Thanks, everyone, for joining us. My name is Sean Diffley from the U.S. Media and Telecom Research Team. I am pinch-hitting for Brian Nowak. We are very excited to have the CFO of Uber with us today, Balaji Krishnamurthy. He has been at Uber for seven years, before becoming the CFO recently, in the last nine months. Quick disclaimers. Please note that for all important disclosures, including my personal holdings and Morgan Stanley disclosures, they all appear on the Morgan Stanley public website at morganstanley.com/researchdisclosures. Some of the statements made today by Uber could be considered forward-looking. These statements may involve a number of risks and uncertainties that could cause actual results to differ materially. Any forward-looking statements made today by the company are based on assumptions as of today, and Uber undertakes no obligation to update them. Please refer to their 10-K for risk factors.
With that, Balaji, thank you so much for being here.
Thank you, Sean.
We wanted to get into some of the key learnings that you have had now that you have been in the CFO role for about nine months, after working in strategy and investor relations prior to that for six years. What would be the one or two key learnings you have had that you think are underappreciated about the company? I know you have been doing investor meetings, just anything that you think is misunderstood about the Uber story.
Yeah. I have been at Uber throughout its public existence, and I have gone through a few different roles. Thankfully, no surprises, which is great for your first year. What we have been busy with this year is just a lot of the growth priorities for the business that are still top of mind for all of us in the management team, both in the core and the investments we are making with AVs, and also selectively finding these opportunities where we can invest from a capital allocation standpoint to be strategic about what kind of acquisitions we could be making. Those have been the areas where most of my energy has gone through this year.
I think what is underappreciated perhaps is at the scale at which we operate and the sort of daily use case for many people in this room that Uber has come to be. Generally, investors look at it and have always thought about this as a potentially more mature business than it is. We have been able to find ourselves in this position where, even at the scale at which we are, we are doing nearly 16 billion trips this year, we are operating with about 215 million monthly active consumers. We are still growing 20% on a gross bookings basis or thereabouts, and this is the Q3 was our fifth consecutive quarter of 20% growth, and we delivered 35% EPS growth in that period as well. What we see underneath that story is just kind of the penetration headroom that is still left.
We have in our top 10 markets, less than 15% of consumers use, or 15% of the population uses our service once a month. In the U.S., in dense markets, only about 50% of the consumers have used our service once in the last 12 months. In sparse markets, that number drops to 10%. We have lots of runway here for this business, and that is true for both mobility and delivery. The other side of the equation, which is on autonomous vehicles and what is a future growth vector for us, we fully appreciate that investors look at that story and see that as a big risk for our mobility business.
The way we think about this is fundamentally a massive opportunity for mobility, and the core belief that we are investing behind is over time, human labor, as it always has done, will continue to inflate, whereas autonomous vehicles, as fundamentally a tech hardware solution, will continue to deflate. Over time, that is going to put us in a place where autonomy unlocks a price point that is lower than what we can deliver today, and that allows us to really expand the TAM that mobility can serve.
Great. I want to pick up on the last point on the AVs. You have been very passionate about autonomous. On the Q3 call, you outlined some clear targets. So 700 fully driverless vehicles on your network by the end of this year, over 2,000 by the end of next year, to tens of thousands by 2028 and beyond. Let us talk about some of the things that have to be done to get there. What are the operational hurdles? What are the milestones? How should we be thinking about that progression?
Yeah. That is a great place to go. I think when you look at our roadmap, what we have said is 700 fully driverless cars on our network by the end of this year, 2,000 by the end of next year, and then scaling to 10,000 and beyond in 2028, as you mentioned. How we get there is that we expect to be live with about 10 different autonomous partners globally next year, and we will be live in nearly 20 cities as we go through the next year. We are right now in the process of building our infrastructure in nearly 40 cities, which is what we will need to be ready for in 2028.
When you think about what gets in the way of these deployments and what needs to be true for us to be able to scale out, you need at least three components to fall in place. The first is the tech needs to be ready, right? When you look at these deployments, the safety bar is exceptionally high. Society is not going to accept robots running around our roads and being just as safe as humans. They need to be significantly more safe than humans, right? Particularly if there is a prospect of job displacement in conjunction with these deployments. What we have been working on is to go slow. We have focused on deployments where initially we have launched with some safety drivers.
