dominic Ong, GM Autonomous

Transcript

Grab's Autonomous Vehicle Playbook for Singapore and Southeast Asia | Dominic Ong, GM Autonomous · Autonomy Insiders

• Daniel Abreu Marques: Welcome to Autonomy Insiders, the show where global industry leaders in autonomous driving unpack their real-world insights. Dominic Ong: In heavy rain situations, when there are occurrences of flash flooding, you actually see demand spiking nine to ten times, And interestingly, this kind of demand shock caused by rain is actually positively correlated with autonomous ODD contraction. we're actually aiming to roll out about 50 autonomous vehicles in the next few months, so that's actually a step up of our existing fleet fivefold. and we intend to map the whole of Singapore Daniel Abreu Marques: Welcome back to Autonomy Insiders. I'm your host, Daniel, and most conversations about autonomous vehicles are conversations about San Francisco or Phoenix or Wuhan. So cities with wide lanes, clear marking, and expensive drivers. And Southeast Asia is none of those things. It has some of the densest and most improvised traffic on the planet and some of the cheapest driver labor in any middle income region. So on paper it should be the last place. Robotaxis arrive. And yet in April this year, a residential estate in northeast Singapore became the first neighborhood in Southeast Asia with a public autonomous passenger service. And my guest today runs that program. Dominic Ong is the general manager autonomous at Grab And Dominic, I'm so happy to have you on the show today. Welcome to Autonomy Insiders. Dominic Ong: Thank you so much, Daniel, for having me. Really excited to be able to share our story on your platform. Daniel Abreu Marques: So before we get into any of the technology, so let me start with the basics because I think our audience knows the AV industry well, but maybe not everyone knows Grab at the full extent. So maybe give me the two sentence versions of what Grab is and also what share of Southeast Asia's daily movement run through your App Dominic Ong: No problem. Well, I kinda wanna say that if anyone has traveled to Southeast Asia and has tried to arrange for transportation on your own, you likely would have touched the Grab app. in a nutshell, Grab is a NASDAQ listed founder-led super app company that has businesses in mobility, food delivery, grocery delivery, digital payments, and financial services. So currently we are present in over 900 cities, eight countries in Southeast Asia. And Our total transaction volume currently accounts for about 0.5% of the GDP of the entire Southeast Asian region. In fact, today I want to say we serve over 54 million monthly transacting users. The company was basically founded with a mobility service as our foundation. we wanted to offer safe ride share rides, in a geography where this wasn't always readily available. And we pride ourselves on being extremely, extremely hyper-local. You know, basically, how do we make advanced technology? Technologies work in the Southeast Asian context. And of course, we also have a mission to want to drive Southeast Asia forward, by creating economic empowerment for everyone. So the driver partners and the delivery partners and merchant partners who are actually working on our platform, their earnings, how they are able to create a sustainable living for themselves on our platform is something that's very, very important to us. So in a nutshell, that's grab. And sorry it took more than two sentences. Daniel Abreu Marques: Yeah, that's that's impressive. And you are the general manager for Autonomous and I think that Dominic Ong: Mm-hmm. Daniel Abreu Marques: title exists at Grab for the first time probably. So what are you actually accountable for? Dominic Ong: Okay, well, just want to take a bit of a step back. the role may be new, but I've actually been looking at autonomous vehicles for over a decade. in fact, before I joined Grab many years ago, I used to work for the Economic Development Board of Singapore. And back then I was responsible for developing what was then a very nascent autonomous and automotive industry, Now I joined Grab ten years ago, and in this last ten years, I've actually always been looking at the autonomous portfolio. in 2016 we announced a first ever Autonomous on-demand rideshare partnership with Nutonomy back then in Singapore. And in 2018, we made a small investment into Andrew Ng's autonomous driving company called Drive AI. So in fact, I've been actually looking at this space for a long time, despite the title being a relatively new one. I will say that, as the GM of autonomous, I do have like a three-pronged vision, for autonomy. The first is, how do we actually make autonomous technology cost effective, such that it is commercially deployable in our part of the world. The second mandate that I have is about how do I localize this technology properly in Southeast Asia so that customers are actually willing to pay for it. And this includes things like making sure the infrastructure that is surrounding autonomous technology actually supports it properly in our environment. The third mandate that I have is actually looking out for potentially interesting technologies that could be relevant, five, ten years down later for Southeast Asia. It's a similar philosophy to why we looked at Nutonomy back in 2016 and drive AI in 2018. Now obviously what I'm responsible for ladders up to this entire three-prong vision, I look after strategy, partnerships, all kinds of investment deals. I'm responsible for the unit economics, and most importantly, the entire operational setup required to make autonomy successful in Southeast Asia. Daniel Abreu Marques: Super interesting. And maybe let's now get started with the premise of this whole program. So I think almost everywhere the business case for AV starts with expensive or also scarce drivers. But I think Dominic Ong: Mm-hmm. Daniel Abreu Marques: in most