This Week in Startups - The Global Expansion of Self-Driving Vehicles

Episode Date: March 11, 2026

This Week In Startups is made possible by:Circle - http://circle.so/twistSentry - http://Sentry.ioDeel - http://deel.com/twistPlaud - http://Plaud.ai/twistToday’s show:It’s self-driving time! We��re going deep on one of the most exciting spaces developing in the world, autonomous vehicles! We’ve got 3 experts on the show to talk to us enlighten us, Ben Seidl of Autolane, Ming Maa of Moove, and Nathan Parker of EdgeCase.What’s going on in the world of self-driving? How has the reliability of autonomous vehicles improved? What challenges are we still facing in the industry? Will the US reign victorious, or is China sneaking up? Let’s find out on TWiST!Timestamps:00:00 intro02:21 Uber teams up with Zoox! 03:02 Does EdgeCase work with Zoox? 03:03 Operational design domains 06:47 The challenges of bringing self-driving to new environments. 00:10:53 Circle: The easiest way to build a home for your community, events, and courses — all under your own brand. TWiST listeners get $1,000 off the Circle Plus Plan by going to http://circle.so/twist. 00:13:26 Plaud: If your work depends on conversations — interviews, meetings, calls — you need a Plaud NotePin. You can check it out at Plaud.ai/twist and use code TWIST for 10% off! 00:17:19 How has reliability of autonomous cars improved? 00:19:18 How Ben Seidl came up with the idea for Autolane! 00:21:08 Sentry: New users can get $240 in free credits when they go to https://sentry.io/twist and use the code TWIST 00:27:47 How Moove helps autonmous vehicles go to market! 00:30:49 Deel: Founders ship faster on Deel. Set up payroll for any country in minutes and get back to building. Visit https://deel.com/twist to learn more. 00:33:28 The logisitcs of managing fleets. 00:37:08 Why is it called Autolane and not multi-modal lane? 00:40:50 Does EdgeCase work with smaller self-driving vehicles as well? 00:41:52 How autonomous systems will interact with one-another! 00:43:14 Why AV’s should not talk to other AV’s on the road 00:45:07 The state of the self-driving market 00:49:12 Why Tesla isn’t involving themselves in California 00:50:34 Why the US is leading the way with regulatory clarity in autonomous vehicles 01:02:51 Ben Seidl says federal regulation is a necessity 01:05:20 What is holding back self-driving growth in the US? 01:06:26 What companies make the most money in self driving Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.comCheck out the TWIST500: https://www.twist500.comSubscribe to This Week in Startups on Apple: https://rb.gy/v19fcpFollow Lon:X: https://x.com/lonsFollow Alex:X: https://x.com/alexLinkedIn: ⁠https://www.linkedin.com/in/alexwilhelmFollow Jason:X: https://twitter.com/JasonLinkedIn: https://www.linkedin.com/in/jasoncalacanisCheck out all our partner offers: https://partners.launch.co/Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarlandCheck out Jason’s suite of newsletters: https://substack.com/@calacanisFollow TWiST:Twitter: https://twitter.com/TWiStartupsYouTube: https://www.youtube.com/thisweekinInstagram: https://www.instagram.com/thisweekinstartupsTikTok: https://www.tiktok.com/@thisweekinstartupsSubstack: https://twistartups.substack.com

Hello and welcome back to Twist. My name is Alex. It is Wednesday, March 11th, 2000, 26, and today we are here to talk about the future of autonomy and the current state of affairs in the Robotaxy and Autonomous Commerce World. This week in startups is brought to you by Deal. Founders ship faster on deal. Set up payroll for any country in minutes and get back to building. Visit deal.com slash twist to learn more. Century. Your team should be focused on shipping features not chasing down bugs. New users can get $240 in free credits when they go to century.io slash twist

and use the code twist. Circle. Circle gives you everything you need to build and scale your community-led business. Twist listeners get $1,000 off the Circle Plus plan at circle.s.o slash twist. We brought a great panel of folks here from all around the industry

to help us understand where we are and where we're going. So please join me and welcome. to the show or back to the show, really Ben. Ben Sidel from Auto Lane. Auto Lane provides software and hardware to facilitate autonomous commerce out there in the real world. Ben, welcome back. Thank you. Appreciate it. Pleasure, man. And then we have Nathan Parker from Edgecase. Edgecase helps autonomous vehicle companies find edge cases and problems before they deploy into production so they don't have any issues when they hit the streets. Nathan, I hope that was close. Glad to have you here.

Well done. Thanks, Alex. Thank you very much. And then we have Mr. Ming Ma from move, which is the biggest fleet manager in the world for the autonomous industry. You may know him from his work in the best at Grab. Ming, welcome to the show. Thank you. Excited to be on. I am too. I have to say that of all the topics we talk about on Twist, this is my favorite probably

because I despise driving. And I find it insane to this day that we let 16-year-olds here in the States get the behind the wheel of a suburban and go 80 miles an hour down the freeway. It just seems like a terrible idea. And so for me, as a lazy person and a safety-conscious dad, I just want things to go faster. So that's my prior going into this. Now, the biggest news today in the world of self-driving is that Uber and Zooks are going

to partner up. Uber, of course, the American Ride Hail Giant. Zooks, the Amazon subsidiary, have been testing in Las Vegas, Ben. Now they're going to work with Uber in Vegas this year and then expand to Los Angeles in 2007. So a relatively measured pace of expansion, but a good partnership. I'm curious from your insider perspective what this means for the industry. I think it's fantastic. I mean, it certainly validates the Uber strategy of aggregation of all

these different players. I mean, they've taken a very long-term view on the industry that essentially this technology will be widely disseminated across many, many partners. And Uber, you know, seeks to become that expedia of AV mobility. And I think adding another company like Zooks into the mix is further validation of that strategy. So I think it's, I think it's fantastic. And I I also believe that Zooks really needs this. I've taken a couple rides in Zooks, both in San Francisco and in Vegas. And it's a fantastic experience, but we're at a point now with AVs where if every single AV company has its own app, it's hurting the industry.

And so I think with Zooks being relatively constrained, a few vehicles, smaller geo-fences, it's very smart for them to open up the demand to a wider audience. and get a lot more people using their amazing purpose-built AV technology. So I think it's great. Nathan Ming, have either of you been in a Zooks? I haven't. So I can't really verify what Ben just said about it being a great experience. I have not.

So I can't speak firsthand on Zooks. We really need to do a group field trip to Vegas, apparently, because I've seen videos of this, and it looks tremendous. But I think they were doing just like an early circuit through Vegas, and it didn't seem that widespread. So the way that I see this, Ben, is that if they're expanding, and they've sorted out whatever early kings they might have had. Yeah, they're also testing in San Francisco.

