Enabling Uptime Infrastructure and Operations for Autonomous Ridehailing-- Ride AI 2026

Well, afternoon uh everyone. Uh as James said, Chris McN here, head of mobility research, Everore ISI, and it's my absolute pleasure to host this panel on AV infrastructure and uptime. Um

literally just two minutes for me singing for my supper before we get into what is going to be a great panel. I

like to say that normally when we hear about AVs, the conversation tends to quickly focus on the vehicle. We talk

about the sensors, the stack, the AI.

But I want you to imagine a world where the tech works perfectly. But an AV business could still fail if the vehicles can't get charged, can't be maintained, or can't access the right

infrastructure at the right time. In the

future, lost minutes are going to be lost revenue.

In mobility, I think the product may be the ride, but the business itself is going to run on uptime. So historically,

as we heard a lot from today, while the sexy talk may be around driving software and capital light, we are going to talk about AV picks and shovels. We're going

to talk about hard assets, the physical in physical AI of AV fleets and their charging data um and service depots.

Now, we're joined today by four leading companies who represent those critical pillars. um Mingm from move AV, Nha

pillars. um Mingm from move AV, Nha Palmer from Terowatt, Brett Howser from Voltera, and Paul Dong um from Moya. So

today we're going to ask one central question over the panel and that is as ride sharing begins to commercially scale over the next couple of years, what actually drives uptime and who's

going to own that? Um so with that, let's dive in and let's get to uh the panel. Um I think one of the best places

panel. Um I think one of the best places to maybe start is around this idea of AV infrastructure and the misperceptions.

Um so the hidden constraints on uptime.

I think what most people see is that or at least the perception out there is that once AV works the rest is just operations. Um but in reality uptime is

operations. Um but in reality uptime is going to be constrained by the things that we don't see. the charging cues, maintenance cycles, repositioning, and the cleaning. So to to the panelists to

the cleaning. So to to the panelists to to start, could we talk about some of those hidden drivers uh of of downtime?

So maybe um Ming, we could start with you and then uh uh Nha from Terowak could jump on and welcome.

Perfect. First of all, thank you for having me here. Um I'm very quickly losing my voice, so just please let me know if you can hear me and I'll just the uh the volume mic. So when I think

about uptime um I think about this at two different levels. Uh the first level is at the vehicle level and then the second is at the fleet or the market level. Now at the fleet at the vehicle

level. Now at the fleet at the vehicle level um it's just about physics. It is

how quickly can you charge a vehicle and all of the other workflows that have to occur in order to get a re ready uh vehicle out there. Whether it's

cleaning, whether it's software updates, it's managing the tooling, all of that should fit within the charging window of a car of a vehicle. Um, and obviously as

charging windows shorten or get faster either because of larger batteries or faster chargers, then all those other ancillary workflows also have to get more efficient to stay within that that

window. Now, when you then zoom out and

window. Now, when you then zoom out and look at it at a fleet level or a market level, that's where it gets really interesting. So in any in a mid-tier

interesting. So in any in a mid-tier market, we may have somewhere between four to five to six different infrastructure assets or facilities

located um throughout uh the different city. And the key to thinking through uh

city. And the key to thinking through uh uptime is how do you orchestrate this entire footprint of infrastructure to

act as one true mesh network to support autonomy. So the problems that we think

autonomy. So the problems that we think about, the problems that we're addressing are questions such as where do you route a vehicle most optimally uh when it's looking for a charge. How do

you minimize the dead miles around the overall fleet? How do you maximize or

overall fleet? How do you maximize or minimize the ETAs for our customers? And

so one mental model that you can think of is how do you how do you best load balance the total supply of electrons in a market against the total demand for

electrons in the form of rides uh as that changes during peaks and off peaks um during market. And so it's that orchestration layer that we're building that software tooling that we're building which

maximizes the utilization the effectiveness of a fleet at a market level.

That makes sense. Na do you want to build on that? Yeah, absolutely. And I

love your mesh network uh description there because we think about that a lot.