For example, with WeRide, when we launched with them in Abu Dhabi, for the first deployment, it took us nearly eight months to go from launch to taking the safety drivers out. Now we have expanded to Dubai, where we took three months to do the same. By the time we exit this year, we will have nearly 200 fully driverless cars running in the U.A.E. with that partner, and that number will scale in the next year. Similarly, in the U.S., we have worked on getting to a driverless launch with Motional, which will happen in this quarter. We have been working with them with safety drivers in Vegas. We recently launched with Zoox, which is, by design, a fully driverless launch. So Vegas will have two options on our app fairly soon. Then this week, we got to our first driverless mile with Nuro.
It is in testing, right? It is not open to commercial rides yet. But in Q1, we expect to put riders in the back seat in a fully driverless mode in the Bay Area with Nuro as well. As we look beyond that, we have a number of partners like NVIDIA, Rivian, Avride, Wayve, et cetera, that start launching and scaling in 2027, 2028 at various timelines. That is the first piece. The second piece you need is infrastructure, right? Unlike original ride- hailing, where all we had to do was sign up drivers, run their background checks, and off you go, here, you are going to need infrastructure on the ground, which means that you need to work with utilities. The lead time there tends to be 18- 24 months. You need to meet city codes, et cetera.
There is just a lot of involved work that goes into that equation as well. As I said, we are working in about 40 cities to get that going. The beauty for us is we have a really, really rich data map, heat map of where demand exists, and we are able to say what would be the spots in the city which are at the 90th percentile for minimal deadheading, and you can get power wired in. What we see is that our competition, who are trying to launch potentially 1P services, they do not have that visibility, and they are putting their spots in potentially inferior spots. We will have an advantage there as well. Then the final piece is regulatory approvals, right? Today, by regulation, AVs are not legal everywhere. We have to work with the industry to get cities to approve this.
Local transportation is local, and you will have city jurisdiction on a lot of this stuff. We need to work through a lot of these reservations that cities are expressing, and that will be the third gating factor that we have to get through.
Great. That is really helpful. I wanted to talk about AV industry structure and the competitive landscape. I think a common investor concern is that you face supplier concentration risk and potential disintermediation by vertical-specific 1P players. Maybe you could talk about how you respond to these concerns and what you think your biggest value add is. Then obviously, you have a utilization gap. How durable do you think that is versus the 1P players?
Yeah. I think the concern here stems from this belief that we are going from a completely disaggregated driver supply to a world where you've been working with potentially one or two or three partners. I would say there's a little bit of a nuance here which is important, which is that ultimately what we believe will happen is AVs are an enabling technology that goes into the OEM supply chain, and typically OEMs will have multiple partners they work with in their supply chain for any solution, and AVs are no different from that. Secondly, the concentration that starts materializing over time is more one of going from individual units of driver supply to potentially fleets of AVs rather than the software itself, right? Which is some concentration, but we do believe that fleets have to have local expertise in every market in which they operate.
We are going to see many, many fleets materialize on our network rather than a few concentrated fleets globally, right? I think a few things that I would go through which are important to keep in mind as you look at why Uber adds value here. Fundamentally, ride-hailing demand is very variable throughout the week, throughout the year, and hybrid networks are likely to have higher utilization than a 1P demand channel, right? We are seeing that on the ground today. We are seeing that in the U.S., for instance, where we have deployments, our utilization lift is at least 30% higher than mature AV markets with 1P deployments, and it's as high as 100% relative to newer deployments in Austin or in other markets that have come through in the rest of the U.S., right?
Utilization is the first piece of the equation, and if you have a 30% plus utilization lift, you can easily justify as high as a 23% take rate, which is, by the way, higher than our current take rate net of insurance, right? That's one. The second thing, which is also important, is fleet management costs. Investors are very focused on what is the software, hardware cost, but the reality is, at maturity, the biggest part of the cost structure is actually fleet management costs. On that part of the equation, we believe, because we can co-locate a number of our partners into the same sites, we have an ability to pick sites that are going to be superior to others as well, as I said earlier.