It's only for employees and friends of employees, but they've been here in San Francisco for, I believe, almost a year. So, you know, taking Azooks through the hills of San Francisco, I did it one time on a Sunday night when there was almost no traffic, and it was just, it was magical. It's very, very cool, completely surreal experience because if you're not familiar with what a Zooks is, They refer to it as a Zooks because it's not a car.

Yeah, they seek to differentiate a Zooks from a car, meaning the Zooks has no steering wheel. It is not forward or backward facing. It's omnidirectional. So you get into it more like a tram. The doors open up like this. You get into it. You're facing each other as you ride. So I've taken it with three friends before.

and you know, you can have a group conversation with four people. Everyone's looking at each other face to face. And then you kind of get out. We have a picture of the Zooks coming up here. There we go. That's helpful. So it's like it's a totally different thing. And it is absolutely special.

So yeah, I mean, I think it's going to be great. What they're doing in Vegas is seven different stops, I believe still at the moment. When I took it at CES, it only had seven stops. So it's only point to point in Vegas that you can't take it anywhere you want to. But I believe when they expand here with Uber, it's going to be an entire geo-offense area. In San Francisco, if you are to take it, it can take you to any address within the geo-fens. It's not point-to-point. It operates a little bit differently than most robo-taxies do in the sense that you can only go destination to destination in Vegas.

I'm absolutely furious that I left San Francisco right before all the self-driving cars actually started to work, which is a cruel, cruel irony. I saw all the early testing and got to have none of the fun. Ming, when it comes to managing large numbers of cars for AV companies, and I do want to get into the economics of fleet management a little bit later, but with Zooks as our current topic, is it different, harder or more economical to support vehicles that don't operate like traditional cars? Because I presume that if you have a Jaguar eye pace,

it's probably pretty easy to fit into other fleet management operations, whereas Zooks probably has more custom parts, it's harder to park refuel? I don't know. I'm just curious how that impacts your operations and how popular that model is. First off, thanks for that question. I actually think there's a lot of different complexities involved.

That's not just tied to the specific type of vehicle or the specific type of the platform, but it's also just tied to the city itself and how every city is very, very different. So when you think about fleet management, one of the biggest challenges that we have is really around how do you fine-tune the system to think about how every city is,

different. Phoenix, for weather, for example, just pick one little thing. It's typically above 110 degrees during the summer, and sometimes it reaches 120 degrees. And in those types of conditions, we have to be very thoughtful about energy management on the vehicles, because the batteries drain just much faster than normal. You also have these very, very strange weather events called hubboobs, which are just these massive wall-like dust storms that engulf the entire city. And when that happens, you have to be very, very careful to ensure all the sensors, the LIDARs, the cameras, all the radars are working properly. Miami, which is the city that we launched in January, as a very different set of challenges, because obviously hurricanes, rain,

all of that occurs. It seems like almost every other hour. And so we have to be very thoughtful about how to protect the fleet against flooding risks, which is very important for electric vehicles. London, of course, will have its own different set of challenges. So I think not only is the format of the vehicle very important, but it's the locality and the unique differences in every market that makes the fleet management very interesting and challenging. Nathan, does Edgecase work with Zooks? We can't publicly disclose it. We work with, unfortunately, we do have several partnerships that are public. To pick up on what Ming was saying, though, one of the things around safety that's interesting with regards to what Ming was saying is something referred to as an operational

design domain. And so the companies, what they're doing is they're doing what Ming just described. They're looking at a certain geographic area. They're looking at the autonomous vehicle. They're assessing where is it going to operate? Is it going to be on surface streets? Is it going to be in city streets? Is it going to be kind of over the road trucking? What's the weather? Is it, you know, do you have a lot of rain, do you have a lot of ice? Do you have snow? Is there dust? All of the things that would potentially impact the performance of the autonomous vehicle, all of those things have to be solved and designed. At some point in time, you know, these companies are graduating into, you know, building fleets of vehicles. And so now you have to think about how operationally,

how are you going to make sure that those systems still perform at that high level. Yeah. So that's what we're seeing at least from a safety side. They'll clear one ODD, you know, and then they'll say, okay, great, that's cleared. We solved a lot of the hard engineering problems. Now we can go to the next ODD, which generally from a robo-taxie perspective means one city, we can now go to another city that's a lot of the conditions are the same as the previous city that was cleared. And that's why, in the case of Zooks, if you go from, I mean, let's say, just say Vegas to

L.A., both are relatively flat, both are in the same part of the United States, similar climate. So you would probably have some ODD overlap, is my guess. Yeah, absolutely. So the core engineering problem they solved in this case in the original location. So now they can go to other like regions very quickly because now they're just solving for the delta, right? They're not solving for the full design. They're solving for just the delta from where they previously launched. So because they're going from like to like, it's kind of a modest step by them. Would you have been more impressed by Zooks going from, I don't know, Vegas to name a Canadian city, Ottawa, where it's cold most of the year, apparently, because it's Canada. I think it's smart business. Most of the autonomous companies that we work with and talk to, they're very methodical on their expansion plans. And so they're thinking, what does the engineering, what does the tech do today? How do I continuously expand that? And then they think about, you know, they look at it through that lens of technical expansion, commercialization, and who are

the commercial partners in those new areas, and then the regulatory piece, you know, or do we currently have the ability to operate the way we expect to operate from the laws and rules of the road governing AVs. Building a community business is really hard. And if you're a first-time founder or independent creator, you may not be prepared for everything you need to accomplish. But now there's Circle. The complete community platform for creators and brands who are building new customer groups. Maybe you're teaching a course or you're starting a membership program. Circle's going to help you with every single aspect of forming and maintaining your new community from creating a branded website to tracking and monitoring all discussions, announcing and planning events, email marketing,

and everything else. It's extremely fast to get started. And you maintain total control over your new community's design. The branding is yours. And your data is your data. It's not shared with anybody else. Circle is by far the cleanest and fastest way to set up a home for your new community. Maybe you got a book club, maybe you're teaching a course. Or like me, you have an accelerator and a pre-accelerator, Founder University. We've been using this product for five years at Founder University, well over five years now, it is the easiest to use all in one platform and so powerful. So try out Circle today and get $1,000 off the Circle Plus plan by visiting Circle. So slash twist. That's Circle.S.O. slash twist.

Is it wrong of me to be a little bit surprised that this far into commercial self-driving operations that we're still having to approach most cities on a case-by-case basis for each group? Like whenever Waymo announces, they're going to go map a place. I know they're doing preparatory work, but perhaps it's a little surprising from where I sit that this is not more generally applicable yet this technology and still requires that much in situ preparation, if that makes sense. I don't know if this is best for Nathan or Ming or Ben, but why am I feeling slightly disappointed learning about how we're approaching safety in this case? Maybe Nathan's this is best for you. The companies take safety very seriously. And so it's not one of those things where we'll hope it goes okay.