Uh Terowatt, we build EVA AV charging hubs uh for fleets specifically and we think a lot about that location piece and it's really nice to say we want a mesh network. Actually delivering the

mesh network. Actually delivering the mesh network in locations can be very very difficult. uh you know, we're

very difficult. uh you know, we're looking for the right location for the customer obviously to build that mesh network for them, but there has to be power, there has to be zoning, there has to be a site that's available for us to

actually build. And so when we think

actually build. And so when we think about the hidden uh kind of factors that are there, it is finding those locations that can be incredibly difficult in our process. And so um you know, we work

process. And so um you know, we work very closely with customers who know their fleets intimately to try and understand what that network needs to look like in the future. I think that's

one of the hard things too given where we are with this industry. Many fleets

are just entering a geography for the very first time. They're looking for their first location. But when we're thinking about it, we're thinking about what is the second, third, fourth, and fifth location going to be. And when we

think about sighting our sites, we want to have that understanding from day one.

So, it's something that uh you have to be planful even if it's something that might be five years out.

Well, let's build on that because I think the the second concept I wanted to talk about was this idea of speed, right? So, deployment velocity for this

right? So, deployment velocity for this AV infrastructure and what could be some of the bottlenecks and so and and NA thank you for sort of touching that because I think you know it's not just

going to be location. We know real world deployment is going to sort of often lag ambition. So um maybe talking about um

ambition. So um maybe talking about um the issues whether it's power, permitting, real estate, capital, what are some of these bottlenecks and how do we think about realistic timelines of

for example depot and and and charging uh deployment? Uh Brett maybe you want

uh deployment? Uh Brett maybe you want to take that and then and then Paul you can follow on.

Sure. So kind of like Nha and Terat Voltera is a builder owner and operator of charging depots for autonomous fleets. We've got um like 10 depots

fleets. We've got um like 10 depots today across the US about 400 stalls should have up to 650 stalls by the end of the year. Um 10 depots they all kind

of look the same but none of them were actually the same as we in terms of development. You know every jurisdiction

development. You know every jurisdiction has a different way of permitting. Um

utilities are very different to work with. In the city of LA we have a couple

with. In the city of LA we have a couple of sites. If I have a site on one side

of sites. If I have a site on one side of the street it could be LWP. The other

site could be Southern California Edison. And there are two distinct

Edison. And there are two distinct processes. So the key is being able to

processes. So the key is being able to get into these markets, understanding the nuances of the zoning because by right most places you're actually not allowed to do EV charging by right, right? You've got to find an alternative

right? You've got to find an alternative use. So being able to get in,

use. So being able to get in, understanding the uh the land use, the the pathway to power and being able to deliver that to a customer like 12 to 16 months from when they've contracted is

critical for the success of the industry. And are just a follow on to

industry. And are just a follow on to that. Are you seeing that it's mostly

that. Are you seeing that it's mostly green field or should we think about a combination of green field and brownfield for for sites?

Well, I think look, it'd be great if you could do a brownfield development. I

mean, when you people think about putting charging deep, you know, deep in the ground, they think I've just got to pave a park repave a parking lot and then put some charging stations in. But

when you start adding the complexity of data offload, of buildings, of security, all that stuff, it adds complexity. And

these uh building departments have not dealt with this before. They don't know how to zone and permit these things. So,

there's a lot of education that has to happen, a lot of handholding.

And then you might in one city, you might do a great job of getting one group of people to understand and then the next project goes to somebody else and you've got to educate the same group or the different people within the same group all over again

and very much a city by city approach.

Paul.

Yeah, I mean everything that you guys have said is I mean a bottleneck to a certain degree and you know for us at Moya America and real quick for Moya

America we rebranded. We're Volkswagen

ADMT and we just rebranded as a new company well an existing company called Moya America and Moya operating in Hamburg uh Germany right now. But I

think when it comes to like operations as a whole, for us as an organization, what we're trying to do is, you know, we have a vehicle that's purposely built

for ride hailing, which is currently the ID buzz. Offloading that, what we also

ID buzz. Offloading that, what we also offer is this service as far as an experience goes that we do operate operator enablement, we do um remote

assistance, we also do a lot of training as well. But when it comes down to the

as well. But when it comes down to the operations, you know, I think that what you were talking about is like there's a lot of, you know, permitting issues that

we run into. And when we start looking at different cities, the urban course especially, where are we going to go to because we're landlocked to a certain degree. We have existing buildings that

degree. We have existing buildings that are there. And at the same time, 50

are there. And at the same time, 50 years ago, 70 years ago, 100 years ago, when these cities were built, the power wasn't coming in those cities as we're seeing power is needed today in in

today's world. So for us, when it comes

today's world. So for us, when it comes to locations, we have to basically use the existing infrastructure that's here and it goes back to the permitting as

far as additional power that we may need for our hubs. So I totally agree with you guys as far as that being something real. Another bottleneck when it comes

real. Another bottleneck when it comes to operations and I'll get real real is when it comes to the technical assistance when it comes to AV units when you're trying to hire real world

people because what we're trying to do right is at the end of the day with level four and level five is we're we're removing the driver.

