Third Bridge专家称Lyft自动驾驶业务未来12–24个月对平台增长贡献有限

[[00:01:33]]
Lyft completed about 237 million rides in Q1 2026. From a commercial and production perspective, what percentage of rides do you think could realistically be AVs [autonomous vehicles] in the next 12-24 months? Could that volume be sufficient for the company's ride growth YoY to be higher than the recent 8%, which is also expected for 2026?

[[00:02:20]]
This is going to be a huge range. It really depends on Lyft's ability and actually even less so on Lyft's ability, but on each of the AV partners' abilities to scale up. That range could go from 1,000 all the way to 5,000, perhaps more depending on each of these partners' ability to… Both on the technical front and also on the actual physical front of just getting physical AVs on the road. That being said, each of these AVs are going to be much higher utilised than a human driver. I think it's safe to assume that each AV can drive for 20 hours per day, and then, let's say, something realistic might be… I'm thinking maybe 20-25, perhaps up to 30 rides done per day, and so that will be much higher than an average driver.

Even if we're talking about that, we're talking about maybe 1-5 million rides per quarter, and that is a drop in the bucket compared to the quarter of a billion rides that Lyft was doing in Q1. I think from that perspective, it's likely to be an incremental number in the next 12-24 months, but that being said, if these numbers… If the deployments go well, Lyft proves itself as a capable partner for these AV players, I think it's going to help Lyft be able to further land partnerships with even more AV players. Then once you have a lot of players and then each of them is deploying 300-500 each, then we're talking much hopefully, bigger numbers over time, but I think that's my expectation in the next 12-24 months.

[[00:04:15]]
You mentioned that each AV can drive for about 20 hours and 20-30 rides per day, which is much higher than the average human-driven trip. When it comes to human-driven taxis, what number of rides per day would you expect in the next 12 months? Is it 5-10, 10-15 or 15-20?

[[00:04:55]]
There's a lot of inputs into that. The AV piece, I was mainly talking… AV, you have no reason to keep it as much utilised as you can, vs human drivers, it really comes down to per driver, how much they want to drive on Lyft or Uber on these platforms. Most drivers are not doing 12-hour shifts necessarily on either of these platforms. I think from that perspective, it's difficult to control from a per driver perspective. That being said, I do think from a utilisation perspective, if Lyft were going to have AVs on the platforms, AV would be the one that Lyft will keep highly utilised and then your drivers are the ones being a buffer. When you have less demand, you want to keep the AV utilisation high and then the driver utilisation might drop. I think from that perspective, it's hard to put an exact number on that.

[[00:05:55]]
Would you be comfortable giving a range?

[[00:06:04]]
I think perhaps half or… Half is actually pretty good. That's my guess at this point. If you have both AVs and human drivers in the platform, I think maybe half on the human driver side.

[[00:06:25]]
As AV is utilised increasingly more, would you expect that to increase the ride volume growth YoY beyond the 8% expected for 2026?

[[00:06:49]]
Yes. I do think so, and that comes from a couple of factors. One is AV expands the market for who wants to take a ride-share. Ride-share rides as a percentage of all rides that could be on ride-share is only a single percentage. In San Francisco, for example, with the entrance of Waymo, we've seen additional people start using Waymo, Lyft and Uber. Some of these are a couple of different use cases. One is a safety use case. For example, women, late at night, they would much rather prefer an autonomous vehicle from a safety perspective. Sometimes we see parents also have their teenagers inside of a Waymo and then vs a regular ride-share. Those use cases are pretty specific to, I would say, autonomous rides.

I think with the addition of autonomous, that actually expands the total market. Depending on Lyft's ability to capture that demand, I do think it could accelerate growth beyond the 8%, so that's assuming current prices. Later on, I think as AV scales up, I think it has a lot of advantages in terms of the cost side. One, obviously, you don't have the physical driver whose pay is a vast majority of the cost of a ride-share ride today. Then the insurance cost should come down significantly given that autonomous vehicles are up to 90% safer than human drivers. I do think those two pieces, in particular, will come down as autonomous vehicle scales down. Then as we know, ride-share is very price-sensitive, and so to the extent that autonomous is able to bring down the cost of a per ride-share ride, I do think that will also continue to expand the market.