They're very, again, methodical on how they prove to themselves and then kind of the ecosystem around them, whether it's commercial partners, regulators, that the system's going to behave the way they expect it to behave when they go into a new geographic location. So it's not one of those things that they are going to hope it works. They absolutely have to prove to themselves and have strong conviction when they go to another location, it's going to perform the way. they expected to perform. So I actually could see that being perfectly normal. All right. We're going to take a very, very teeny tiny break to let Lon tell us about something. And when we get back more about cities, safety, and how one company is blazing a different path. But in the meantime, Mr. Lon Harris, take it away. You know, if you're keeping a lot of different variables straight, like these gentlemen are in their safety work and you need to keep track of everything, we recommend

getting yourself one of these plod note S-pins. You can see me wearing one right here. All you do is you hit the button, you could start talking. Not only does it pick up everything that you're saying, but it actually cleans it up, it makes it nice, it keeps it organized, and it recognizes everyone else in your area. So if you were talking to a group like this and you had to keep track of what everyone was saying, it actually recognizes your coworkers, it tags it in your notes. It's an incredible way to keep things straight, whether you're keeping track of what you're doing, like on a walk or on the drive to work, or if you're collaborating with others in a meeting

and you need to remember exactly what everyone said, along with the nuance and the fine points of what they were making, we can't do, you can't do any better than getting one of these Plaud note S pins. And if your work depends on having conversations, doing interviews, conducting meetings, taking calls, you need to get a Plaud note pin. Check it out at plod.aI slash twist and use the code twist for 10% off. That's p-l-a-u-d.a-u-d-a-i-slash-twist. Thanks, Alex.

Pick it up from where we left off. Neuro is going out into Tokyo, and they are doing it without doing the usual mapping and preparatory work. They're calling it, I think, zero-shot autonomy. So as we talk about safety and the importance thereof, how is one company able to effectively go around or not skip a step, but pass one by without doing more work like Waymo does, for example? This is an interesting kind of new path for an AV company. I would say we're seeing some version of this on the military side and some of the autonomous vehicles companies that we're working with because they're going into kind of all-steer environments. They're going into terrain that they don't necessarily have

the ability to previously map. And so they want the capability to really, you know, lay in a vehicle and go, if you will. And so although a lot of the companies, again, that we've historically worked with definitely operate with trained data. The fact that there's a company, you know, kind of aspiring to do something like this on public roads. It's not entirely new, but it definitely will be something that we'll be watching. I would say, again, from an edge case perspective, our point of view would be, you still have to make sure it's safe, whatever that system is. You still have to prove that you're approaching autonomy slightly different,

but you still have to think about how do you demonstrate that the system's going to perform the way you expected to perform. I think just to add on to Nathan's comments, it's very, very challenging to get from two nines of reliability to five nines of reliability and covering all of those different corner cases city by city is one of the biggest challenges for autonomy. So, for example, the color of the streetlights are slightly different for different cities, at slightly different heights, depending on the cities that you're in.

When you're driving on the left-hand side of the road versus the right-hand side of the road, the conditions are very different. when you look at the curbs in different cities, some curbs are fairly tall, some curbs are fairly short. And so all of that influences how you think about pickups and drop-offs and navigations throughout the city.

You also remember, there's also an auditory component to this. So Waymo's understand when there's a first responder coming by the sound of the siren. So if there's an ambulance or a fire truck coming by, they'll know to pull that vehicle aside and let that first responder

go through first. But those sounds, as you can probably imagine, vary a little bit slightly, you know, city by city. And so really capturing all of those little nuances is what makes it five-nines of reliability. To get from zero-nines to two-nines and then from two-nines to five-nines Ming, is it like, you know, 20% of the work to get to two-nines and then 80% to get to five-nines, or am I inverting the amount of work per nine in terms of reliability? And what we're finding is we're finding new ways to make the system more and more robust every single mile that they drive. So we talked about weather as one of the key things to really understand.

Some cities in hot climates, for example, there's a lot of dust in the air, and being able to really model and calibrate the sensors to detect and really understand dust and fine particles is an important component. Once autonomy reaches the snowy cities in the north, being able to really understand snow and how that affects the sensor. stack is going to be very important as well. So I think all of this is really coming together now to make a really waymo the world's safest driver. I came into this conversation, incredibly excited about the quick pace of progress and self-driving. And now I'm discovering it's a little bit more

piecemeal than I thought. How has reliability of autonomous cars improved through the eyes of auto lane? Because you're dealing with them in the real world. So do you see improvements in how they drive and do these edge cases that take you from two nines to five do they show up in you know your interactions with them as a company you know the company that we're most uh integrated with is is actually Tesla and I I drive a hardware for Tesla every day 100% on FSD so I can mostly speak to that I've taken a lot of Waymo rides and we orchestrate Waymos on private property um at some of the centers that we support. But, you know, the Waymo capabilities because you can really only learn about them, at least we can, by riding in them, that's harder. So it's all about the amount of time you have

in the vehicles. But with Tesla, you know, we have a fleet at Autolene of Tesla's hardware 4. And, you know, we're driving them all day, every day. So every time that there's a point release for FSD, you know, that's being tested across our. organization and myself on a daily basis. And, you know, we, I came up with the idea for auto lane in August 2024 because I bought a brand new Tesla and was astounded by how good the FSD quality was even at that point. And kind of had the epiphany that if this technology continues at the rate that it's improving, then this is clearly a general purpose technology that's going to change the world. And if that's true,

then what we need to see is pretty significant and steady progress similar to the world of LLMs in order for this epiphany to become true. And so we've bet, you know, a pretty big asymmetric bet at Auto Lane on the idea that autonomy is not only going to improve at a consistent rate, but that it will eventually come to personally owned vehicles as well. So I can say over the last 18 months of building Auto Lane, if you looked at FSD as an example from August 2024 until today, it's significantly better. I would even maybe say twice as good.

I mean, it used to do, even in August 2024, it would do things that weren't safe, things that would make you not want to use it again. And unless you were kind of an intrepid early adopter, you probably gave up on it. You know, but I could kind of see through the fog and understand that like it is, if they keep improving at this rate, then these things are going to go away as Ming's pointing out, you know, step by step. And we've seen that happen every single quarter, if not every single month for the last 18 months. Debugging sucks.