It allows us to get to a fleet management cost per mile that is going to be superior to what you can see with others, right? That's the second piece. The third piece, which is also very important, is platform costs. Platform costs, I would think about those as customer support, insurance, payments, marketing, just general ops for these deployments, which are true for ride-hailing with humans, and it will be true for ride-hailing with AVs. That is a piece of the equation where we have a significant advantage, starting from day one, relative to anyone else, and that stems simply from the fact that we're doing 16 billion trips on our network, and we have economies of scale that nobody else can match, right?
So maybe just to put a finer point on that, if you looked at Uber's platform costs, those items that I just mentioned, at the time of our IPO, when we were doing about 7 billion trips, versus our platform costs today when we're doing 16 billion trips, the leverage we've driven there is about 7 percentage points of our gross bookings. That is substantial. That's almost as much as our mobility margins. Anyone coming into this space competing with us has to offset a bigger number than that, because their scale is going to be in the millions, not 7 billion.
That's really helpful. I wanted to ask about the infrastructure investments that are required for AV, which you've just touched on a bit. Obviously, on the call, you talked a bit about some of the financing structures that you're going to employ for this AV fleet financing. Can you talk to us about how we should think about the balance sheet and some of the dynamics and considerations there?
Yeah. If you look at all of the commitments that we have already announced with our various partners, and you just added up the commitments on equity investments, the infrastructure that we're building out, as well as the vehicle offtake commitments we've made, which we've said it's over 120,000 vehicles over the next many years, those add up to about $10 billion, right? We are good for that over the next few years if it had to go off our balance sheet. We have the capacity to fund it if we needed to. Now, the way we see this playing out, though, is that today and the next year, where the scale is relatively small, and we want to have some time to market speed, we are funding this off of our balance sheet.
The equity investments are really linked to milestones that help us get visibility with our partners. So, for example, when I talked about Nuro reaching its first driverless mile in the Bay Area this year, sorry, this week, that is a specific milestone against which they unlock some capital from us, right? We have that contracted in when we initially engaged with them, and we know by when they're going to get there, right? So that's good, and we'll continue to invest behind those kinds of efforts. The infrastructure piece, again, the 40 cities I talked about, because we need those ready by 2028, we are going to invest behind that off our balance sheet. But these are typically leases, so you're talking about 10+ year leases here, which again, the number becomes not very material.
On vehicles, through the next year, we are talking about 2,000 cars or so, really modest deployment in the next year. As we go beyond 2027, what we would like to see is that we are not constraining the scale-up of these deployments to Uber's balance sheet, and bringing in third-party capital that can help us turbocharge the expansion efforts here. There will likely be two variants of the third-party financing efforts that we are talking about here. The first is to finance infrastructure, and that looks very much like real estate, and so it is not that complicated to fund that. We do believe that over the next few months, we can bring that to life and show investors what that looks like.
The second piece is vehicle financing, which the variables that we need to work through are revenue- generation capacity, useful life, and residual value. So there is a little bit more complexity to that, but through this year, we have seen a lot of financing efforts that have gone after the data center and GPU space, which have similar characteristics on the GPU side. In this case, we do have a higher visibility to what kind of revenues you can generate with these cars. So, we feel good about the investor engagement we have there, and we will provide more details as we go through next year.
Excellent. Given that we are in Barcelona, I wanted to ask about the international AV landscape and how it could differ from the U.S. So maybe just hit on some of the regulatory environment concerns and other issues we should be thinking about in terms of speed to scale to commercial service on AVs as we think outside of the U.S.
Yeah. The good news is that we have seen a lot of energy from a regulatory standpoint, and not just in the U.S. Even in Europe, we are beginning to see a lot of markets where the regulators are actively engaging, and there is a desire to be leaning forward on technology. We have talked about the Middle East already, but the Middle Eastern theater, we are seeing a lot of activity from various markets, and we do expect that that is going to be open over the next one to two years in most places. There, our partnerships with WeRide and Pony.ai are likely, and Baidu as well, we are likely going to see those deployments scale up fairly meaningfully over the coming years.
Then in Europe, we have seen engagement in Madrid, in London, in Germany, et cetera, which we are right now working through the nuances of the regulatory efforts there. We expect that as we go through the next year, both with WeRide and Pony.ai, we will scale up a lot of these deployments in Europe. In London, where we are also working with Wayve is a great partner of ours. We do believe that their technology can generalize fairly quickly. As they prove out their path to fully driverless L4 deployments, we expect to go global with them as well. There's a number of partners already, and some of the long-term vision we've been painting for our AV strategy is going to be real in international markets from the get-go.