of your markets, drivers are not that expensive or scarce. So what is the actual problem autonomy solves for Grab? Dominic Ong: Okay, so maybe I just wanna take a couple of steps back first, and just provide a little bit of context on Southeast Asia, which I mean you've kind of already started off very nicely talking about Southeast Asia at beginning, Southeast Asia is a really dynamic, region that comprises a mix of cultures and urban and suburban areas. And I mean you can tell this from the GDP per capita because it ranges from 2,000 US dollars in Cambodia to over a hundred thousand US dollars in Singapore, So we're actually dealing with like a wide spectrum of economies, and cultures here. And amidst all of this, the main philosophy of Grab is how do we deliver technology At Southeast Asian prices, So, for example, the average ride-share fare in the US is 14 US dollars, However, if you look at like two-wheel motorcycle rides in Vietnam, they're about one US dollar on average. So the question here is to make the economics work, do we actually try to deliver like a 14 US dollar experience, and have it one fourteenth of, the effect or do we really try to deliver the entire 14 US dollar experience at a $1 price, And I mean we've basically chosen to offer the best possible technology, at prices that are relevant to the region. And this you can imagine is not an easy technical feat. because of everything that we've done to try to deliver a world-class technology to Southeast Asia, I think people actually, see us as a reliable everyday service, akin to running water, Because they rely on us for their commute to work, they rely on us for food delivery, grocery delivery. I mean, and we really try to meet people where they are. So with all of this context in mind, you can imagine that for our business, reliability is basically P0 to us, So our mandate is really that of reliability. Look at us like a public utility service, like running water. And basically, the problem that autonomy has to then solve is how does it contribute towards this P0 priority of reliability, And actually, I see autonomy, like contributing to this end goal in two different areas by first of all helping us solve what we call peak hour gaps. I was looking at our 2026 data, Basically, in a city, in a very densely urban city like Singapore's, the demand can easily spike about four to five times during peak hours, Then there's always the question of are the drivers available and online and are they able to serve the demand during these peak hour intense periods, And that's where I suspect that autonomy actually has a role to play in actually helping up smoothen what we call these peak hour spikes, Because you don't want to run a business in which Like it's super spiky. At the same time, I want to say that autonomy could potentially help us solve connectivity gaps as well. So, meaning, you have situations where, let's say there are townships that are out of the city center, sometimes they aren't always the most popular locations for our driver partners to want to serve. can autonomy actually help us plug that gap? So I think in helping us solve these two problems, it basically led us up to that end goal. of reliability. so for now we're looking at autonomy to help us continue our philosophy of being like the most reliable platform in the region. Daniel Abreu Marques: This makes sense and also your answer to all of this is what you call a hybrid model. It's a term I think that is also often Dominic Ong: Yeah. Daniel Abreu Marques: used in the in the industry. but maybe let's also define it. So what actually is the hybrid model for you and also maybe what is the alternative you reject? Dominic Ong: Yeah. So for the hybrid model, how I define it is it is autonomous vehicles and human drivers operating on one network, booked via one app, with the same pool of demand, but dispatched interchangeably, So meaning we have the intelligence behind the scenes on the platform determining whether or not an autonomous vehicle or a human driver would best serve a customer's request, The platform will choose between them trip by trip. in this situation, I would say that the hybrid model is one which really encompasses the best of the both, instead of forcing a choice between human and robots. And in relation to the P0 being reliability is for what I shared in the earlier question, the hybrid network actually offers reliability in the most practical manner from the perspective of wanting to administer and to deliver a mobility service that is truly, truly reliable, now we talked about peak hour gaps and connectivity gaps. The other thing that I wanted to raise, was that in the context of Southeast Asia, rain really constantly. contributes to an extreme variance in demand, In heavy rain situations, when there are occurrences of flash flooding, you actually see demand spiking nine to ten times, And interestingly, this kind of demand shock caused by rain is actually positively correlated with autonomous ODD contraction. And what I mean by this is that when demand spikes due to rain, you end up with autonomous vehicles that are maybe not able to support, ride requests, in in areas of heavy rain because the vehicles haven't been trained or because the floods actually prohibit the smooth operation of these vehicles, So what do you do in a situation like that when you still have to serve the customer? because to reject the ride request means that you will end up disappointing them and they will leave the platform. Now, the second reason why I feel that it also offers reliability in a very practical manner is because you don't have a situation where you're overscaling the fleet to meet peak demand. Because as we as we had ascertained, in a densely urban environment, peak I mean demand sometimes can be very spiky due to a variety of different reasons, And if you had chosen an alternative, the alternative being you only have 100% autonomous