[[00:08:50]]
How would you expect ride growth to trend over the next 12-24 months, if you think that potentially AV could increase that? What do you think it might look like in 12 months?

[[00:09:08]]
I think the challenge is the 12-24 months' time period. I think given that time period, depreciation and cost of these things is still going to be massive. I do think from that perspective, I don't necessarily expect huge changes on either the growth trajectory or the EBITDA trajectory in the near term. I do think possibly in the next 12 months at least, the cost of AV ride will probably be on par, perhaps a little bit more depending on where… If it's a new region, you're just getting started and then the fixed cost isn't sufficiently spread over enough rides, I do think that might actually make it worse in the near term. I think it comes down to maybe a question of when is Lyft or these partners able to get enough scale so that the cost is spread over enough rides.

[[00:10:06]]
The market is expecting ride growth to increase to 13-14% in 2027 from about 8% in 2026. How much of that do you think is down to AVs or other factors?

[[00:10:30]]
I think for Lyft in particular, if we're talking about maybe 1,000-5,000 active vehicles by that point, and that is less than 2% of the total platform value. I do think 13% is pretty unrealistic for 12 months at least. I don't know about 24, maybe at that point, the scale is larger and maybe the economics gets better faster than we expect. I think that's… I would expect that growth not to come from AVs necessarily, but from the investment in luxury rides, international growth and things outside of the US.

[[00:11:10]]
You would say that the growth that is expected to come in rides is not from AVs – it's from growth in international markets and also premium.

[[00:11:22]]
Exactly, and partnerships. That's where I would bet that most of the growth is coming from rather than from AV deployments.

[[00:11:34]]
You mentioned that the cost of an AV ride would be on par with a human-driven ride for the next 12-24 months. How long do you think it will take to get to a point where AVs are cheaper than human-driven cars in regions such as San Francisco, where this technology has perhaps got a wider uptake than in other places?

[[00:12:27]]
It comes down to a couple of things. One is how fast can each of the AV partners bring down their hardware cost. The best way to do that is you have your sensors directly built in on the OEM assembly line. There's none of this after fact retrofitting piece that's super expensive. That's one piece. Teleoperations, can you have one remote operator be operating 50-100 vehicles, and that's very few AV partners are actually being able to do that right now. Then you're cleaning and charging all of that. How can you actually spread that enough over enough vehicles to make that fixed cost go down significantly. This is really a guess from my perspective, but definitely not in the next 1-2 years, but maybe five years, plus or minus.

[[00:13:24]]
This is in markets where AV adoption is already higher than in other markets, so we're talking about somewhere such as San Francisco here.

[[00:13:31]]
Yes, San Francisco already had a few years on that. I'm thinking, for example, in terms of Nashville, if you get started in there, it probably takes another 3-5 years to get those costs down. The expansion will take some time. That's why I'm expecting on average in the US, San Francisco has already started before that.

[[00:13:59]]
Would you say that San Francisco, in the next 12 months, will be at a point where AVs are cheaper economically than human-driven cars?

[[00:14:10]]
I also doubt that. I don't think… Maybe Waymo. Waymo by themselves perhaps, I don't know how far they negotiated on the insurance side, in particular. They had some good conversations with the same insurers that everyone has. I don't know how far that conversation has gone. I do know they're bringing down the cost of their AVs pretty significantly. Maybe for Waymo in San Francisco in the next one or two years.

[[00:14:45]]
What will be the major driver of that? You mentioned insurance there. Is that the main bottleneck?

[[00:14:57]]
In terms of bringing the cost down?

[[00:14:59]]
Yes.

[[00:15:00]]
I think for Waymo, a couple of things. One is scale. Do they have sufficient scale of vehicles that the cost of operating, cleaning, charging, all of that is able to be spread over sufficient rides. Insurance for sure is a big part of that. Then also, can they switch out their Jaguar fleet to something much cheaper over time. I think they're in the process of doing that. I think those three pieces, I would say, are probably the biggest drivers there. Waymo, in particular, is different than some of these other partners that Lyft has are much earlier, like May Mobility, there's… I would not expect them to be able to get there remotely close in the next 1-2 years. That's how… I think Waymo is the leader here. San Francisco, in particular, has been the market that's been strongly saturated with Waymo. I do think San Francisco can be an early signal for other markets, but I don't think anywhere else is like this.

