Operator: Ladies and gentlemen, thank you for joining us, and welcome to the DoorDash Q1 2026 Earnings Call. [Operator Instructions] I will now hand the call over to Weston Twigg. Weston, please go ahead.

Weston Twigg: All right. Thank you, Elizabeth. Good afternoon, everyone, and thanks for joining us for our Q1 '26 earnings call. I'm pleased to be joined today by Co-Founder, Chair, and CEO, Tony Xu; and CFO, Ravi Inukonda. We'll be making forward-looking statements during today's call, including, without limitation, our expectations for our business, financial position, operating performance, profitability, our guidance, strategies, capital allocation approach and the broader economic environment. Forward-looking statements are subject to risks and uncertainties that could cause actual results to differ materially from those described. Many of these uncertainties are described in our SEC filings, including our most recent Form 10-K and 10-Q. You should not rely on our forward-looking statements as predictions of future events or performance. We disclaim any obligation to update any forward-looking statements, except as required by law. During this call, we will discuss certain non-GAAP financial measures. Information regarding our non-GAAP financial measures, including a reconciliation of such non-GAAP measures to the most directly comparable GAAP financial measures may be found in our earnings release, which is available on our Investor Relations website at ir.doordash.com. These non-GAAP measures should be considered in addition to our GAAP results and are not intended to be a substitute for our GAAP results. Finally, this call is being audio webcasted on our Investor Relations website. An audio replay of the call will be available on our website shortly after the call ends. Operator, I'll pass it back to you, and we can take our first question.

Operator: [Operator Instructions] Your first question comes from the line of Shweta Khajuria with Wolfe Research.

Shweta Khajuria: Let me try 2, please. One is on product and the other is on partnership. So first on product, could you please talk about how you envision your product develop over the next 12 to 24 months as you integrate more of agentic and AI capabilities. So will we have an opportunity to sort of communicate via voice and put a cart together and execute a transaction, even saving us more time or better search and discovery or whatever it may be, if you could please talk to that. And then the second one is on partnership. You announced extension and expansion of your partnership with Lyft. As you think about the greater value proposition around local commerce and becoming the operating system for local commerce, how do you think about travel as an adjacency with Uber partnering with Expedia? Is partnering with Airbnb and Booking.com type partnership a value add or something else? Your thoughts on that would be great.

Tony Xu: Shweta, it's Tony. Maybe I'll take both of those and feel free to add in anything you want, Ravi. Look, on the first question with respect to product -- the DoorDash philosophy and story has always been the same here, which is we have to create the best end-to-end shopping experience. If we do that, we will continue to be the ones that innovate, lead. We'll continue to deliver great results like the ones that you saw in the quarter and in the many years leading up to the results that we've just shared. And so there's not one way to do that. You talked a bit in your premise, Shweta, this idea that you should be able to, with the assistance of agentic-like tools to have better discovery search experiences, and we agree with you. I think that we absolutely will have agentic ordering experiences in which it will be a lot easier for customers to do many things that they do today with much lower friction to discover things that they perhaps didn't know, existed on DoorDash to formulate complicated queries and solve those in the best possible way. And the most important thing in delivering this is making sure that we actually can do it so that we don't just win on discovery and the upper funnel, but the end-to-end experience because what's the point of having the best discovery experience that we can't bring you that exact item or that if that exact item were out of stock or it didn't meet your personalized preferences that we can't actually solve for that need. And so for us, the way I think about it is there's no one thing. There's no one trick. It's making constant and continuous improvements to the selection quality, the accuracy of the catalogs, making sure that we offer the widest choice in terms of affordability and different price points, offering certainly the best quality of experience in speed, in timeliness, in accuracy, and then obviously, in customer support, which, I think, also is having an agentic revolution in and of itself. And so you will see all of these things play out in the DoorDash product experience. The most important thing, though, is that we have to build the best end-to-end experience. And we are the only company that has the most robust catalog, much of which is actually about the physical world that does not exist in any digital repository that cannot be scraped and that we ourselves uniquely own access to because of all the work that we do to actually build up a repository of the physical world. And so that is -- that's something that we will continue to build, I think, greater and greater advantage in, especially in the world of agentic commerce. Your second question on membership and kind of this idea of how will partnerships evolve. The way to think about it is that membership experiences and the benefits that kind of live underneath the umbrella of membership programs, they kind of only matter if they are best-of-breed experiences to customers, right? This is why you see different customers, for example, choose a variety of different memberships even for the same product, like if you take streaming, for example, some people prefer shows of a certain format on one network, whereas some others prefer shows of a different format on a different network, and that's why they end up having multiple membership programs and things like this. And what it really -- and there's so many examples in which I can give you of this where being best of breed is really ultimately what customers care about and why they will choose to either adopt your program or not adopt your program. And as you saw in some of the results that we kind of discussed record engagement in DashPass as well as our other membership programs around the world, what we're doing is we're building the best-of-breed product experience when it comes to eating and in shopping increasingly as we go outside of the restaurant category. Now, there's a long ways to go, right? I mean there are 20 to 25 occasions for eating alone every single week, so over 100 every single month. And then if you add in shopping, it's even higher than that. And on that combined sum, we are a tiny fraction of what's actually available and addressable, which in some sense, means that there's a large runway and opportunity for us to become even better in breed in terms of what it is that we can offer. And if we can keep doing that, I think we're going to be just fine. I think that's why you see that -- you see it in our numbers. You see it in our growth rates, both in the U.S. and outside of the U.S. We are gaining share virtually in every single market, and we're growing at near historical highs pretty much in all of our geographies. And so I think that's happening even at the scale that we've developed over the last few years because we're continuing to build the best-in-breed experiences in our categories that have a very large runway for growth.