vehicles. Then basically you would have to overfleet to cover that peak demand as well, And during periods of low demand, you end up with a situation where a lot of the vehicles end up remaining idle, creating a utilization problem, and because of the need for us to be extremely, extremely practical and prudent, in our part of the world, we really want to be very conscious of like how we're utilizing capital, how we're actually buying assets. And finally, one thing that I think is very close to my heart as well is. In a hybrid network, you continue to allow for room to have white glove services that are delivered, by driver partners, So for example, if you have families with children and they have problems with the car seat with luggage, maybe you have passengers with certain disabilities who actually need help, you have elderly folk who are trying to take the vehicle and maybe they just need a bit of help, with opening the door or walking them to the entrance of the medical facility. And all that, there will still be that dependency on having a human in the vehicle. And so I think that the hybrid network allows for us to cover all of these situations effectively without disappointing the customer too much. The fact of the matter is that autonomous technology today, in 2026, isn't quite yet perfect. And hybrid allows us to continue running a trusted and reliable service that doesn't hurt confidence. when autonomous vehicles fail because you really don't want people to try it once, twice and then, they end up leaving the platform because they're like, well, it doesn't really fit into my daily needs, Yeah. Daniel Abreu Marques: Yeah, super interesting. And maybe one question that you cannot answer right from the spot, but would be super interesting, how how probable is it that you have such an flooding or rain that you cannot drive with autonomous vehicles in South East Asia? Is this something that is really common or is it more more scarce? Dominic Ong: yes, extremely. it's extremely common in Southeast Asia actually. in particular in countries like Singapore, Malaysia, certain parts of Thailand as well, I mean these are countries that are classified as, being within the monsoon region. Thunderstorms are actually a fairly frequent occurrence, in the period between August to December. Yeah. and then you end up with a situation like when there is heavy rain. we have seen our demand spike easily nine to ten times, So imagine you have that kind of crazy spike and yet, you are constrained in being able to serve the customer, because the vehicles cannot operate, during those conditions. Yeah. Daniel Abreu Marques: Yeah. That's super interesting. maybe let's go there directly to your deployment. So as far as I can tell, Punggol is the first public autonomous passenger service ever deployed inside a residential neighborhood in Southeast Asia. So maybe let's Dominic Ong: Mm-hmm. Daniel Abreu Marques: build it from from the ground up. So can you walk me through the division of labor between Grab Your autonomous vehicle partner WeRide and the land transport authority. So specifically so who is the operator of record, who owns the vehicles and who carries also the risk if something goes wrong? Dominic Ong: Yeah, absolutely. So we have a very Unique and great relationship, with WeRide and with LTA, as far as this Punggol pilot is concerned. in the case of WeRide, they are the provider of vehicles, they are the L4, technology stack owner. we have done a partnership with them in which they are offering the vehicles to us. In the case of LTA, they are the regulator. but at the same time, also to some extent, I would say that they are the innovator as well, because when we did The Punggol pilot, we actually co-created a lot of what the pilot would look like with them. So I remember that the conversation took place along quite about I want to say over a year and a half ago, where we said, Hey, we have some ideas about how we want to do this. Can we really like sit down with you and work through all the details? Because back then LTA was really seriously contemplating how to enter into service a fleet of autonomous vehicles in Singapore. So they're the regulator Meaning they put all the safety frameworks, all the guidelines, all the testing requirements in place. But at the same time, I also see them as an innovator as well, being able to sit down with the private sector and say, hey, All these requirements, all these boundary conditions, let's try to do it in a way that doesn't become prohibitive to technology innovation. Now, the role that Grab plays is that of the operator on record, And what this means is that for the purpose of this pilot, we actually own the vehicles on our balance sheet. we operate the vehicles, we integrate the vehicles onto the Grab app such that they are available and you're able to book a ride with them. At the same time, also, we have invested heavily into what I call the localization layer, And maybe I'll just take a little bit 30 seconds to kind of describe what this means. the localization layer basically refers to, grab investing into software and hardware capabilities to actually help our autonomous vehicle partners. So imagine you could be an AV company who comes to Singapore with, maybe Not a full stack, but we could work with you and have our engineers kind of sit down with you and try to make up for any kind of gaps that you have in the launching of the Singapore service. But of course, you could also have AV companies who come to Singapore and maybe they have like a full stack, They say, hey, we want you to just adopt everything, and we'll say, okay, no problem. We will basically build on top of your stack and try to improve efficiency and try to make this technology. Work in the Singaporean context. And concretely, what does this actually mean? For us, it comprises hardware. So, for example, we have our proprietary KartaDashCams which are inward facing