What we are looking forward to proving out is just our ability to show that utilization lift at scale and then hone our capabilities on fleet management and just getting consumers visibility and comfort with these deployments.
Great. We've talked a lot about autonomous rideshare. What about autonomous food delivery? I think this is one of the areas where you're making some clear progress. What would you say are the biggest gating factors to scale there, and how would you compare and contrast autonomous food delivery versus rideshare?
Yeah. This is another area we're very excited about, and I would say in this space, again, we are partnering with a number of players. You should expect that autonomous delivery actually matures into distinct form factors. The first one is kind of the sidewalk/bike lane bots. These are, unlike autonomous mobility, where we are going through a deep investment phase right now, and every trip we're taking from a human driver and putting on a robot is going from a profitable trip to an unprofitable trip. Sidewalk delivery actually can be quite attractive from the get-go, and we are seeing that the economics are reasonable, and they can be a good form factor that we can scale.
Now, the challenge on sidewalk delivery is really that you have added friction at both ends of the delivery relative to a human delivery, where a human comes into the store, picks up the order, drops it off at your doorstep, and there is very little friction. There is a little bit more friction cost that you have there. So that might be the piece where the ecosystem needs to solve that before it can really go wide. The second form factor, which we are very excited about, is drone deliveries, and those have the potential to be quite transformational because the experience is significantly better than, potentially better than, humans, right? Because you can get the deliveries done faster. If your food arrives faster and hotter, you are always going to be happier, right? That is kind of the promise there.
What we are seeing is, we just recently announced a partnership with Zipline, which we are looking forward to scaling. The goal we have jointly set is to get to 1 million daily deliveries by the end of 2029. It is an ambitious goal, and we are looking forward to proving that out. But we have already started with them in Texas, and by the end of this year, we expect to have 10 live drone delivery sites with our drone delivery partners.
Excellent. I wanted to switch gears to some of the near-term mobility supply dynamics. You obviously referenced on the call that you had some supply-related headwinds in Q2 and 3Q in Brazil and India specifically, but you are seeing some recovery there as a result of some of your targeted investments. So maybe walk us through what is the latest, what are these investments, and how should we think about the international mobility supply dynamics as we move throughâ
Yeah.
âthe rest of this year and into 2027?
Yeah. Maybe it is prudent to start from the highest level of what our supply growth looks like. If you look at our overall supply base, we had about 10.8 million monthly active drivers as of Q3 2026. That number was 9.4 million a year ago. What that tells you is that at the scale at which we are operating, to support the growth we have, we have to add millions of new drivers every single year, and that is a very meaningful acquisition effort that we have to keep in place every single year. Secondly, from an annual seasonal cadence standpoint, September through December is the strongest demand period for our business, and we kind of have to step up supply almost on a weekly basis to meet that demand.
What we are seeing right now is a combination of three issues that are kind of putting downward pressure on our supply growth efforts. The first is that gas prices are high globally. Again, when gas prices are elevated, you see an impact on new driver acquisition efforts because our existing drivers know how much they earn on Uber, and even the increased gas prices have a more muted impact on driver health relative to new drivers who do not yet know how much they could earn, but they do know how much gas costs. That is an issue, and it is a pronounced issue in low-fare markets where gas price is a bigger part of the driver P&L.
The second issue we have seen is localized competition for two-wheel supply, in particular in markets like Brazil and India. Some portion of that is driven by incremental competition in delivery, which is a category we do not participate in in those markets. In Brazil in particular, you are seeing a lot of elevated competition from Meituan, Didi, et cetera, against iFood, which has a knock-on impact on two-wheeler supply for our Moto business. That is the second. The third piece is really just tighter immigration flows and larger developed markets, and it was a pronounced impact for us in Canada in particular. We are stepping in with some targeted investment efforts here across various markets, and it is focused on driver acquisition as well as engagement and retention efforts with our existing drivers.