[[00:16:06]]
When an AV ride-hailing company launches in a European city – London or Paris, for example – how long do you think it takes to get to a point and scale where AVs might be cheaper than human drivers? Is it the same 3-5 years as you mentioned with Nashville, or is it slightly longer than that?

[[00:16:32]]
I think Europe, in particular, is going to be so difficult. The regulatory framework is very fragmented and also really strict on some of these licensing agreements. That's one thing. That's going to take years. Then even, let's say, Lyft succeeds in London, other regulators watching, but they are not necessarily bound by the same rules. They will want to do their own things, and that's going to take a long time to get the regulatory side figured out. Then also unions are super strong in Europe. I believe all of the black car or a big chunk of the black car drivers are represented by some union. I think there's going to be significant pushback to adoption of AVs in European markets. I think that's going to slow this down even more so than in the US.

[[00:17:32]]
What about in a US market? Is 3-5 years from launch roughly how long it takes to get to a point where AVs might be cheaper than human-driven cars?

[[00:17:46]]
Potentially, and that is assuming there's no… On a technical maturity side, that's assuming a lot of these partners are at the Waymo level, which might be a strong assumption for some of these players.

[[00:18:06]]
How far behind Waymo do you think other partners are? Are they 12-24 months behind, or is it much longer, such as three years behind?

[[00:18:27]]
I think some of these are at least three years behind. It really depends. I think the Chinese players, a lot of them are at the Waymo level or very closely behind it. A lot of the native American players outside of Waymo are pretty behind with the exception of maybe Zoox and Tesla, but it's unclear either of those will be interested in partnering with Lyft. I do think from that perspective, you have almost three tiers of players in the US. There's, like, Waymo, no one else is close to that. Second tier like Zoox, Tesla, probably that's it on the end of that line, and then everyone else.

[[00:19:17]]
Hypothetically, if AV deployment lowers the average fare per mile over the next 12-24 months or beyond that, and we expect fare per mile to come down as a result of AV, what impact do you think this might have on gross bookings? Would ride-hailing platforms such as Lyft be able to keep their fares at the same level, or would you expect them to reduce their prices to reflect the fact that AV is cheaper than human-driven?

[[00:20:02]]
I think it depends on the demand side more than the supply cost in terms of how each of these ride-share players might actually price AV rides. On the one hand, for certain types of customers, people who are really into AVs or business travellers who want the quiet and privacy of an AV ride, you definitely can make these, charge them as more of a premium ride, similar to maybe an Extra Comfort or a Black ride in terms of pricing vs on the other hand, if you have certain AV rides that are just ETAs or these are going to be, by definition, longer than for human drivers, and so for people who are just trying to get somewhere ASAP, AV may or may not be a good option in that case. I think price discrimination is probably going to be more interesting than just to say AVs are always higher or always lower. I think it depends on the time, the use case and then sometimes the preferences of the riders.

[[00:21:11]]
Do you think AV could have a positive or negative impact on gross bookings going forwards?

[[00:21:23]]
Definitely, I think, positive, it brings in additional use cases that previously people even haven't turned to ride-share for. I do think it grows the demand in that sense. I think the question is from an AV displacing human drivers perspective, how much of that is going to happen, and then whether Lyft is able to balance utilisation of AVs and keep it sufficiently high so that these things are almost fully utilised as much as possible. I do think it comes down to ultimately that cannibalisation and how much Lyft is able to dispatch things efficiently.

[[00:22:11]]
To what extent would you expect AV to expand the current ride-hailing market in North America? You mentioned new use cases where people are now using AV ride-hailing platforms – notably parents ordering for their teenage children or women not feeling safe at night – vs human-driven ones. By how much do you think this is expanding the market? Is it 5-10% or 20%? What expansion do you expect in the next 12-24 months or beyond?

[[00:23:08]]
San Francisco, I think the whole market expanded about 20%. San Francisco, I would say, is the best case scenario, just given how accepting of AVs this market is. I think maybe that's a top side of the estimate, so probably something like 15-20% is my expectation.