And it takes up time your team could be spending on awesome new features and products. But now there's a better way. Centries, AI-powered debugging agent Sear. isn't just guessing about what might have gone wrong with your system. It's analyzing your actual data. And because it has context, it spots buggy code before that code ruined your entire day. Plus, SentryWorks alongside coding agents like Cursor, passing along an idea for a fix that can then be applied to your code-based

and sent to a human team member for review. End-to-end automation all the way from bug detection to pull request. So join the millions of devs and companies like Claude and Disney Plus who use Sentry to move faster. Check them out at S-E-N-T-R-Y-O-S-T-T-R-Y-O-S-T-T-E-O-S-T-T-R-E-O-S-T-R-E-R-E-O-R-E-M-O-M-O. From my vantage point and CEO of Adelaan and working with these different and testing, and I've tested Wii-Ride, I've tested Zooks,

I've tested Wave, I've tested, obviously, Tesla, Waymo, so almost all of them, and, you know, they're all really good. They're all really good and they're all pretty close to each other. So I'm very bullish on the progress that will continue to be made here. And hopefully that can be disseminated across all AV companies. Yeah, we're going to come back to you in Second Bank. We have an audience question about autonomous commerce that I want to get to.

But Nathan, from the Edgecase perspective, have you guys seen the types of improvements in the cars that you help safety check, similar to what Ben just described in how FSD has improved. I know you can talk about individual customers and clients, but have you seen a similar rate of them getting better at avoiding the sorts of issues that you need to flag? The short answer is yes. So you start to see some of their metrics like interventions per mile, some of those industry standard things.

I think the thing that's also interesting is the problems they're looking to solve. Like if you talk to some of the AV players, like what are some of their biggest problems? three to five years ago is some foundational engineering problem. Now they're solving like ecosystem problems or how do I scale? How do I work with customers? How do I build the commercialization engine to support the deployments that are coming? So I think it signals to us that they're definitely getting very close to launch and several have

because they're solving different problems now. They're solving very much a commercialization, go-to-market problem. Of course, there's still safety cases that have to be pulled together and some fundamental engineering work that has to be done. But the aperture has been opened on the problem set. So do you eventually fix yourself out of a job at edge case once you've solved the edge cases? Because there's only so many cities. There's only so many ODDs to use your term. Does those case eventually get out of this business once it's solved?

The true answer is no. I'm surprised to hear you say that. Yeah. So one of the other, and this is back to some of our background, part of the challenge is getting the system to market. What we also think is the bigger challenge is going to be keeping the fleets operational and keeping them performing to the safety standard. So you can imagine in the future, you know, a lot of the deaf fleets, they're, you know,

five to call it 50 vehicles. The problems, parts are going to go obsolete. You're going to have things that change in the configuration of these platforms all the time. So you have to not only now understand the safety profile of a single vehicle, you have to understand the safety profile of 1,000, 10,000, 100,000, a million with precision to understand, much like the airlines, right, you know, a grounding of a fleet. What is the autonomous vehicle community going to do when they have an incident? So we're built, that's part of what we've been building around our platform is that that real time monitoring, real time observability to understand are all of the vehicles in the fleet still performing to the safety standard that was set out, you know, way back when the engineers were building.

Okay, Ming, we have to bring you in here because this is, this is very fascinating. When I think of fleet management, in the back of my head, I'm imagining some high school kids parking cars in a back lot and then picking them up again and driving them out. Clearly, that's not the case based on what Nathan just told us. So how hard is it to keep the sensors clean and calibrated and to ensure that cross generations of different Waymoes that all the technology is working? And how do you test to ensure that the cars that are being used that you're helping to manage are at exactly 100% readiness to take people just as we expect airlines to have their planes knocking on wood, always ready to go. go. It's a great question. And I think the first, I think the first mental model is to, is to understand that managing an AV fleet is 100% completely different from managing a traditional fleet. When you think about the goal of a traditional fleet manager, it's really just BIS.

It's butts in seats. So you spend a lot of, a lot of energy looking for drivers. And then when you find a driver, you want to get that car out of your parking lot as soon as possible, and then you completely forget about that car until it comes back for servicing. So you're completely hands-off. That's traditional fleet management. Managing an AV fleet is almost the exact opposite in every single motion. So we spend zero percent of our time thinking about filling seats. Instead, we spend a lot of our time to Nathan's point, thinking about the motions around recording. every single vehicle when it comes back into the facility, how do we ensure that it meets the safety standards

that our writers deserve? And so having that traceability, every time we balance or repair a tire, how to ensure that the torque values are recorded for every single nut before it leaves the facility. And that process, that motion is much more akin to an airline level of safety than it is to a traditional fleet company.

Now, we also think a lot about the demand curve for right sharing. So very different from traditional fleet companies, we think a lot about how does the writer demand change throughout the day? And then how do I match the supply curve to that demand curve as best as possible? Now, companies like Wayne will have invested a tremendous amount of engineering to develop very sophisticated tools for managing these supply and demand curves. curves, but we find that every city is very different in certain small little ways. And so what I'd like to think about is how do we help them fine-tune the system for every city to borrow that analogy from AI? You guys raised a bunch of money last year.

And it sounds like you're doing all of the tricky and difficult and hard work, but not getting as much of the praise. And I presume not the entire economics of self-driving. So, man, have you considered rolling your own self-driving? self-driving car as well, because I feel like you have, you know, the muscle and sin you required to do it. So I'm curious why not go upstack, if you will. Our expertise is around operations and really helping the AV companies go to market in the

most efficient and the most sustainable way, you know, possible. So I think ultimately why companies like Waymo choose to partner with us is because, you know, we accelerate the commercial scaling of their ambitions into cities on a faster timeline. and in a more economic way. So what that means if you take a very concrete view on this is, first of all, we develop all the infrastructure needed to bring a Waymo to a city, and there's really three types of infrastructure that we work on. The first is charging locations where we service the vehicles,

we prepare them for our right-sharing services. Second, we also set up the service facilities that's needed to perform maintenance and repair, and to, again, to Nathan's point, ensure that every vehicle is as safe as possible and as roadworthy as possible. And then the third is we do have really parking infrastructure for situations when we have to defleat the fleet due to either bad weather events or other issues. San Francisco losing power, for example. Exactly.

Perfect, perfect example. And in I think any market that we operate in, we most likely have multiple facilities for each of these three. arch types. And the real challenge for a fleet operator isn't orchestrating this very distributed footprint of facilities to make sure that the overall fleet works as efficiently as possible. So that's really on the infrastructure side. The second thing that we do is all the operations needed to ready the vehicles for ride sharing. And the objective to Nathan's point is really to ensure, first of all, our P0 is safety, that all vehicles are as safe and clean as possible for passengers

while we maximize the total utilization of the entire fleet. So when you step back, we're not going to go upstack, if you will, into the R&D and the AV development. But the mental model that I would leave you with is we help Waymo to really open up a new market, light up a new market, without having a single person from Waymo needing to step foot into those markets themselves. We've got a brand new partner here at Twist. And as fate would have it, we love and use their product. If you need to hire, manage, pay, or equip your team members anywhere around the world, you need deal, D-E-E-L.