We are in a fortunate position where the marketplace was well-funded from an investment standpoint on both demand and supply, and we are just tilting it a little bit further towards supply right now. That is allowing us to be in a position where, while we had these issues, we continue to have supply stress. The pressure stabilized in October, and the trend lines are improving. We expect that for Q4, we can deliver something consistent to that 17% growth we delivered in Q3, and that is how we operate. We are doing this without really a big trade-off on profitability. Our EBITDA mobility margin was at an all-time high in Q3. Our EPS grew 35% in Q3, and that number at the midpoint in Q4 is near 37%. We are in a good spot from a profitability position.
Great. Earlier, you referenced the big growth opportunity in sparse markets. Maybe just flesh that out a little bit for us. I think those markets are growing 1.5x faster than dense marketsâ
Yep.
âin mobility. I think Australia is an example that you've used, but maybe speak to your strategy in these markets. How can you accelerate growth even faster there?
Yeah. Uber has always had this legacy of being a product that has had a great product- market fit in dense markets, right? This was true for both Uber and Uber Eats, and historically, that has been a big strength for us. What we have realized over the years is that everybody who lives in these sparse markets also moves and eats, right? So, we have an opportunity to appeal to those consumers, but we didn't have the right products to meet those consumers' needs, right? So, we've had a concerted effort over the last few years to innovate from a product standpoint, and then from an operational standpoint, we also need to make the investments to go and invest into these markets, whether it's in driver supply efforts or think about selection growth on the delivery side.
You need sellers who can go into these sparse markets and actually acquire the merchants. So, we've been investing in both of those businesses quite actively, and that has allowed us to get to a spot where the sparse markets' growth for mobility and delivery is significantly faster than our dense markets' growth. So, just to call out a few things. For mobility, we introduced products like Reserve, where consumers who are willing to pay a premium in exchange for reliability, where they want certainty that a driver will show up at a certain time, Reserve is a great product for them. On the other end of the spectrum, we introduced Wait & Save, which is consumers are expressing their willingness to trade time for price, right?
They are telling us they will wait longer if we can find them a driver who can pick them up at an affordable level, and that has been a great product fit for sparse markets as well. As of Q3, Wait & Save was at an $8 billion run rate, growing 100%. Once you have these kinds of products that allow you to get some consumer demand, you start getting to a spot where your flywheel starts working. You have demand, you can put more supply into the market, and that allows you to start adding more products into the pipeline, and products like Teens, which allow us to go after an incremental family member, seniors, women preferences. All of these products are now building out a full portfolio that is allowing us to serve the consumers in those sparse markets better.
Similarly, on the delivery side, I know your question was specific to mobility, but I will round that out with delivery because it is a concerted effort for us. On the delivery side as well, we have been making a deep investment in selection growth as well as our product, where you need to have a different kind of a product motion for the couriers in sparse markets because demand density is not the same, and we have made those kinds of changes. Now we are turbocharging that with partnerships that can help us reach consumers in a very efficient way. The partnerships that you saw us announce with Costco, McDonald's, as well as Amazon recently, these are going to be ways for us to reach consumers in sparse markets quite efficiently and just accelerate our delivery business further.
Great. Let us dig in there on the delivery growth angle. In the U.S. delivery, you saw an acceleration for the second quarter in a row, and internationally, really strong results as well. Maybe unpack for us what is driving the growth between core U.S. restaurants, international, and grocery and retail.
Yes. We had our second consecutive quarter of healthy acceleration in the U.S., and similarly in LATAM, we accelerated by a very strong amount in Q3. If you look at other international markets where we already have a strong number one position, like the U.K., Turkey, these markets, we got to an all-time high category position in Q3 as well. What we are seeing is very broad-based momentum for our delivery business globally, and it is driven by selection, quality. It is driven by the product improvements we made to reduce defects. Underpinning all of this with an Uber One offering that really delivers really good value to consumers. At this point, we have about 55 million Uber One members globally that drive about 70% of our delivery gross bookings, and that penetration has increased 20 percentage points in the last couple of years.
When you see those sort of themes come together, that really just helps us drive a lot of growth in that business. Specifically back to the U.S., where we had acceleration, what we are seeing there is our efforts to, one, the dense markets still have good penetration headroom, and we are seeing good momentum there, and we have a strong number one position in a number of key markets like New York, L.A., Miami, et cetera, and those markets are doing really well. But in addition to that, we are making a deep push into the sparse markets where we have a lot of runway and our competitor has entrenched lead, and we are chipping away at that. We are also, as I said, going after some of these partnerships that can drive outsized returns for us.