[[00:23:34]]
Is there something unique about San Francisco that means that those 15-20% market expansion figures are possible? Might it be smaller in other markets?

[[00:23:50]]
Yes, that is my expectation. I think San Francisco is obviously very tech forward in accepting of new technologies. Here, we can basically assume everyone is willing to take an AV ride, all else equal vs other places that might not be the case, that might take even more time for consumers to get used to AVs on the road. I think that's maybe one of the reasons for 20% probably being the higher side of the estimate. Then also San Francisco, the ODD is fairly easy. There's some hills, but you're not talking about unpredictable weather or anything tough like that. I think from that perspective, also, it's a little bit easier for new AVs to deploy. That's why I expect other markets to grow not quite so fast as San Francisco.

[[00:24:40]]
How long would you say it would take for typical cities, not San Francisco, to grow by 15-20%?

[[00:24:52]]
I think that's probably going back to my original estimate, five years, plus or minus from there.

[[00:25:06]]
Is 15-20% the limit, or could it go beyond that?

[[00:25:11]]
San Francisco has grown 20% without the drop in prices. That's the initial assumption because Waymo here are priced at roughly around the same price as in Extra Comfort on Lyft. If these players can bring their costs down significantly, get adoption, I do think if prices can come down half of what they currently are, then we're talking about much bigger growth. Having said that growth in terms of total number of rides. I think if you multiply, then the total bookings might not be the… Might end up being flat. I think it comes down to, is AV able to expand the market faster than the price elasticity accounts for. I do think once you account for that, still, it will probably be an expansion in the overall market and just taking share away from private vehicle ownership.

[[00:26:20]]
When you say that Waymo is priced the same as Extra Comfort on Lyft, that would perhaps put it at the more premium end of the market or perhaps mid-tier. Obviously, AV is a new thing. Lots of people are talking about it on social media and posting Instagram reels and TikTok of them in it. It's not cheaper than a human-driven car yet – at the moment, it's positioned as something premium. Some people might argue this is just a fashionable fad and people will revert to cheaper human drivers at some point. What would you say to someone that is perhaps a little more sceptical?

[[00:27:17]]
I think from what we're seeing from consumer behaviour, the people who have tried Waymo just love it, and they will continue to use it in a very sticky way. I think that's what convinces me that this is not a temporary fad thing. It really creates use cases where you didn't have any other better choice before, and so you never could have a business meeting in a vehicle while driving before. This is one of the first times you actually can do that. I do think these use cases are very real and very sticky. From what we've seen from consumer behaviour, I do think the retention on these things are super high. I do think this will just be a generally a growth story from a demand perspective.

[[00:28:04]]
How much of an issue are things such as weather and geography to the successful roll-out of AV taxi fleets in cities?

[[00:28:33]]
I think that's a pretty big determinant… Pretty big factor. One, in terms of how these AV players think about which markets they choose to go to. I think from a crawl, walk, run perspective, they will tend to start with the easier weathers… Cities with easier weather and they expand as their technology becomes more mature. That being said, I think, this comes down to how the technical side of these AV players. Waymo at this point can handle snow and all sorts of challenging weather vs some of the other players, as I mentioned, maybe the third-tier players are barely handling ODD in a clear environment. For them, it will add years for them to be able to handle snow, sleet and those kind of things.

[[00:29:31]]
Under what types of weather conditions would you say AV works the best?

[[00:29:40]]
I think San Francisco is a pretty good example. A couple of things, not just weather. One is the street layout. US cities tend to be grid layout. That's actually pretty easy, vs some of the European cities are hard for even humans to drive in. What the streets are like, very narrow and difficult to navigate. I think that's going to be much more challenging from an ODD perspective. From a weather perspective, anything that affects visibility, affects, for example, snow, ice, freezing conditions, those are pretty tough. Unless you have self-cleaning sensors or a lot of redundancy on your sensors, I think that's where the stormy weathers can make it very difficult. Even Waymo at this point is not able to handle all types of weather, it's able to handle vast majority of weather. That's another piece. Then the self-cleaning stuff, that also… I think the Waymo drivers at this point are all self-cleaning. I think Tesla, for example, if they insist on their current technology stack, I think that's also going to make them difficult to drive in inclement weather. I think it really depends on the per player, what does their tech stack look like. I think generally speaking, I would say that's maybe some of the things that affects how easy or hard it is to handle ODD.