They're going to take care of all the annoying human resources tasks that you don't have time for, payroll, compliance, visas, and onboarding. so you can stay focused on achieving product market fit or scaling your business or finding more talent. Deal scales with you. They do all your chores, all the hard work, and they do it perfectly from the first hire on. So there is never a need to switch platforms or transition to a new system in the future. It's future proof. And with deal, you can set a payroll for any country in just minutes and get all the complicated visas and paperwork settled right away, allowing your business to grow without borders. And that's the smart way to grow now. Talent is not limited to San Francisco,

the United States, or this continent. It's everywhere. That's why more than 37,000 startups and fast-moving companies are already using Deal to accelerate their hiring and growth. Find out more by visiting deal.com slash twist. That's D-E-E-L.com slash twist. Also, I happen to be a shareholder in the company. I got very lucky they acquired one of my startups, and I am stoked. It's an amazing company. I want to double click on one thing you said, which is having multiple facilities to maintain self-driving fleets right now on a per market basis. Given that we're usually talking about double digits or triple digits numbers of cars in an individual market from an individual provider, I'm a little surprised you need that much square footage on the ground. Does that

make it harder to aggressively roll out AVs because the maintenance and care and parking and so forth takes up that much space? It's absolutely one of the rate limiters to growing fleet in any particular market. And I think one of the biggest challenges is how do you provision the right amount of power into these facilities, either for charging or for other purposes? So you're looking at, you know, these are facilities with, I would say anywhere from three, five, ten megawatts of power. And getting that provision from the local utilities can take anywhere from 12,

18 plus months depending on the city. So as you think about the rate limiters of growth in autonomy, the infrastructure, provisioning the power, getting the right locations, locations that are close to end demand. That is absolutely one of the key challenges for autonomy. Ben, a couple questions from the audience about what you're building. Seems that the consumers watching the show are very curious about

when all the stuff we're talking about is going to impact their daily life. So first question, a little bit generic, but roll with me here. When will we be able to order coffees from AVs, just general commerce on the road? Super excited about this and trying to help build it. So talk to the people. The first thing I'll say is autonomous commerce and auto lane helping to build this platform. This is not the first time this has been tried or piloted. So even last year, DoorDash and Waymo worked together in Phoenix on some things.

Before that, we had Nero, Neuro built an entire multi-billion-dollar business, making purpose-built delivery vehicles, partnering with people like, or companies like Kroger and Walmart and others. You know, so there's been iterations of this, of course, for quite some time. And the cleanest thing to point to right now is the sidewalk bots. You know, these are, these are a form of autonomous commerce that are happening. every day in many major cities in the United States, Miami, you know, Los Angeles and others. So, you know, autonomous commerce is not a brand new idea necessarily, but the way in which we think about it is, I think, different. It's more akin to the Uber strategy, which is to say,

we believe that robotics and largely speaking, autonomous vehicles are going to be the next delivery form factor. And I don't know if that's necessarily even debatable. It's not going to be 100% of all delivery demand, but it's going to be, I think, the majority over time. And so what does that look like? Well, I believe absolutely that it's going to be multimodal. So you're talking about drones. You're seeing drones, drone delivery ramping up quite fast with Zipline and Wing and Walmart and Amazon. That was a very, very enthusiastic conversation. Actually, maybe quite optimistic that's going to happen sooner than we think, finally. Yeah, yeah.

But, you know, that form factor has its significant limitations as well. And so do the sidewalk bots. And so do purpose-built vehicles. Like, I don't know if you've seen the DoorDash dot that they've created, that they just launched in Fremont last week, I believe. But that's a purpose-built autonomous commerce delivery vehicle. You know, that's faster than a sidewalk bot that can go. on roads, but also smaller than a car and not flying.

Here's what it looks like, by the way. Show the audience. I think it wants to see it. Just imagine a meatball with stroller wheels and a terrible Honda Civic spoiler all put together. Yeah. It's cute. It's definitely interesting. And I think, you know, we're starting to see all these form factors.

I don't know if you've also seen there's a company called River that's doing doorstop, autonomous delivery with these wheeled sidewalk bots basically that are like half robot dog half sidewalk bot and they can yeah i've seen that up staircases and things like that so you've got like all of these form factors happening uh which which i believe encompasses a new type of commerce called autonomous commerce um yeah exactly those and those are those are live in a couple cities as well so I think it remains to be seen which form factor is going to be dominant, but Auto Lane's purpose here is to orchestrate autonomous commerce. So we aren't necessarily making, we aren't making a bet on one form factor or another. We are simply stating that all of these

form factors and others that we can't yet imagine on this call today will come to fruition and will create an ecosystem of delivery options that all function in a very similar way, meaning they don't have drivers. None of those that I just mentioned, they're all autonomous. They're all driving with nobody inside of them. You can't really take them over easily. And so these are autonomous delivery mechanisms that if, let's just say, imagine a world in three years where all of those form factors I mentioned show up to a Walmart.

And I didn't even mention the other one, which is autonomous trucks that's also taking off. Right. So you've got, Ben, why is it called? Why is it called auto lane then? Why isn't it called multimodal lane? So I feel like it's some pretty false advertising because I, you know, I've always thought of you most as a place where self-driven cars can arrive to pick up people, drop off things, essentially, you know, creating the, I guess, orchestration is the word of the year.

The orchestration for that, but it sounds much more like you're thinking about pretty much an all the above strategy for what gets delivered. I certainly think that autonomous vehicles, cars are the form factor that's going to win. amongst all of these, but I, but I wouldn't go so far as to say that there isn't a use case for every single one of the form factors that I mentioned previously. I think it's, it's more of an ecosystem of solutions and depending on the need for the delivery, you're going to get one of these options. But, but absolutely, you kind of need to work backwards from the actual retailer itself

and say, put yourself in the shoes of Walmart and then say, I've got three drone companies coming in. I've got two autonomous trucks coming in. I've got two autonomous passenger cars, Tesla and Waymo. Then I've got maybe humanoids and robot dogs and all the sidewalk bots all coming to get different Walmart orders. That is either going to equate to 12 different different integrations with 12 different companies or it's going to result in some type of universal orchestration layer. That's what Auto Lane is. And so when I think of the word auto-representing autonomous. Oh, that, okay.

That's very reasonable. But yeah, I mean, I think for sure the focus that we have is absolutely on autonomous passenger cars, Tesla's namely, because if you look at all of those, I'll just kind of finish my thoughts here on these form factors. But like each one of those has significant drawbacks, either regular. regulatory-wise, in the case of drones, operational-wise, in the case of sidewalk plots, in my opinion. So you've got significant drawbacks that equate to none of those options serving everybody. What is the form factor that hypothetically could serve everybody?