Operator: Your next question comes from the line of Michael Morton with MoffettNathanson.

Michael Morton: One for Tony and then a quick one for Ravi. Kind of following up on what you were just speaking about, Tony, AI and partnerships. As the AI platforms become more capable, there's a concern from investors that like personal agents could layer themselves in between the on-demand marketplaces and the consumer. I would love to know Dash's long-term strategic view on this and if there's a risk to your business to becoming an API or logistics offering to these, and why or why not you'd want to work with one of the third-party AI platforms? And then the quick one for Ravi, as you've been operating Dot for a bit now in some cities. Are you willing to share any learnings about what percentage of the U.S. delivery market you think is addressable for AVs and then maybe thoughts on how to incentivize consumers to come out and meet the Dot or where the opportunity costs are around cost to serve with AV?

Tony Xu: Michael, I'll start on your question related to agentic commerce and just kind of agents and whether or not there's any intermediation or disintermediation risk. I mean, I really think, like, what's instructive here is what we've seen historically with all top-of-funnel programs, right? For at least a decade, you can argue companies like Google or Apple and many other large platforms were top-of-funnel drivers to a lot of different commerce platforms, ours included. Take, for example, Google Food Ordering, which allowed you to order through various Google channels, whether it's Google Maps, Google Search, I believe there are a few others, too, in which you can order restaurant delivery that started in the mid-2010s, and it went for about 8 years before they shut it down, where you can order delivery from any one of these Google surfaces. And from a traffic perspective, they absolutely could drive a lot more traffic than virtually anyone else could to any one of these restaurants. Yet the retention of that traffic was a fraction of what platforms like DoorDash saw. And as a result, customers effectively moved all of their shopping experiences to DoorDash. And I would argue something similar happened with Amazon, where perhaps at the beginning of the 2010s, Amazon was not a leading player in the product search kind of category. But by the end of the 2010, Amazon ended up owning a significant percentage of all product search terms related to commerce. And you may ask me, why did that happen? And what lessons can we learn or borrow from history to kind of instruct what's happening in this moment and in the years to come? What I would say is customers ultimately don't care about any top of funnel, DoorDash included or any of these agents. What they care about is, did they get the order that they wanted? Did they get the item that they actually were looking for? Did they get it in the best possible experience? That means in terms of price, the speed, the timeliness, the accuracy. Obviously, if something were to go awry or wrong, that it was fixed appropriately and quickly. And when I look at it from the customer's perspective, they're going to ultimately judge us on the best end-to-end experience. And so -- and that's what we're focused on maniacally at DoorDash, in which we're not just trying to build aigentic ordering experiences on DoorDash to make the discovery or the search experience easier, kind of echoing what I said in the previous question, but we're also building a catalog, a digital catalog of structured information for the physical world, collecting where every banana sits or every ripe or unripe avocado to every size shoe in whatever color and style that a customer is looking for. All of that information about the physical world, of which there are billions of items, tens of millions per city and getting that annotated and having that unique and proprietary to DoorDash, which we don't have to share with anybody. And I think if we can do that and improve our discovery experience as time progresses, given the power of some of these agentic tools, I think we're going to be the best end-to-end shopping experience for customers. And ultimately, that's how we're going to get judged. I think that's the reason for why, for instance, even our restaurant delivery business, which is the oldest of the areas in which we operate continues to grow at above historical highs because we're constantly trying to build the best end-to-end experience and be best of breed in doing so. So it doesn't mean we're perfect. We got a long ways to go. And it doesn't mean that it's a guarantee that we're going to be able to get there. But if we can keep executing like we have, I think the numbers will continue to speak for themselves. And these top-of-funnel players will be partners of ours in which they'll drive a small percentage of our traffic and a lot of that will be a choice that we'll have.

Ravi Inukonda: Michael, it's Ravi. On your second point around Dot, right? Look, I mean, I think we are very happy with the progress that we're making. Maybe I'll talk a little bit about the vision. Look, the vision for us is we are building an autonomous delivery platform because ultimately we think different formats are needed for different types of deliveries. That's how we build the most efficient network. We're obviously happy to partner with others. We're happy to build ourselves. I think there's going to be different formats, both on land as well as air that we are working on. Look, we are early on this journey. We are scaling. And what we are trying to do is obviously operate at scale, manufacture at scale. That's going to be important for us. We've seen good results. We've launched it in a couple of markets. I think in terms of the end customer benefit because I think that was one part of your question, it's going to be a combination of the key things that we focus on, right? It's going to help us with speed. It's going to help us with quality. It's going to potentially help us with overall range of deliveries. But the key, I would say is the work that we are doing is starting to look good. We are early in our journey as well as the overall progress that we're making is going really well according to the plans that we made at the beginning of the year.