[[00:31:22]]
What do you mean by ODD [operational design domain]?

[[00:31:27]]
ODD stands for operational design domain. It's basically a set of conditions that includes weather or road types or lighting where within an ODD, a self-driving vehicle can operate safely, outside of ODD, it cannot. For example, in San Francisco, Waymo cannot go to the airport. The airport is outside of their ODD. They started trying to work on that at the moment.

[[00:32:07]]
What proportion of the typical urban ride-hailing use cases can be performed within the ODD [operational design domain] of AVs as of 2026? The airport use case in London, for example, is quite a big one for someone such as Uber. Obviously, Lyft isn't quite here yet, but we've got Freenow, which was acquired by the company in July 2025. That might exclude 10-20% of the types of rides that people use ride-hailing platforms for.

[[00:33:04]]
I'm just looking at San Francisco, which is on the higher side of the market. Outside of the airport, the vast majority of the city is within the ODD. That's probably not super representative vs other places. If we're talking an average American city with grid layout and average weather, which includes some snow and inclement weather, probably I'm expecting 50-70%. That's mostly a guess right now.

[[00:33:44]]
In order for use cases such as airport, for example, to be fulfilled, what needs to happen, and how long until the ODD broadens out to that use case in particular?

[[00:34:01]]
Airport in particular, is very different. Airport is less about whether the technology is there, but more about showing the airport regulators that Waymos or other AVs are not going to contribute to more congestion. What airport regulators are very sensitive about is they don't want AVs or any vehicles for that matter idling next to the curb, causing a lot of congestion issues. This will require Waymo to work very closely with the airport authorities to basically prove that, "Hey, if you give us this staging lot for AVs, we will make sure we will get in and out of there and actually reduce your congestion issues." That's more about buying trust with the regulators rather than their technology is not there.

[[00:34:59]]
Does that mean there's no technological reason why AV platforms could not serve the airport market – it is a regulatory issue?

[[00:35:15]]
It's a little bit of time to get it. The regulatory lack of trust is due to issues where in a crowded environment, sometimes AVs do get stuck. I think to the extent that Waymo is able to build trust with the airport authorities and over time, I do think once they've accepted a first player in, the next time, they know exactly what questions to ask and how to get the AV players to demonstrate their technical abilities there. I think the second and third players have a much easier time at that airport. That being said, that does take a bit of time.

[[00:35:55]]
How should we think about efficiency and utilisation? My understanding is that Lyft has a hybrid dispatch model in markets where there are both human drivers and AVs. How efficient is that or could that be at keeping AV assets utilised during off-peak hours? How might poor fleet utilisation erode profitability for AV partners over the next 1-2 years?

[[00:36:31]]
I think this is one of the biggest reasons why AV partners are even interested in working with either Lyft or Uber is because of the ability to actually fulfil this fleet utilisation number. For example, Lyft is able to… During peak times, everyone is on the road, AVs, human drivers, everyone. Then during low times, you just keep the AVs utilised, and then you start asking basically keep drivers off or dispatch more rides to AVs. By doing that, you can pretty much… If you think about the wave of a demand graph, you can keep AVs fully utilised until the trough of that demand, which is quite high actually. I think from that perspective, partnering with a Lyft or Uber from an AV player's perspective is very beneficial in keeping this utilisation high vs if you think about just Waymo going by themselves during the… In the trough times, their vehicles are probably sitting pretty unutilised. That's just a huge cost on their side. I do think from this partnering perspective, it's a win for the AV partners.

[[00:37:55]]
Are you saying that fleet utilisation is not really a problem for AVs?

[[00:38:02]]
If they partner with Lyft and Uber. Lyft and Uber can absolutely make sure these vehicles are as highly utilised as possible.

[[00:38:12]]
Could there be any backlash from human drivers if it is perceived that Lyft, Uber and others are trying to utilise AVs as much as possible and perhaps prioritising AV journeys over human rider journeys? Could that impact Lyft and Uber's top or bottom line?