cameras and outward facing cameras, That we install into the Wii Ride vehicles so that the regulators actually have a very clear view of how what the passengers are doing within the vehicle, as well as an additional pair of eyes on what the vehicle. Is seeing ahead of it, We have what we call the maker-checker model, which is that because of the tight partnership that we have with WeRide our team also sits down with them and reviews all the edge cases, the data that comes from the car, what it's actually telling us, and we say, hey. These circumstances are unique to Singapore. This is how you should tweak the software. This is how you should tweak the driving characteristics. So in a way, we regard them really as a partner in that we sit down side by side to do these things versus saying that, hey, this is entirely your responsibility as a software stack provider, And the third aspect of this is that we also have invested into building a fleet management middleware that actually plugs into different AV OEMs as well. Because what we want to do also is we want to say, hey, we through this middleware, you're able to actually have oversight on all the vehicles, how they operate, and exactly how they are functioning on the grab platform itself. So this is what I term Daniel Abreu Marques: super interesting. Dominic Ong: the localization layer. Daniel Abreu Marques: really interesting, yeah. So basically you then also help to tweak the AV partners virtual driver to Singaporean edge cases. So that's that's really cool. Dominic Ong: Yes, I will say that we we try our best to do that, because we also understand that in the Singaporean context, there are many local nuances that, may not always port over automatically, And I think the philosophy here is that we just want to be a very helpful partner to you. Daniel Abreu Marques: Makes sense and maybe a bit more about your first pilot. So the first pilot, was on a fixed route more or less. And now you also have announced an on demand pilot. So for a Punggol resident, what actually changes and also maybe why now? Dominic Ong: Well, I have to say that we're actually extremely excited about this step up for Punggol, but maybe I'll just take a little bit of time to address how we started this entire Punggol pilot. we started testing in Punggol, and as I told you, the approach had been co created with the Ministry of Transport and LTA in Singapore. We wanted to actually put autonomous vehicles in a neighborhood that actually had a mix of commercial. commercial and residential, Because you wanted people of different profiles and, different backgrounds, to be able to be exposed to the vehicle. we did a fixed route shuttle because we wanted to really try out this technology using a very phased approach. And this was very much in line with in general, how the Singapore government rolls out innovation. Because they are I would say that the approach is usually very considered, very deliberate, and it's particularly important to do so with autonomous vehicles given that safety is obviously paramount. So we said, okay, let's try it on a fixed route shuttle. Let's really like make sure these vehicles can actually be localized properly. And let's make sure we socialize this technology with the public because so far, the public has only ever seen this technology in San Francisco or in different parts of China. They haven't actually seen it in Singapore. So let's enlist their help also, try out the rides, and get as much feedback as we can from them so that they also co-create. What the eventual rollout will look like. And in a way, also, it it's this is extremely crucial for trust. So as we look towards the ride hailing service rollout in Punggol. we're actually aiming to roll out about 50 autonomous vehicles in the next few months, right? so that's actually a step up of our existing fleet fivefold. and we intend to map the whole of Singapore and hopefully, bring the service beyond Punggol. Because I feel like we've done a lot of work in Punggol. People in Punggol are already very, very familiar, with our service already. we really want to bring it to more parts of Singapore. Now you may ask the question also, like, why are we going into ride hailing? Now. This pivot was interestingly actually due to user demand because during the entire fixed shuttle service duration, we were collecting a lot of feedback from residents, and many of the residents were extremely forthcoming with their feedback. And they told us that they actually wanted flexible bookings, direct connections, basically allow us to book this service exactly as how we would book a grab. And that was really what gave us the confidence to us and the regulators to say, okay, we're ready to take this bold next step forward. to go into autonomous ride hailing Daniel Abreu Marques: And you went from forty thousand kilometers in early May to about ninety thousand now. and miles and also disengagement is what the industry always quotes publicly. But it'll also be interesting to hear from your end. So what is the metric your team actually watches internally? Dominic Ong: Okay, so miles per critical intervention, that is definitely an industry standard. And it is relevant in the initial phase when we're trying to prove out the technology. And by the way, currently we are at about over 120,000 kilometers with no critical intervention so far. So that's something that we continue to to amass and to work hard on. miles per critical intervention is important at the initial phase where you're trying to prove the technology, right now we're in a phase where we're trying to prove the operations. And so vehicle uptime, how much time in a day is usable for the ferrying of passengers, is a metric that we're actually watching very closely right now because it tells Tells us how efficient our depot operations are. And we've just set up a depot in Punggol and this is where we're really bringing in a lot of