When you think about Costco for a minute, which was working primarily with Instacart historically, we started that partnership with a subset of the U.S. We were operating with them in about 17 states since 2021, and now we have expanded to all of the U.S., and we believe that head start that we had in those markets has allowed us to build a product with Costco that is going to be a good experience for consumers, and we can drive growth there. Then McDonald's, where, on their first-party delivery service, we have a good partnership with McDonald's that goes back many, many years on the marketplace, but on first-party delivery, we had a much smaller business with them, and recently they flipped that to we are the preferred delivery partner for McDonald's globally, which will obviously have an impact in sparse markets.
Then third, the partnership we just announced with Amazon, where our food delivery inventory is now available on the Amazon app, and Prime members get access to free delivery over $25, et cetera. That channel has the potential to be a pretty meaningful demand needle mover for us in sparse markets, and so far, we have seen incrementality come through better than what we had anticipated going into that deal.
Excellent. I wanted to turn to the competitive landscape and consolidation and competition in delivery, which is always topical at this conference. Obviously, your competitor is acquiring Instacart, and that follows Deliveroo last year. You have obviously, you have agreed to acquire Delivery Hero. Maybe talk us through how you are thinking about building out the international delivery, which markets we should be thinking about, where the biggest focus is, and then touch on quick commerce in those comments, please.
Sure. I will start with our own consolidation efforts. We announced an intention to acquire Delivery Hero. We are very excited about that transaction, and right now we are going through all of the engagement with regulators and ensuring that we can meet their concerns, if there are any. So far, it has been a constructive conversation that is moving as we expected. In parallel, we are going through the tender process where we just did our early close, and next week we will have the extended close on the tender, and we feel good about what we are seeing there as well. When we think about the asset itself, the quality of Delivery Hero's businesses in certain theaters is exceptional.
When we look at their Middle Eastern business, for instance, with what they have built with Talabat and HungerStation, they have a clear leading position, and we have a clear leading position in our mobility business there. When you combine those two assets, you can actually create a really strong platform, similar to what we have in a lot of markets where we have organically built Uber and Uber Eats. Similarly, Latin America, we feel very good about that business, and then Korea, of course, they have a massive business there, which allows us to extend into a large top five delivery market where we have historically not operated.
I would think about this as the Korea angle, by the way, is similar to what we have done in Turkey, where we did not operate, but we were able to acquire our way in, and now we are the market leader and growing really fast there as well. When you think about dark stores specifically, our view here is somewhat nuanced. We do not believe dark stores work everywhere, but there are certain markets where the market characteristics allow for those stores to be meeting a consumer need and allow us to operate it in a way where we can run that efficiently and at a cost point that makes sense. Definitely, we see an opportunity in the Middle East. We see an opportunity in Turkey. We have historically operated dark stores in Taiwan.
There will be selective places where we think this opportunity exists, and we will operate against that. On Instacart, of course, it is a good asset. We have a lot of respect for that team, and we think they have built a good business. Ultimately, our view here is that there are some pieces of our opportunity that we can build organically, and grocery and retail in the U.S., we see a long runway for organic expansion there, and we will continue to execute against that. From a merchant engagement standpoint, we feel good about the opportunity we have to go deeper with a lot of the merchants who have worked historically with Instacart, and they will look to expand with other partners.
Great. I wanted to talk about reinvestment and profitability. I think there's been some investor concerns that you're reinvesting a lot of the profits that we're seeing come out of the business. Maybe just talk about some of these key areas of investment. We've hit on some of them earlier. Obviously, AV part of it, driver supply is another. How should we think about you balancing profitability and reinvestment across the businesses?
Yeah. The principle with which we operate is that we want to reinvest to sustain long-term growth. But the bottom line is it needs to also translate to the highest profit maximization outcome for the business over the long run, right? So we're not just reinvesting for the sake of growth. It needs to have a path to driving higher profit dollars relative to the counterfactual over time. What we have done so far is actually very consistent with that, and our track record speaks to it. If you look at the last three years where we had said we would grow mid- to- high teens CAGR gross bookings and high 30%-40% EBITDA CAGR, what we are on track of delivering is a track record where gross bookings has compounded at over 20% and EBITDA has actually compounded at over 40%.