[[00:38:45]]
They definitely could. I think this is a pretty controversial area for drivers. That being said, as we mentioned before, some of these… AVs can't operate in certain areas or use cases. Airports, for example, you can actually have more human drivers go to airports and fulfil those rides that are pretty profitable for them as well. It's more of a, if AVs are better utilised doing short-haul rides within urban dense areas, then you want human riders doing these rides to maybe the airport, to suburban areas that may be more difficult for AVs to do for whatever reason. I think in the near term, if Lyft is able to balance this right, then you can have your cake and eat it too. That being said, if this happened very suddenly overnight and drivers saw their earnings drop by 30-40%, then that could potentially get pretty tricky in terms of how Lyft and Uber to manage. I do think if they manage this in a gradual perspective and then do it in a way that's humane to drivers and try to communicate as much as possible and some of the rationales for this and explain, again, AVs are hitting different use cases, that people wouldn't have historically come to ride-share for anyways so they can grow the pie together. I do think there's ways to manage some of this backlash so that it doesn't become a major issue for both players.

[[00:40:20]]
Is the ultimate goal to replace human drivers with AVs?

[[00:40:25]]
I don't think you can fully replace human drivers. Just given the reason, the demand is very ups and downs. During the peak times, you always want human drivers to come in and start driving, vs otherwise, during peak times for Waymo, you just have to wait 30 minutes, and so that loses a lot of trust for their customers. I do think this hybrid world is the best of both worlds for a ride-share perspective.

[[00:40:58]]
What proportion of rides do you think is ideal for human vs AV in order for the CX to remain broadly similar to what it currently is – good waiting times, being able to get anywhere you want to through one platform, etc – but also in a way that structurally improves something such as EBITDA margins for a company such as Lyft, if we do get to a point where AVs are cheaper than human drivers? What would you say is the ideal mix between human and AV, if hybrid is the way forwards?

[[00:41:50]]
This is very long down the line. We're talking about AVs that are significantly cheaper, and you can technically meet all of your demand up until the trough with AVs. That's the ideal case. Probably two-thirds AVs, where your daily demand, commute demand, all of that is mostly met. Then you have maybe 30-40% humans to meet peak demand. If you have events, extreme weather, for example, situations where AVs can't function. 30-40% with a lot of elasticity. If you can get people to come in and drive for two hours when the platform really needs it, that's a lot more useful than to have a lot of people who are just on standby for the whole day. I think from that perspective, I think both Lyft and Uber will need to invest in very flexible incentives that gets people to stop whatever they're doing to start driving for ride-share immediately.

[[00:43:02]]
Over what timeframe would you expect a scenario such as that to happen? Might that happen in the next 3-5 years, or is it going to take a much longer time?

[[00:43:15]]
Much longer time. This is probably talking about 10 years out at this point.

[[00:43:27]]
You mentioned that one of the roadblocks in Europe has been unionisation. Obviously, that culture isn't as strong in the US. Even if there was a backlash to a relatively small number of AVs coming online in the next couple of years, what could drivers actually do about it that would harm the top or bottom line of Lyft or Uber?

[[00:43:59]]
I think the example in the US. I'm thinking about is the New York market, where basically Waymo was not allowed to enter because of all the pushback on that front. I do think drivers unionised or organised can go to regulators and basically say, "Here's all the reasons why you should not allow AVs into the market," and some regulators will be open to that. I do think it comes down to how much backlash pushback there is and how well each of these companies is being able to balance both sides, allow enough opportunities for human drivers, while at the same time, also introducing AVs in a safe and cautious way.

[[00:44:46]]
Why would AV software providers or OEMs – such as Waymo, Zoox and Tesla – opt to share revenue with a company such as Lyft or Uber, rather than run their own D2C app?

[[00:45:04]]
A couple of things. I think one is the utilisation problem we talked about. With Lyft, you have ability to keep your entire fleet utilised at all times vs doing it yourself. You can either have enough vehicles to service peak time, in which case they're going to be sitting empty most of the time when it's not peak, or you just don't have enough vehicles and then people turn away from you when the wait becomes too long. From that perspective, I think it's less effective for them to not have a partner on the ride-share side. Then also from the cleaning or charging piece, a lot of these… It comes down to how interested are these players in being the owners of the fleet management part and whether they're looking for someone else to take that part off of their hands so they can focus specifically on the tech. A lot of AV players are more interested in the development of the tech right now rather than fleet management and all that stuff down the line. I think from that perspective, Lyft and Uber can also provide an opportunity, Lyft through Flexdrive, Uber through partnerships with other players that helps them take that piece off of the hands of AV partners. I think those two are probably the main reasons, but the main one is the keeping their fleet highly utilised.