know-how on operational efficiency, how we actually set up the entire space, how we think about the vehicles ingress and egress. all these things have a bearing on the vehicle uptime, and so that's a metric that we're actually looking at, very carefully right now. In the next phase, when we start doing autonomous ride-hailing that's actually when we will start being concerned with utilization as a metric, Which is how much of the vehicle's available time is actually occupied with a passenger. And I think eventually when we are really comfortable and we start offering the service widespread in many, many different areas of Singapore and we're actually running it commercially, that's also when, we will look to getting a metric which is hopefully a combination of, operational efficiency, uptime and utilization. Yeah. Daniel Abreu Marques: and you mentioned already also the depot and you said you built that in Punggol. So do you envision Dominic Ong: Is it? Daniel Abreu Marques: also to operate all the depots you need and also do the fleet operations or do you want to outsource it to third parties providers? Dominic Ong: that's an interesting question. that is actually an important strategy p ton that we're working through right now. I think that what we see in other markets is that in the case of Waymo, let's say let's say we take Waymo as an example, they have markets, they have cities in which they handle their own fleet operations and they own the fleet themselves. They have cities in which they have contracted this function And I think it's great that in the US market you actually have choice. In the Singaporean market, I would say that because this technology is still up and coming, we're still at a very early stage where you don't have an ecosystem, a thriving ecosystem to actually support the different models. So I would say that as a start, we will look to relying on ourselves to do fleet operations. Also, because for the initial batch of WeRide vehicles that we are putting on our platform, we actually own the assets as well. So we will start with that. And I think as we start scaling up the fleet, we will also be looking at whether or not there are available fleet partners who may be prepared to actually take on this function. And I suspect that as we get closer and closer to commercialization, once there is regulatory clarity on the direction that we are going and the speed at which we are going in the scale-up, that is probably the juncture at which there will be a lot more interest coming from existing fleet owners to say, hey, I'm willing to actually. take this next step. Yeah. And so the work that we're doing hopefully will be an encouragement and a guidance, for many of these players, who might be watching the space but not ready to take the step yet. Daniel Abreu Marques: And maybe now let us widen the lens a bit because Punggol is Singapore and it's only one place in all your and your entire footprint. So you now have real mileage. And could you maybe also rank the top three things about the whole Southeast Asian roads that could break an AV stack trained in China or the United States. So we now hear that it's rain and flooding but maybe there are also more things that could Break the stack. Dominic Ong: Yeah, so before we actually touch on the physical edge cases, I actually wanted to talk about an economics edge case, which is Singapore, I'm not sure if you're familiar, but Singapore is likely the hardest place in the world for, autonomous vehicle economics to work out. And that's really because of the taxation regime, because The entitlement to put a vehicle on Singaporean roads actually costs more than six figures in US dollars. this was the approach that had been adopted by the Singapore government to I guess to moderate the level of car ownership in Singapore, because Singapore is a place of very limited land, and we don't want a situation where we end up having to build too many roads or having the existing roads be too congested. And so this approach towards car ownership has actually been in place for decades already, And so, in a way, this presents also a bit of an economics edge case, Because you can imagine that an idle autonomous vehicle here will be an Extreme extravagance. It's not just a small inefficiency problem. It's like, my goodness, it's really costing you a lot on a per hour basis, And so I would say that you cannot hide a badly utilized vehicle in this market. And in a way, one of the like extreme things that we do want to test out is whether the economics actually works out, And obviously, if the if we're not able to get a high level of utilization in the vehicle, then that would be one of the things that would break the entire market. Model, Now, but going back to physical edge cases, which I believe is what your question was about, I would say that the top three are likely going to be rain, as I mentioned, because a tropical downpour changes lane visibility, sensor return, and pedestrian behavior all at once. In fact, we have seen situations where a tropical downpour can actually reduce visibility to near zero. so it's really difficult even for human drivers to navigate that situation. And we haven't actually put these autonomous vehicles to that kind of a test yet, so rain is probably number one. The second would probably be motorcycles, because two-wheelers are actually a very common modality of transportation in our part of the world. Singapore included, For example, a lot of the grab delivery riders actually they deliver food using motorcycles. So you have situations where at a red light, suddenly you have a swarm of motorcycles kind of like surrounding the vehicle, And you don't see that in the US, Because motorcycles, I mean, are not as cheap or as commonly relied on as a mode of transportation. The third one, which is not so much I guess, something that will break the stack, but rather something that the stack definitely must