It's a passenger car. Why? That already happens today. That's how we serve everybody on DoorDash and overeats, and Amazon is with passenger cars. and trucks. So my take on it is that is the form factor that's gonna take the lion share of autonomous commerce deliveries.

And that's what we're absolutely focused on. But understanding that it's not a binary industry. Yeah. Nathan, we talked about your company and helping people de-risk the AV world. Do you guys also work with smaller self-driven vehicles? like we're discussing here, or is it only on the Robo Taxi-ish side of things? Yeah, historically and currently, we typically work with passenger vehicle companies,

autonomous trucking companies, and then we have a very large defense business. So I spent a lot of time on the defense side working with how they're thinking about deploying autonomous systems at scale. Which is, I'm sure, a pretty big topic just given the expansion of autonomous warfare in both Europe and the Middle East. Probably not a topic for today, but I take it, Edge case is a little bit busy on that front.

Yeah. Yeah, we have a lot of, a lot of kind of pressing needs by our customers, and we're happy to support them. I'm just, I'm curious about how we're going to have all these different autonomous systems integrate because we're talking about the number of different providers on the Robotaxy site, you know,

everywhere from people in China to Europe, to the United States, lots of companies, lots of ideas, lots of different foreign factors. You know, Benj had told us about the different smaller devices that are also going to be kind of at play here. Do we have like the right standards built like, I don't know, car to car Bluetooth or whatever it's supposed to be to ensure that these technologies can interact with one another and therefore not have a large AV truck, stomp my DoorDash dot robot that has my lunch in it? Because to me, like, I feel like people

are building individual things, but maybe not mesh them together, Nathan, as much as they they should be. Yeah, you said two things that are really interesting. One was the standards comment, and then the other was how the autonomous systems will interact with one another. On the standards, we actually worked, you know, with one of our passenger vehicle partners doing a standards assessment, thinking through like, what are all of the standards out there that they need to make sure they're in compliance to?

And there's like over 200. And so then the question is, okay, how many of those need to inform their design? and then eventually in operations. And so that is all part of the work that we do with our partners to think through. Like, how should you be thinking about standards? So historic, I won't bore you to death with a bunch of standards, but like ISO. You can't bore me. I love this.

Okay. So ISO 21448, it's safety of the intended function. That's a pretty common one. You have UL 4600 and 26262. Those are kind of the kind of the, kind of the, exemplars, the primary standards that a lot of autonomous vehicle companies use and talk through. As far as how the autonomous systems will interact with one another, it's an interesting problem, candidly, I haven't thought a ton about it. My expectation would be is that the perception system

will see that as an object and it will detect the object and it will perform based on how the perception system trains the system perform for, you know, a small, you know, dot robot that's, you know, either in the middle of a road in a crosswalk or on the side of a street. I think there is a very good argument that you do not want AVs to directly talk to other AVs on the road because when they start talking to each other, that creates a dependency. And the more dependencies there are, the less reliable the overall system, you know, could become. Can I interrogate that? Because that to me sounds like saying having a backup parachute is a bad idea because it builds dependencies. Like I mean, what's wrong with having a

back channel as a last ditch, say? Ideally, you want every vehicle to be completely self-sufficient and able to sense and understand the world around itself in as complete a manner. I think over time there may be, you know, AV to AV communications, but that shouldn't be a reliant or a condition, if you will, when you think about safety. Ben, just one more question for you on the autonomous commerce side of things. I can track how quickly Waymo is growing because they release numbers, paid trips per week, and they're shooting for a million this year by the end of the year. I have a much smaller aperture or visibility into what's going on with autonomous commerce.

Can you describe the state of the market for? us, demand, number of companies involved, consumer attraction, just kind of define that for us and let us know how quickly it's growing. I'm very curious. That's a good question. I've actually admittedly not looked into kind of the aggregate because each one of these industries has a couple startups, namely in the drone space, you got wing and zip line and they don't, you'd have to cobble together each one of their private reporting information,

which is all at different timelines. So almost all these companies in autonomous commerce are private. I think besides serve robotics is probably the only publicly traded one. And so I think it's really early days for that reason. It's not something that you can just, you know, cobble together public information from, you know, annual reports or 10Ks and figure out what's going on. you have to have kind of deep in industry insider connections to kind of just get a grasp on

who's doing what, you know. And so any information I have on that is not something I would want to share in public, but, you know, but I think you can, you can get a good sense of what's happening by looking at those individual companies, you know, Coco in the sidewalk block world and serve. and if you follow each one of them, you get a taste of what they're up to. And I would say the overall summary for the industry would be all of them are growing, all of them are expanding, all of them are increasing their ambitions. And so there's progress being made in each one of these small privately held companies,

startups. But I think in aggregate, it's still a very, very small industry, arguably, you know, almost doesn't exist. It's not, it's not taking a, any kind of notable market share from Uber Eats or DoorDash, I believe. So yeah, I read serve robotics earnings every quarter because they're public. And it's always like, we're going from 50 to 100. And I'm like, you get them. And then I look at their market cabinet is like $48. But I love that they're public. I get to see what's going on. Not everyone needs to be, you know, be open AI. They're, they're, they're fantastic operators. But I, but I do think, uh, I,

do you think we're incredibly early days on the commerce side? The ride-hailing side is, of course, quite mature. I mean, you can get those numbers. As you said, there's also something called RobotaxiTracker.com, which is run by a really sharp kid out of UT Austin, and his name is Ethan McKenna. And he's actually got a nice Robotaxie internship that he announced on X the other day, so he'll be joining Tesla. But we, We work with him and he's kind of producing the, I would say, industry tracker for all of what's going on. So you can see how many Tesla Robotexies are out there amongst Austin and SFB area.

You can see Waymo growth. You can see the ODDs, as we talked about earlier, the geo-fenced service areas. Everything you want to know about the industry on autonomous ride ailing is pretty much on that website. Or if you want, you can go look and see everyone's betting. on. Yeah, I'm tracking this because I would like to see Waymo Tesla competition be as, as, as, hardcore as possible, because that'll lead to lower prices, faster out, you know, roll out and so forth. But people are really bummed out about like the chances of Tesla actually expanding into Cali, like the first half of this year. And I don't know. I was surprised by this level of pessimism.

I thought they were going to be able to go faster. And that's a relatively small market over an implying market. But I just thought they would be already kind of trying to take. take away the Waymo Crown in San Francisco been? And yet... Yeah, so did I, actually. But when I put myself in the shoes of the executive team there, especially the team working on Robotaxi, you're talking about the introduction of a technology that's going to completely transform the world, the economy and society. So if you hold such a powerful technology in your pocket that you're internally testing and in some cases limited publicly testing in Austin. It makes total sense to me why the rollout is slower than anticipated.