[[00:46:40]]
How should we think about the competitive dynamics between Lyft and Uber when it comes to AVs? Is Lyft at any advantage or disadvantage, given that it is just a ride-share company, and does that give the company any advantage regarding revenue-share terms with AV platforms and developers over the next 12-24 months? Or is Uber at an advantage because it's a bigger platform and has many different product lines across ride-share, delivery and other things, and does that put the company at any advantage over Lyft regarding revenue-share terms with AV developers?

[[00:47:36]]
Both players obviously have advantages on that front. I'll start with the Uber advantages, as you mentioned, so cross-utilisation. Uber can have the same supply serve both ride-share and Uber Eats because they've been doing that with their drivers for a long time. I think from that perspective, they're the bigger platform. If you want to start somewhere, most players will start with having conversations with Uber, whether that goes anywhere, that depends. Uber's cross-utilisation and then their global scale, even though Lyft is also working on expanding to other markets. I think volume commitments from Uber probably is more meaningful given how much bigger they are. I think that has some advantages and also the fact that Uber has a lot of free cash flow. They have been investing in a lot of early AV players just to lock in their position with these tech players later on. I do think that maybe Uber's advantages. Then Lyft's advantages is, I think from a partnership perspective, Lyft is a much better partner to work with. I think Uber has had an interesting relationship with Waymo.

I think it's a little bit of a frenemy relationship given that they just recently broke up over an existing market. I think most AV players probably will see Uber as both a competitive threat and a potential partner, vs Lyft has positioned itself as more of a fleet operations partner, a marketplace partner, someone who is not interested in threatening your distribution channel. That has also benefited Lyft in terms of the third-party partnerships, such as the partnerships with United, with DoorDash, etc. I do think that gives Lyft an advantage in terms of being the better partner to work with. Then Lyft also owns Flexdrive, which is a fully owned subsidiary that is able to manage some of the physical fleet management piece. I think Uber does it through third-party partners on their side vs Lyft does this more in-house. It's held closer to the core business. That allows these AV partners to work closely with Lyft in terms of deployment, in terms of Nashville, for example. From a better partnership perspective, I think that makes it easier to have those continued conversations and to have Lyft as a long-term partner.

[[00:50:28]]
Is Lyft a better partner for AV platforms than Uber?

[[00:50:32]]
Yes. I think that Lyft has definitely positioned itself as the better partner.

[[00:50:36]]
How does that manifest? Does that mean it gets better revenue-sharing terms with, for example, Waymo, Zoox or even Tesla if it is involved in the same way? If that's not the case, how does it actually manifest?

[[00:50:57]]
I think being a better partner means maybe one, you're more likely to sign with Lyft for future markets. It could affect the revenue sharing. Right now, I think in a Lyft and Waymo situation, Waymo has probably more leverage in that sense, and so it could help, but maybe not. Then also from a… As Lyft plays as a better partner to one player, the other players see it, and so I think it helps them build better partnerships in the future with other AV players. I think all of those things could influence future developments, but it comes down to relative leverage on both sides.

[[00:51:51]]
Looking at a typical AV trip on Lyft, Uber and so on, who actually controls the customer relationship, trip-routing data, in-cabin display, design and monetisation through advertising? Would it be the ride-hailing platform such as Lyft, or would it be Waymo or other AV platforms?

[[00:52:26]]
It depends on the relative leverage. Maybe we can break it down to several pieces. One is the customer relationship and the billing. Where does the customer actually book the ride from? If the customer is booking it from the Lyft platform, then Lyft owns that customer relationship fully, vs a customer books the ride through Waymo, then obviously, Waymo owns that relationship. For co-dispatch fleets, that depends. That depends on negotiations and how that ends up. For example, May Mobility, people are booking that through Lyft, so Lyft owns the customer relationship. Then once that happens, the next part is, okay, who's routing the vehicles. Lyft, obviously, if it's booked through Lyft, Lyft obviously knows the destination, the intent, where is the pickup location, all of that. Lyft will pass that information to the AV operators, will then tell their vehicles, "Here's the path that you should take to get there."