be tuned to, is the right-hand drive requirement. Because in certain markets in Southeast Asia, the you actually have like right hand drive vehicles, Which basically means that all the traffic directions, has to be tuned for like a different side of the road. even things like how the passenger enters the vehicle, exits the vehicle, how you cross like a big junction and things like that has to be recalibrated completely. and I I wouldn't say that it's it's it's an obstacle, it it's rather it's a challenge that needs To be worked through. And I know that many of the players, are experiencing this for the first time in London. Daniel Abreu Marques: I read that also I think it's also WeRide that want to also create an right hand drive version of their GXR. So Dominic Ong: Yeah. Daniel Abreu Marques: and some people ask so okay, what is the point? Do you then put the steering wheel on the right hand side or what is the difference? can you talk about that? Dominic Ong: I think that there are two things here. First of all, I think it's really great that WeRide has decided to invest into creating a right hand drive vehicle because it really shows their commitment towards right hand drive markets. And I honestly wanna really like salute Tony Han for that. Basically, it underscores his commitment towards cities like London and Singapore. one thing that also is extremely helpful is the fact that the GXR vehicle that we're using, has one sliding door. So by configuring the vehicle in right-hand drive mode, it also means that the sliding door actually opens on the correct side. Meaning, for passengers to board and get off easily. Versus if we continue on with the left-hand drive vehicle and you end up having situations where the door has to open onto incoming traffic. Yeah. Daniel Abreu Marques: Yeah, this makes sense. Alright, so we now talked a lot about the the general ride-hailing part of your business, but you also have another half of your business, which is delivery, which you introduced also in the intro. So Dominic Ong: Mm-hmm. Daniel Abreu Marques: Zelos is also an autonomous logistics company that was one of your four original MOU partners, and Vietnam was announced as autonomous vehicles and delivery robots market. So in a dense South East Asian City, is then autonomous goods movement maybe also a better product market fit than moving passengers, or where does it actually sit in your priority stack? Dominic Ong: Yeah. well I wanna say that using robotics for deliveries is actually an incredibly interesting use case and it ties in with I guess what we term as urban embodied AI in general, We really do see that there is an interesting use case that is emerging, which is basically using robotics for deliveries to complement the actual human delivery service. There are two specific problems that robotics actually helps us solve in this situation. The first problem is actually the first and last mile problem, Because we actually find that in many situations, especially when you're doing a delivery service in I guess very heavily built up urban environments, you have delivery riders actually waste spending a lot of time trying to, deliver a package or a food order within a high-rise building and you're waiting for the elevator, you're trying to locate the apartment unit. Sometimes you have to pick up the food in a massive shopping mall as well. So it's very difficult to find where exactly you're supposed to be going to pick that up. And we find that actually Actually, there is about 10% of a delivery rider's time that is wasted, having to navigate all of this, Versus being on the road in transit towards the delivery destination. So we feel that first last mile is probably something that robotics can actually help us a lot with. And at the same time also during supply peaks, I talked about how demand spikes, during peak periods, right? And I mean for food delivery, for grocery delivery, they all have separate peaks. food delivery is during the lunch and the dinner time crowd. We basically hope to also use robots to kind of like complement the actual delivery riders to kind of smoothen out that curve a little bit. Now, to this end, we're actually considering Innovations from many different players. Zelos is obviously one of them, and we're trying to work through suitable use cases where we can actually use their solution in different parts of Southeast Asia. we also have an in-house robotics team that has a solution called Carri So Carri is actually an indoor outdoor robot as well, so it's very unique in that the robot can actually travel indoors and some part outdoors, and so in a way it kind of solves that problem of having to navigate a very, very complicated complex or campus, And you can imagine a situation in which a delivery rider just shows up at the entrance of a mall and Carri kind of like comes and he picks up the order from Carri versus he has to go navigate, the depths of the mall to actually find the restaurant or the cafe to pick the order up from. Yeah. So these are some of the examples of emerging use cases that we're seeing and we're actually very excited to continue experimenting to find out how best can they actually complement our existing service. Daniel Abreu Marques: Super interesting. And maybe let's also talk a bit more about the strategy behind all of this because I mentioned you have an in-house solution, but you also have taken equity in players like WeRide May Mobility, Momenta, also Vay so why pay for something you could arguably also probably get with a commercial commercial supply agreement Dominic Ong: no, that's a that's a very interesting question. we ask ourselves this question all the time, But as I told you, I've spent quite some time looking at this space already. And I think one conclusion that I've drawn from all the different partnership deals that have happened over the years is that it's really critical to actually drive strategic alignment