Yeah.

But operations it's a very you need humans when it comes to operations. And

for us, I think that one of the big things is when it comes to humans is trying to find the technical knowledge of people that can work on these vehicles when it comes to preventive maintenance because that's something

that can completely hold up your operations. So, when it comes to the

operations. So, when it comes to the technical people as far as trying to find those technical technicians that can work on AV technology, it's pretty difficult out there right now. You have

dealerships that are trying to compete against trying to for EV electricians, people that can work on vehicles. So for

us, I mean, I think that that is something from an operational perspective we're looking at. Um, but I think that I could probably talk another 3 minutes on some operational some real world operational stuff that we run into.

Absolutely. On the on on the bottlenecks if we then maybe let's talk from bottlenecks to sort of throughput and and and uptime. I mean, I think there's not much of an argument that, you know,

the AV world is going to be EVs, right?

So let's discuss maybe charging as a as a throughput problem. Um I think charging is often discussed in megawws but in fleet ops it's really going to be about throughput. How many vehicles can

about throughput. How many vehicles can we cycle per hour? Um can we talk about how your companies are are tackling this constraint? Niha maybe to you and then

constraint? Niha maybe to you and then and then Brett to to follow.

So uptime is table stakes right and uptime can define a whole host of activities. Generally for charging we're

activities. Generally for charging we're thinking about the uptime of the charger and again that has to be table stakes for any kind of fleet. they're not going to be able to run their business if they're dealing with low uptime. So from

our perspective, that is just what we deliver. It's the core of what we

deliver. It's the core of what we deliver. It's like a non-negotiable. The

deliver. It's like a non-negotiable. The

other things that are so important and it makes some of the site selection even more difficult is thinking about things like flow. Uh do you have multiple

like flow. Uh do you have multiple ingress egress points? Are you on a busy street? Are you facing an alley? Are you

street? Are you facing an alley? Are you

with a dead end? All of these different design decisions are so critical to a fleet being efficient uh and really driving that throughput time. Uh if

they're cleaning and if they're doing other activities there, can they happen simultaneously to the charging? The

chargers need to be in the slots need to be spaced out to allow that. Um there

are a lot of things that go into the details of the design that will drive that throughput in a really critical way and really align with the operators, right? training people to be efficient

right? training people to be efficient with the spaces that we have. Um it's

really kind of thinking of that whole stack when it comes to throughput and our part in this is making sure we design the site for the maximum efficiency.

Absolutely.

Yeah, she stuck the landing. That's it.

The efficiency with, you know, queuing um throughput. I mean, so you've got to

um throughput. I mean, so you've got to have the right power not only for the vehicles that are coming today, but the vehicles that'll be coming tomorrow, right? I mean, when these companies and

right? I mean, when these companies and and Ming can probably talk this better than I can, but when they're looking at, you know, trying to get, you know, five chargers or five vehicles per charger, that five is going up to eight or 10.

So, the throughput there is going to be really efficient. And then, so you're

really efficient. And then, so you're going to get these chargers 75% plus utilization and you have to have 98% uptime. So, failure is not an option. So

uptime. So, failure is not an option. So

making sure that you have the right systems in place, the right people in place to make sure that our customers can operate on their sites is she said table stakes.

Yeah.

I'll add one more thing. You know, these are hard to find sites. And so to the extent they're able to serve more vehicles with the same site, that's the greatest efficiency that can be found anywhere. And so we are working really

anywhere. And so we are working really closely with customers to say how do you get that from 500 to a thousand to to maybe even more with higher power.

Yeah. N I think you just gave me PTSD talking about flow. Bring me back to business school and managerroio economics. So I I I I I do understand

economics. So I I I I I do understand that that math problem. Um maybe if we start to think about a little bit, you know, f further out um you know to the

idea of what sort of nextgen means for AV and and and and charging depots. So

looking ahead, does the AV infrastructure frontier become the automation of these depots themselves, whether that's, you know, robotic cleaning or um automated charging um and and definitely in this question of big

data analytics, you know, the role of of of AI. So um maybe we can start with uh

of AI. So um maybe we can start with uh uh Paul on that and and then move, you know, no pun intended, uh to uh uh to Ming to talk about this future state.