I do think they should be matching their comms to those internal desires because clearly they don't actually intend to or they're hitting some type of roadblock. They don't intend to go as fast is what their public statements have laid out. But, you know, if you were to nitpick that and say, well, this is disappointing, and I feel it's a little bit disappointing as well. But if you put yourself in their shoes, I mean, this is not something to take lightly. And if you look back on the cruise thing, all of us on this call on the operator's side would be dramatically set back if Tesla were.

to have any significant issue when it comes to autonomy in particular. Waymo is well. Wouldn't Nathan then get like 10 times as much business? No. The industry would be paused for a very significant time and it would be a winter for this industry. So when I look at it, one cruise level crisis could do that still today, even with all the momentum.

I thought things were much less fragile than they were back then just because. I'll let Mick, Ming answer that. But I would say if Tesla had one cruise level incident, it would be a winter for 12 to 18 months. Meng, what's your take on that estimate? Let's hope that never happens. I think the good thing in this industry is we have players that are very conservative about how they approach the regulators and how they approach the public. And so I think we are in that public trust-building phase of the industry. That blows my mind.

Have you ever driven on a road with other people, like humans? Have you seen what they do? I just don't understand why we hold robotaxies to the standard of they have to be literally Jesus Christ behind the wheel. It's that or it's your 15-year-old cousin with a permit. I completely agree. But you got to, I mean, once you start looking at this from a societal level, which is what has to be done in this industry now

because we're not talking about, does this technology work or do people want to use it? Both of those answers are yes, very loudly yes. Then the next question to answer is, well, how is society going to adopt this, adapt to this? Those are really complicated questions that even us on this call are not able to kind of parse

because every city, if you've been tracking kind of the response of a place like Boston or a place like New York or a place like Seattle, Portland, Oregon, very anti-automation, anti-AI, anti-autonomous vehicles, anti-Big Tech. So we're about to enter into this kind of societal phase, I would call it, where it's less about does this technology work, is it safe? Do people want to use it? Is it cheaper?

Is it a good technology? Of course, those things are all yes. but whether or not people, humans, want this at scale at this moment in time is a huge question mark. And it's not a resounding yes across the country. And it's not a resounding yes, even across major metropolitan areas in the United States. So that part of it to me is like the crux of how fast this industry is going to move or not is public acceptance and public perception.

And so if you look at what's going on, if you look at the congressional testimony that the Waymo safety lead had to give and talked about the remote operators and all the flak that they caught for that. And politicians who are against this and citizens who are against this are looking for any small slip up that they can point to to can this. and I don't know if you ever you saw the uh I'm forgetting the name of this cat but a cat got hit by oh yeah we covered that on the show yeah yeah yeah I mean like San Francisco tried to ban Waymos over that there was like an entire army of people in one part of the mission but those those people banded together and did a whole PR campaign political campaign to try to get this and it got enough news that like that's just a cat like RIPs it was called Kit Kat kit Kat. So one way

But like if you take that as a flashpoint, then as an example of how hot button this topic is, one cat. One cat can change the entire perception of this technology, which is wild to me. I want to, I want to, well, here's the local coverage from Mission Local. I used to drink right around that area, so I'm pretty familiar with it. RIP, Kit Kat, and if anyone has a pet, they know how much this matters. I don't mean to make a light of it. Either do I. But I just rank, I just rank cats below humans.

And so when we're talking about overall safety, that's where my focus is. Ming, you were at Grab during what we might call like the late ride-hailing boom years, if that makes sense. You were there from like 16 through 20, 2023, I think. Yeah, that's about right. It certainly did not feel like the tailwind, but certainly it was well on its way. Yeah. I cover Uber.

Not trying to say that I'm cool, but I got to go to like the Uber launch. party in Chicago in like 2009 or something. So I was early to seeing this happen. That is cool. Thank you. It was fun to be young, let me say. But it seemed that the level of pushback to ride hailing here in the States at least was much more vocal and virulent than what we're seeing with autonomous vehicles and self-driving. I'm just curious if you gave kind of your more global perspective, given that grab is, you know, based in Asia, about. how the autonomous vehicle pushback from people compares to what we saw with ride hailing in the preceding kind of generation of this area of technology?

So it is a very, it's a very interesting question. And I think I might step back before we even talk about pushback and autonomy. I think one of the conundrums that we all faced with the right sharing is that, you know, on the one hand, it's absolutely created a tremendous amount of value, a tremendous amount of good, you know, around the world, no question about it, better accessibility, affordability, mobility. On the other hand, though, the fundamental CVP for right sharing hasn't evolved. The customer value proposition hasn't changed. I would argue the experience of opening up an app ordering a car is largely exactly the same as it was five years ago. There's more categories of cars, but by and large, it's the exact same experience.

The only thing that has changed is pricing. Arguably, it's one of the few sectors in technology that's been priced inflationary because of rising labor costs. And you compare that, of course, with most sectors of technology, which becomes lower costs, more efficient over time. So at Grab, we looked at autonomy as a way to fundamentally change that. And of course, Singapore, as you mentioned, is a great market for autonomy. Everyone follows the rules. Trip pricing is sufficiently robust to cover the cost of the vehicles. And so everything does make a lot of sense.

Where we, and you know, back to the safety point, prior to Waymo's, I would drive my kids to after-school activities probably four or five times a week. Now I send my kids the high school in the Waymo. You live in the Bay Area for context. Yes, in the Bay Area in Palo Alto. And so now my kids are a little bit, find a little bit of cringe, you know, coming up to school in a Waymo. But I feel fundamentally safer having my kids in Waymo. Whoa, you can't drop cultural pop culture knowledge like that and not tell us about that.

Did you just say that Waymo's a cringe? No, my kids feel it's a little bit cringe. I know, but kids are the future. The four of us are the past tense. Why is it, I would have thought arriving at school without your parents would be fundamentally cooler because I remember my mom dropping me off and wanting to die. So what am I missing here, Ming? No, the first time this happened, they took a ride in Waymo, I think they were a little bit self-conscious

and a little bit just self-conscious about the experience. Now, though, they're more than happy. They take away to and from school all the time. So I think the adoption and the acceptance is very, very high. And for me, it's saved at least four or five trips, you know, during the weekdays. So I do think autonomy is fundamentally changing the value proposition that consumers are experiencing, whether it's better safety, whether it's better privacy. And that will have an impact in how the regulators think about the adoption for AVs versus for for right-sharing.