with your partner. Because autonomous technology is still a technology that is being worked out, it is not at a stage of yet being commoditized, the technology is still being refined. And equity investment is actually really, really key for joint strategic alignment and ecosystem integration. because as we are finding out, within the partnership with WeRide there is a lot of work that needs to be done that traverses technology integration, operations, data. Regulations, and you really need your partner to be able to lean in with you to solve all of these things, which are very frequently intertwined with one another. Which is why we decided that an equity investment, where both parties have their incentives aligned and their interests aligned, to be the best way to really drive the joint ownership of the deployment so that both parties can actually grow and mature together. Daniel Abreu Marques: Makes sense and maybe not flip it around now. So from an AV Dominic Ong: Hmm. Daniel Abreu Marques: developer side of the table. So what does Grab bring to them that they cannot build or buy themselves? And I would say demand is here the easy answer. So give me the second and the third reason maybe. Dominic Ong: Yeah, I I think that it varies from partner to partner, so I can give you a few general points, that I guess hold true across many of the different partners that we've had discussions with. I will say first of all, it is real world data because basically across 20 billion rides and deliveries, since our inception, 14 years worth of data. we do have operating data on the different kinds of conditions, in the ride sharing domain and food delivery domain. so in many occasions, partners have actually found this experience actually very interesting and very useful in helping them accelerate, the the taking apart. of edge cases and understanding and predicting how exactly the stack is going to work in sit in different situations. and for us, because We have so many driver partners on our platform with great driving behaviors. We also enrol driver partners also in the process of training up the autonomous vehicle also, so that the vehicle actually performs like a really, really well-trained chauffeur driver, for example. So you can actually see this in the finesse of the ride, when you actually take it in Singapore versus other deployments. so we actually do feel that this experience, this data is something that is very cherished by our partners. the second thing that we could bring to the table is the regulatory relationships. Because as I told you earlier, one thing that we have also set in place as a result of our many years of operating in the region as well, are the deep regulatory. Relationships that we have forged with the different regulators in the region. One of the great things about these relationships is that they allow us to co-create. Because you call them a regulator, but the relationship that we have with them is one where we feel comfortable enough to come to the table to say, we have these ideas, we would like to implement them, and we would like to sit down with you to actually look at how policy can accommodate this new technology. And in many occasions, I would say we have been received. With open-mindedness and collaboration. And I think that's something that's extremely, extremely critical and important, especially when it involves the rollout of new technology and new innovation. The third aspect of this is actually workforce capabilities. And what I mean by this is that in certain markets, we do see that governments actually view gig economy workers as a very integral pillar to the economy. In Southeast Asia, in fact, know in many jurisdictions. The informal gig economy is actually a very very important category of the workforce that the government prioritizes because depending on the economic situation in the country because you have different economic cycles you have ups and downs They actually look to it as a means of, providing transitory employment as people kind of move between jobs. so because of the contribution and how integral they are to the economy, they also do believe that whenever you bring in innovative new technologies, it's also important to take care of this workforce as this informal economy as well, And this is where Grab also plays a very significant role because many of these informal gig economy workers actually they work on our platform as driver partners or delivery partners so we also feel that it's very important that as we get ahead of the technology we also help them transition into potential new jobs that open up as a result of these new technologies forming I mean one could say that as much as there is a lot of uncertainty about whether or not AI could actually impact jobs you also see a world in which it helps to create a lot of new opportunities and in the autonomous space we actually see that in the form of safety drivers, remote operators, fleet management specialists. we actually do need a lot of people who are well trained in the mobility sector to come and support the entry into service of these vehicles. Yeah. So those I would say are the areas that we would help them. And I guess in addition to that, I mentioned something earlier, which is the physical localization layer as well, So our invitation to partners is basically come as you are, we will work with you to make Singapore work. Daniel Abreu Marques: Super interesting. So one partner I mentioned earlier or one investment is Vay and Vay is not necessarily autonomy. It's like a human driving remotely and you invested 60 million dollars, which is also a serious signal. So is teleoperation for you a bridge technology or is it a permanent layer in the stack or maybe also a hatch in case level four takes longer than everyone hopes? Dominic Ong: Yeah, so earlier on, when I was sharing about the role of the GM of autonomous, I did say that one pillar in the vision