I think from the ODD perspective, from the pathing perspective, the AV players has to do that part. Then the next piece is in-cabin experience displays, obviously, I think the vehicle owner is the main owner of that. What does the dashboard look like, what is the displays look like, whoever built the vehicles typically has more control there. Though if somebody booked a, let's say, a May Mobility ride from Lyft, you can have discussions around like, "Hey, can we also put Lyft branding here and there." Those conversations do happen. I think you mentioned monetisation in-cabin, that really depends on negotiations and leverage. I would imagine AV operators want to own that and then Lyft also wants to own that, and so it comes down to can both sides agree on some revenue share or co-branded ads or something like that.

[[00:54:36]]
How do you think AVs might impact metrics such as EBITDA margins over the next 12-24 months? Do you think there could be a major impact? Would it happen over a longer period of time?

[[00:55:08]]
I think the impact will definitely be over a much longer period of time. In the near term, it might be a little bit of an additional cost just from setting up these depots and the fixed cost of setting up Nashville, for example. I don't expect that to be huge from the overall mix. I think in the near term, maybe some costs, but longer term, for sure, that you can get… Hopefully bring overall costs down as we talked about earlier. The revenue share comes down to the negotiation power on both sides. That really depends on if the initial deployments go well, then future deployments will probably have even more favourable revenue-share models, etc. That could change pretty significantly depending on who's partnering with who. I do think longer term, it will probably be helpful for margins with expansion of AVs. In terms of timelines, that's really up to anyone's guess.

[[00:56:18]]
Lyft's EBITDA margin is expected to get to about 3-3.3% of gross bookings in Q2 2026. Sell-side analysts seem to think it could reach 3.4% in 2026 and 3.8% in 2027. Do those estimates seem reasonable?

[[00:56:50]]
They seem reasonable. I don't think most of that is coming from AVs. Most of that is coming from international expansion and premium expansion. That is going to drive margins much more than maybe 1,000 AV vehicles on the platform.

[[00:57:09]]
You wouldn't expect AVs to have much of an impact on EBITDA margins in the next 1-2 years?

[[00:57:15]]
I don't personally.

[[00:57:17]]
Would you expect AV adoption to be more of a cost rather than something that will deliver margin or revenue upside for a company such as Lyft in the next 1-2 years?

[[00:57:31]]
I think from the beginning, since you have significant cost of setting everything up, I do think it's probably going to be a bit of a small cost, not a huge cost. It's probably closer to a cost centre rather than actually generating margins. I do think once it scales up, it will switch to become a margin driver.

[[00:57:53]]
What cost would you expect Lyft or another platform to be investing in launching in a new city that fits inside the ODD that we mentioned? How much does it cost to set up? Is it less than USD 1m, USD 1m-10m, USD 10m-20m or USD 20m-30m per city?

[[00:58:27]]
If I'm thinking about setting up a depot like the 80,000 square feet one that Lyft has in Nashville. One, you have to refit those depots to make sure they work with AV chargers, AVs, etc, need to build in some sensor calibration rigs. Then you also have to work on the fleet operations piece. If I add all that up, that's probably single-digit millions per depot and then how many depots you actually build, then the cost could go down a little bit, but you're still working on getting leases on that stuff, hiring technicians, all the operations stuff. That will cost some money. I think once you scale that up, that cost quickly goes away.

[[00:59:24]]
How many depots would you need in a city? Is it 2-3 million?

[[00:59:31]]
Nashville, I think one is probably enough to get started. I don't know. It depends on the size and how many… What's the demand on that city. I can't do that math right now off the back of my head. I think probably a handful in terms of large metros.

[[00:59:55]]
Let's leave things there. Thank you so much for your time. It's been great to chat, and enjoy the rest of your day.

[[01:00:06]]
Thank you. Bye.

[[01:00:07]]
Thank you very much. Take care. Bye-bye.

Transcription ends at 01:00:10 of the recorded material