So what we're talking about is the future, right? And we're talking about

future, right? And we're talking about the future as far as what this could look like coming up when it comes to operations pretty much. And so for us it

continues how we operate is you know we're enabling our operator to really handle a lot of those type services based on what our knowledge is and is

based on you know our experience as far as what the future will foresee is you know are we looking at you know almost like a manufacturing line when you come into the depot where you have your

cleaning services at the beginning you have your charging your wireless charging underneath the vehicle when you're looking and cleaning the LAR equipment or your different sensors. Is

it kind of like a manufacturing kind of through throughput?

You know, is this what the future is?

And I think that for right now as again we're in the infancy when it comes to what these vehicles can do from operational perspective is that in the next 15 well really in the next three to

five years what is this really going to look like as we start scaling to even more to thousands of vehicles that are in each city with different providers.

So, you know, at the end, you know, what this is going to look like is, you know, something that we didn't anticipate four or five years ago, but coming up here in the next two to three years is going to be pretty much like a pretty good ride

here.

Yeah. Because I mean, obviously, we talk about this idea of automotive grade, right? Also, you know, think about a

right? Also, you know, think about a parent like VW like something, you know, vehicles are supposed to last for 20 years. We don't really have that

years. We don't really have that history, right? Particularly on on LAR

history, right? Particularly on on LAR systems, new new sensors. So I do think preventive maintenance is completely like what we consider preventive maintenance right is traditional

vehicles right every 3500 miles you know you go get your oil change you know from a tires perspective every 40,000 miles you know when it comes to the different kind of applications or categories that

are out there we're talking TNC's right now Ubers and the lifts but there's a whole other category of you know closed communities as far as retirement communities as far as schools and

universities that are going the use cases are going to be so different than what we had prepared for preventive maintenance. So I think when it comes to

maintenance. So I think when it comes to the stacks and the technology stacks that you have that additional preventive maintenance that you know is something that wasn't foreseen 15 years ago but I think moving forward that's something

that we're going to have to address coming up.

Absolutely.

Couldn't agree more everything that Paul just mentioned um at the depot level automation in robotics is just a given.

it's table stakes and that will get done. Um but when you're operating a

done. Um but when you're operating a market at scale um what we think about for next generation depots is really

around multi- depot orchestration. Um

now you can think of these facilities as very small hyperscalers. Um the

difference obviously being in a typical hyperscalers all the GPUs are tied down locked down onto a server rack. In our

case, our GPUs are coming in and out.

They're on four wheels. They come in and out, you know, every every day. Um, but

the question is then, how do you orchestrate um a distributed footprint of infrastructure at scale? And that's the key at scale. Um, and how do you uh how do you really orchestrate the different

elements when there are different capabilities and different functions that are being um spread out. So one

analogy that we think about a lot is um similar to uh content distribution networks like a cloudflare uh or a you know their job is to get the bits and the packets as quickly as possible to

the edge of the network to the demand as quickly as possible in the most efficient way you know possible. We do

the same thing except we're not delivering packets we're delivering vehicles to the edge to the demand as quickly as possible. So how do you build the CDN for physical autonomy and for

EVs? That's the future of depots

EVs? That's the future of depots and and nah or Brett did you want to expand upon you know some of the ideas of nextgen in in in charging when we see you know some of the plans that like for

example Tesla has talked about robotic arms wireless charging you know what what do you you think is is specifically for that vertical you know some of the things that we should be looking out for

I think Ming said it's table stakes right it is pretty amazing you see these vehicles out there driving themselves and then they come into the depot and you have a human charging uh plugging in the charging and the technology is

there. Um there are many different

there. Um there are many different applications that can be used already.

Obviously generally robotics is getting better and better and so it's not very complicated to open the flap and then plug it in and so certainly that'll be something that we will see pretty

ubiquitous across uh I think fairly soon. Yeah, I'd say the initial like the

soon. Yeah, I'd say the initial like the three legs, the initial stool, it's it's uh charging, it's inspection, um and it's um cleaning. If we can do those

three things and make them autonomous, I think we have a flywheel effect and we create a significant efficiencies at the depot. And so I know for us, we're

depot. And so I know for us, we're already integrating those technologies into our systems today. Trying to figure out best of breed technologies. We're

not trying to create those technologies ourselves. find the best ones out there,

ourselves. find the best ones out there, integrate them into our system so that when our customers are using the site, they have a single pane of glass or single source of truth where they can get access to all that information.