So we've actually been able to produce more profits than we committed to and what investors had expected, and that's kind of the position we want to put ourselves in for the next three to five years, right? The reinvestments we're talking about here are in the core. They're focused on product innovation as well as consumer acquisition efforts that we can continue to lean into. Membership plays a big role here as well, and we are continuing to make that sort of investment. Then secondly, we want to be able to invest behind AVs, which in the near term are an investment kind of an area, but over time, they're going to be a significant source of expansion for the business, both from a top line and profit standpoint. You need to be able to invest into that to deliver on the long-term promise, right?
I think it's important from my, especially as I think about my responsibility and our responsibility to shareholders, it is important that we continue to find the sources of cost savings that can help us invest behind this as well. So, for example, we've been very disciplined on headcount growth over the years, and we still went in and executed on a 10% headcount reduction effort in the last quarter, which allows us to get multi-hundred million dollars of savings that we can reinvest into the business, right?
We will continue to balance these things where we are reinvesting, but we are also staying very disciplined on costs. We will strike the balance where we can deliver this high 30% kind, for the recent quarters where we have seen 20% growth with mid to high 30% EPS growth, that is kind of the balance we want to be able to strike, where operating leverage shows through into the next few years as well.
Great. I wanted to turn to the agentic landscape. Obviously, a lot of investor focus and consumer utilization here in the past year with consumer-facing horizontal tools accelerating. Dara on the last call talked about your focus on the direct-to-app experience being frictionless, but you do plan to have some discussions with some horizontal agents. Maybe just walk us through your approach to agentic, how you think Uber fits into the equation.
Yeah. The bottom line for our engagement with agentic experiences is that it needs to deliver a better consumer experience. If it is just recreating existing consumer experiences in a worse way, it is going to go nowhere because consumers will not adopt it, right? If you think about today our apps, for both mobility and delivery, a lot of the consumer engagement is actually quite frictionless. Consumers come to our mobility app, three-quarters of the time it is a single tap that gets them on their way. Our typical transaction on mobility takes less than 50 seconds, right? The second flavor of the consumer engagement is where they actually need customization. You come in, you want to change the sort of car you have because you have five people with you, so you need a bigger car, or you need a car seat.
Similarly, on the delivery side, you may want to explore restaurants or customize your order. For those kinds of experiences, what we have seen is consumers prefer visual interfaces rather than chat-based interfaces, right? That is really the reason why we see so far limited traction with agents. At the same time, we are investing in first-party agentic surfaces as well, where we can, for instance, the cart builder that we have on grocery, we do see pretty meaningful uplift when consumers engage with that, and that is a good sort of experience. What we recently announced with Muse, for instance, our integration with Uber Courier. Courier is our peer-to-peer parcel delivery effort.
That is a really good product experience potentially, and we can see the incrementality, both from our perspective, but for consumers as well, where there's an end state vision where potentially, you want to buy something on Facebook Marketplace and Uber Courier can just deliver that product to you from the person you're buying it from, and the agent, Muse, can just figure out all of that for you, right? That's a superior experience where it makes sense for us to engage with them, and we're delivering jointly something that we can't do on our own.
Great. I wanted to close out with capital allocation. Obviously, you're leaning slightly away from M&A after the announced transactions, and more towards investing in AVs. Maybe just remind us how you think about capital allocation, the potential for capital return, and how you're weighing all of those priorities.
Yeah. What we said was right now, given the acquisitions that we have announced and some of them remain in flight, our focus is on ensuring that we can deliver on the integration hypothesis that we had built out before the deals were announced. For example, in the case of Delivery Hero, we've said that we're looking to deliver at least $1.2 billion of synergies within 18 months of close, and we have a lot of work to do to get those things done, right? It takes up organizational energy to ensure that you can successfully close transactions, and that's why we said that through 2027, we expect little to no M&A activity. Beyond that, we'll reassess where we are, and we'll continue to pursue things that make sense.
At the same time, we are making some investments in AVs, which we've talked about at length earlier. We should still be left with capital that we can return to shareholders. Our position, since we initiated the buyback authorization, has been that we want to return about 50% of our free cash flow to shareholders. So far, we've done more than that. Over the next multi-year period, there's no change to that stance, and we want to continue to return somewhere in that 50% vicinity.
Excellent. Balaji, thank you so much for joining us.