All right, before we let you find gentlemen go, we've been really talking about mostly the United States here from a safety perspective, from a company perspective, from a, you know, self-traving provider perspective. Nathan, what can you tell us about safety testing for AVs around the world? Ben has scared me with this idea that we are one crisis away from losing a whole year's progress in the industry. Is that something that would happen globally or are the safety standards that you're helping to get people to here in the United States as stringent in markets like Mina, Europe all across Asia because self-driving is popping up pretty much everywhere there's developed economy. Yeah. What we what we see is a lot of the standards in Europe are certainly pretty, pretty well defined

and continuing to evolve. We, from our vantage point, a lot of the development, a lot of the progress is happening in the United States and some of the other markets are watching what's going on. I, we don't spend a ton of time in. Asian market. So, you know, my, my bias is probably somewhat of where we're, where we're playing. So I, we'll have to see. I don't, I don't necessarily have a strong opinion on, you know, the standards and how, how companies in Asia are testing their, their platforms today. Leave me with a little nugget about other places looking at the U.S. How far ahead of other markets

are we? And is there something that we're doing particularly well that other markets are struggling to replicate or best? So we talked a little bit about the regulatory environment and how it's a bit of a patchwork across the states. I would say in some other markets, they don't even have that. So even though like a lot of the ecosystem, I think would prefer to have a federal kind of guidance on how these systems should be defined safe

and what's acceptable, the fact that there is state laws that define that. I think is a plus. So I think it's like over 40 states currently allow operation and testing on public roads, which allows these companies that we've all been talking about to really put their platforms in play. So it's not just, you know, you're at a test range, spending countless hours. They're actually operating in a meaningful way in the markets they're trying to pursue. So there's, I think that in and of itself provides a real advantage for the companies that are operating in

us. Speaking about Congress, there's been some movement towards doing something similar to AI legislation by saying at the federal level, we'll set a standard and then states cannot supersede or get in the way of kind of the federal national standard. Do you think that's the right way to approach autonomous vehicle regulation, or are you content with states and the federal government each get any kind of play a role? I think it's interesting on how it impacts the different verticals. So if you think about robotaxies, they could probably get away state by state in some some capacity. If you think about over the road trucking, that's going to be problematic because the trucks are going to go state to state. And so you're not going to want. Exactly.

You pull interstate commerce into place. So at the federal level, it's not just a nice to have. It's actually something that's going to be required. Now, where a lot of the autonomous trucking companies are operating, it's in the Sun Belt. A lot of those states have kind of put consistent regulations in place. But I mean, I think it makes sense to have some type of federal guidance that's consistent for something as impactful as the technology we're talking about. All right, Ben, really quickly, same question to you. State level AV regs or federal standard. And then also are the U.S. regulations holding back startups like Autoland from expanding? I mean, federal regulation is absolutely necessary. There's no way this industry can move

forward in the way that it currently is situated. So with the NHTSA meetings earlier this week, I think that was a really good demonstration of sincerity from the federal government to do something on this front. So we're really hoping for some type of federal preemption or some type of federal regulation that we can build against. And I think every other AV company is also hoping for that. So I actually think, strangely enough, that even though it's patchwork and even though it's a little bit hirky jerky for the industry today and has been for the last couple years, it's still a better regulatory environment and a more open and innovative regulatory environment than most other countries.

So more work to do, but we're not super far behind other markets. If you look at Europe and even if you look at China, the way that they're, They do their licensing and data collection of autonomous vehicles. And then you look at what Europe's done, which is essentially very little outside of the UK. Just this year has started to kind of make some progress on that. You know, the U.S. is obviously the kind of leader here in pushing regulation to allow this type of innovation to occur. And I think this week's NHTSA event really shines light on that effort. So I'm proud of the work that we've done as a country on this front.

And I hope it continues because that's what allows this technology to blossom. We have to have more acceptance of it being okay to do. And then comparing that to standards that are set and that everyone has the same rules. And we can move faster. Yeah. Maybe if we had the federal standards, then we wouldn't be risking a 12 to 18 months winter in self-driving if there is a crisis because we've already put in place the rules for it. All right, Ming, last question of the day goes to you.

I'm very curious about the expansion rate of fleets. I was a little surprised to learn how few cars are part of these autonomous fleets in different cities. So two questions. One, after an operator opens up in a city, say, Austin, how quickly does their fleet grow? And by the end of this year, how many self-driving cars are operating in the United States? I wish I had a crystal ball. I think rather than talking about the growth, I would say, talk about the rate limiters to growth. And I think the largest rate limiter today is the number

of vehicles that are available for autonomy. And so we're seeing the supply chains become much more mature, both for EV batteries as well as other components are quite critical. The maturity of that supply chain will only get better as volumes continue to increase. And as scale manufacturing gets gets larger and larger, you'll see the unique economics become much better over time. So ultimately, we're waiting for the vehicles to get into market. And then the second rate limiter is around the infrastructure and being able to work with the regulators, with the utilities in order to provision the amount of power to light up every city that we operate in.

Who wakes more money in a fully developed and mature self-driving market, the fleet operator or the company that designed the software that powers the self-driving? Another great question. What I would do is I might step back a little bit. So when you think about the unit economics for ride sharing, typically that's very focused on it per ride, per ride, how much do you make and how much do you spend on incentives versus the right value.

For autonomy, we think about it very differently. We think about a five-year total cost. cost to serve. So assuming that Navy can provide services for five years, what is the total cost of owning operating and providing services to the public? And that's very unique, if you think about it, from how traditional right sharing is analyzed. So with autonomy, the key to understand is that when you look over the total five years, the cost of the vehicle is roughly somewhere between 30 to 40% of that total pie. The remaining 60% to 70% is everything that's required to get the vehicle onto the road.

So think about all of the infrastructure costs, the insurance costs, maintenance, charging, cleaning. Now, the cost for the vehicles, of course, will decline over time as we get scale economies. But what the industry really needs to focus on is really laser honing on that 60% the 70% of the total TCS in order to make AVs commercially successful. TCS is total cost of service? Total cost of serve. That's right. All right. Well, that's a really polite way of saying you're not going to answer the question. I'm going to go ahead and say that you're going to make as much money as Waymo because I bet you

there's a lot of work that you're going to do that has nice margins attached to it. Anyways, my friends, this has been so much fun. I have to let you go. I could pest youth questions all day long. But Ben from Auto Lane, thank you. Ming from Move. Nathan from Edgecase. have been real sports i appreciate it uh just quickly where can people find your companies online and is there a job you're looking to hire for then you want to shout out into the audience here and uh we'll go around here so nathan you're first yeah uh edgecase dot a i um we're building out our platform and looking for four deployed engineers so uh anyone that wants to get

close to customers and build out the platform to help solve their problems reach out awesome thank you ben you're next uh yeah our website is go auto lane.com and we're hiring for all roles at this point. So if you're talented and want to be a part of building autonomous commerce, hit me up either on LinkedIn or Ben at goaudelaine.com. Thank you. And Ming, last plug goes to you, my friend. Thank you.

Move.io, 2Os. We're hiring across the board in almost every single function. So if you love operations and scaling autonomy, then please hit us up. All right, everybody. This has been Twist. We're back on Friday. We'll see you then. Bye.