was to keep a lookout for technologies that may be relevant in Southeast Asia five to ten years down the road, Vay stood out as a very interesting candidate as we were considering the entire spectrum of driverless technologies, because I think Vay is an incredible company. I think that they Have really done a lot of great work in integrating their remote driving stack into the vehicles. So, from an engineering perspective, I think that they are really a great company. I would say that we are working with them to actually understand operationally where they stand in this world where autonomy is picking up. We see them as a very interesting alternative technology that has the means of delivering a ride share service or a car sharing service at a fraction of the cost of what the incumbent. Are offering today. And to that end, we are working with them on many different business models to try to achieve this end goal. Now, in the area of autonomy, in fact, I will say that one interesting and emerging idea is basically what we call assisted autonomy. Whether or not the remote driving stack that Vay has can actually be a very nice complement to actually help the autonomous industry, And to that end, they already have a partnership with Kodiak, for example. I know Vay is actually. Actually integrated into Kodiak today. At the same time, could they also be used in vehicle recovery, for example, Because they also happen to have a deployment in Vegas where you have a lot of autonomous vehicle players that are currently running commercial services there. So I think Vay gives us an opportunity to really push the boundaries of how we're thinking about autonomy by presenting a technology that can be very nicely complementary to the incumbents. Daniel Abreu Marques: Sounds interesting. And now let's come to the big news in Singapore. So Waymo just announced they're also launching in Singapore in 2027 and they want to provide commercial rides in 2028. So how are you viewing this development? Dominic Ong: Daniel, as someone who has looked at this industry for a long time, I want to say I am extremely excited about it. I am really glad that Waymo is actually joining us and Comfort Delgro on this joint mission in creating a thriving autonomous landscape for Singapore. there are three reasons for this. first of all, from a consumer perspective, Singaporean riders will now get more choice, because they would be able to choose from existing operators as well as the new operator, Waymo, and and make their own decisions on which operator actually serves them the best. Second of all, I would say that Waymo's entry is a validation, that there is a pathway to commercialization, even in a market like Singapore where the economics are difficult, So in a way, it kind of validates All the intuition that we had about, hey, is Singapore viable as a market, and I feel like Waymo's entry will also end up accelerating a lot of our thinking around the regulatory frameworks, the safety frameworks. So all in all, I would say that it's it's a very positive development for autonomy in Singapore and Southeast Asia. and I'm really excited about them coming. on our end, we're staying laser focused on delivering our service in Singapore, and we're really, really excited about the expansion that we have announced already, the five-fold fleet increase that's coming online very soon, how we're actually going to be carrying passengers on autonomous ride hailing services very soon. and I really look forward to being able to work in lockstep with the Singapore government to really work through the regulatory changes that will need to take place to accommodate all of us while at the same time serving the Singaporean population. Daniel Abreu Marques: And maybe one follow-up question because you you again mentioned fivefold increase of your fleet. And at the beginning you mentioned that you you own the first We ride vehicles. So if you look at the thing at scale, so do you still want to own the vehicles and carry the Capex? or would you want to to have third party fleet ownership as well? So what what would be the model at scale at the end? Dominic Ong: so Grab is actually not a stranger to owning CapEx, know, because in our part of the world, we have actually set up rental car companies in Singapore, Malaysia, and Indonesia where we actually own the assets on our balance sheet. And the reason for this is simple. It's because in our part of the world, many drivers have difficulty getting access to credit to be able to purchase their own cars. And so in having our own rental companies, we've basically made it A little bit easier for them to actually get access to a vehicle to drive on our platform, now we had made these investments in the early days when we first started the rideshare service, and the reason for this was because the ecosystem hadn't yet grown. So we've took it upon ourselves to say, let's make this easy for drivers, to be able to come onto our platform. I see this evolution happening similarly in autonomous as well, which is that at the onset, as I mentioned earlier, you may have fleet owners who are waiting and watching to see if this is actually a viable business to be had. So we happen to be in this phase where the technology and the business model is still being proven. I would say that until we get full regulatory clearance to actually run a sustainable business in many parts of Singapore, it is likely that we will still end up owning the CapEx and operating in CapEx ourselves. Daniel Abreu Marques: Makes sense. So Dominic, we're at the end. So thank you so much for for taking the time today. It was really insightful to learn more about Grab, to learn more about the Southeast Asian market and all the special and edge cases you encounter over there. So thanks again for joining me today on the podcast. Dominic Ong: Thanks so much, Daniel, for having me. Really enjoyed this conversation. Catch up with you soon.