Excellent. Um, and then maybe in the last last couple of minutes, I think what was great about sitting here over the course of the day is I thought the topic of this panel came up in so many of the other uh presenters, you know,

those who are talking about the software, the commercialization is hard, you know, essentially is the keystone of of of what you guys do. And so maybe in the last couple of minutes we could just

go down the line and just expand it you know maybe under a minute or so um what you see as the remaining bottlenecks.

you talked about the bottlenecks in in the infrastructure deployment. Um but it would love to understand if if what you are seeing from your customers you know are there other things meaning let's use

some examples AV software better infrastructure is essentially what what this panel is about better operations coming from you know the the Whimos and the the Ubers of this world or clearer

regulations that city bycity approach will that be aided by by federal but you know I think given that you sit at this bottleneck would be great to hear um if we just went down by the line what do you think is is holding up the industry

from truly scaling.

Oh, number one is operations and that's self- serving obviously.

But uh so today we operate um the Whimo fleet in Phoenix um you know the center of semiconductor fabs um the oldest and the first market for Whimo in the US. uh

we launched uh Miami uh in January, the one of the largest markets on the east coast and we'll be launching London uh towards the back half of this year which is the largest market uh in Europe and

um the the mental model that we have is unlike ride sharing human right sharing the key to really understanding is the is the total cost to provide ride

sharing services over the life of the vehicle. So call it five years.

vehicle. So call it five years.

And when you interrogate that fiveyear TCS, what you find is the capex of that that vehicle which is really encompassing all the R&D system software

etc. that vehicle cost is only somewhere around 20 30 40% of the total TCS. So

the vast majority of that TCS is around the operations, infrastructure, insurance technicians mechanics all of that. And unless you get that 60 70

of that. And unless you get that 60 70 80% um done in the right way, then the overall industry isn't necessarily sustainable. Makes

sustainable. Makes sense, Paul.

Yeah. I just think that the the continuation of expansion as the fleets become bigger that we're going to need to have more hubs in in city cores at

this point because it just feels that a lot of our hubs are on the outskirts of the downtown areas. So, when we talk about dead miles and we talk about utilization, in my world, utilization is

like the big word. Um, is that you you you want to be as close to 100% utilization as possible. The reality is is that's not real. But the thing is is

that if we continue to create more hubs, we have to create these satellite hubs in cities where we can quickly go and charge or quickly do the cleaning that for a vehicle that had a ton of bird

poop that was on there and we got to quickly get on that vehicle or someone that just got out of the vehicle and they had onion rings or they just got done vaping in there, which are the realities of,

you know, commercial um scalability are what people actually do to our cars when they're in there. So when it comes to locations, you know, instead of having hubs that are completely outside the cores of where we're operating, trying

to figure out the inner city locations that we can get these satellite locations in, which means that we need your guys' help to get us the charging and get us set up pretty well in those areas. So

areas. So yeah, so my life before Terowatt was data centers and we lived in what we call the 59's world. And I think that the SLAs's are so important, but right

now we're still in a world of disparate SLAs's. You might have one SLA for the

SLAs's. You might have one SLA for the charger uptime. You may have another SLA

charger uptime. You may have another SLA for the robotic arm uptime. There may be SLAs's that operators have with their, you know, their customers that they're, you know, operating the fleets for. And

it's coordination amongst all these pieces that has to start to get incredibly seamless almost to the point of one SLA. I don't think we reach 59s, but how do you actually make that one

integrated SLA, I think, is the question at hand. And that will require massive

at hand. And that will require massive amounts of coordination between the various stakeholders.

I've got to leave us here with something very insightful, right? Since I'm ending it, you know, at the end of the day, like I to me it still comes back to this permitting, right? It's hand-to-hand

permitting, right? It's hand-to-hand combat. Every place is different. Every

combat. Every place is different. Every

city is different. Um, states have tried to do things like California has AB 1236, which by right allows you to do charging, but it was really meant for public charging. I think there's a lot

public charging. I think there's a lot of education that still has to go on about what it takes to put up a robust um charging depot for fleets and we've got to be able to do that by right so that we can streamline the process of

getting these things improved and getting shovels in the ground as quickly as possible.

Yeah, we def we definitely have to get you to my uh my hometown of of New York.

Definitely we're going to need it that that um well look um with that I think we have maximized this panel's aotted uptime. Round of applause for the panel.

uptime. Round of applause for the panel.