Tony Xu: Yes. One thing, Michael, I'll add to the autonomy story that I think sometimes perhaps is harder to see from the outside is that there's a pretty big difference between just shipping a vehicle or having a vehicle ready for a demonstration and a vehicle that can really operate at scale under any condition and is really battle tested, right? It's kind of like saying, I can shoot a 3-point shot and so can Steph Curry, but one of us is the greatest shooter of all time and one of us may be hit so once in a while. And this year for us, it's really climbing that curve for the autonomy program and making sure that we can harden our -- I mean, it's not just the autonomy, it's the autonomy, the hardware, the remote operations, all the work on regulatory with the different cities so that we can do this at scale and truly be, again, best-of-breed. And I believe the only way you can really do that is if you actually get in there and do all of the things yourself. And so that's what's happening this year with DoorDash Dot and also our broader autonomy program.

Operator: Your next question comes from the line of Eric Sheridan with Goldman Sachs.

Eric Sheridan: As we get deeper into 2026, any updated views around either the depth or the duration of some of the strategic investments, especially in the platform that we've talked about over the last couple of earnings calls. And more importantly, any updated views on how the tech replatforming might position you for different forms of innovation than you envisioned 6-plus months ago?

Ravi Inukonda: Sure. Eric, I'll start. Tony, feel free to add anything. Look, I think we talked about 2 calls ago that we are investing several hundred million dollars back into the platform. Obviously, the largest component of that is our global tech infrastructure stack. It's going well. The biggest component of that is just being able to design and map all the domains, which is what the team has done over the last several quarters. That part is done. Now, we are focused on execution. We're starting to see production traffic go through. We're already starting to see some early benefits come through. I think on the cost side, Eric, which I think was the second part of your question. Look, I talked about the fact that this was the biggest component of the investment that we're making. My view on the overall quantum of dollars that we are investing behind this has stayed the same. It's largely in line with what I had expected 2 quarters ago. And both the program from an execution perspective as well as a cost perspective is going well. And finally, to your point around benefits of this, look, ultimately, the benefit is going to accrue in terms of us being able to do more, us being able to release features earlier, the feature development velocity is going to improve, which will ultimately result in retention, order frequency and unit economics increasing. That was the goal for this. We're starting to see benefits, and I feel good about where the trajectory of the overall program is.

Tony Xu: Yes. The 2 things around your second part of your question, Eric, that I'd add to what Ravi said about innovation is, one is really around the velocity and the second is around the quality. So the velocity increases for the simple fact that instead of shipping one feature, which if we were to do it today, we'd have to ship 3 separate times across DoorDash, Wolt, and Deliveroo. We'd only have to do that once. That's the velocity comment. But the second point is really around the quality in which we can see benefits. What we're doing by choosing to build a new tech stack versus, say, just replatforming a couple of different brands into the same tech stack that we currently have is that you get to kind of take the best-of-breed experiences from different brands and different products and put them into a new product that all 3 get the benefit from. For example, one of the things that we've learned is that there are different logistics challenges in places like London, as an example, or cities in Europe that are a lot smaller, a lot tighter, not always perfectly grided like some of the cities in the United States or in other parts of the world. Perhaps there are older cities historically, and therefore, they weren't really meant for driving under any circumstance or a condition in which you need different logistics approaches, and we can borrow and take the best of what we're seeing from our European operations and bring those over here to the U.S. whereas in the U.S., because we have larger physical geographies that travel longer distances, that have perhaps a greater retail network, that has a larger catalog of items, those are advances that we get to see that we get to port over to Europe. And so that's what I mean by quality. And so I'm pretty excited about -- we're on track, which is great news when you're taking on a project as large and ambitious as the one that we're thinking about. But I'm very, very excited that not only are we already seeing some velocity and quality wins across all of the brands but I think there will be a lot more to come as we actually roll this thing out.

Operator: Your next question comes from the line of Youssef Squali with Truist.

Youssef Squali: Maybe just following up on the prior question and maybe looking at it more from a competitive lens. Can you maybe talk a little bit about what you're seeing in Europe, in particular, maybe Northern Europe with Uber becoming a little more aggressive. There is a line of thinking out there that maybe as you guys are going through your replatforming, it may make you potentially a little more vulnerable to competition. So maybe if you can comment on that. And then, Ravi, thank you for quantifying the support to drivers for Q2. I think you said $50 million. Obviously, we don't know how long this thing is going to last. But is $50 million a good run rate to assume going out for the rest of the year, just assuming we have status quo on the macro environment?

Tony Xu: Yes, I can take the first question, which is around kind of our competitive position in Europe. I mean we've never been stronger is the short answer in Europe. I mean, if anything, Deliveroo is seeing the highest growth rate it had in the past 4 years, and it's actually been reaccelerating in growth each of the months in which we've been operating it. Wolt has seen the highest share performance in each one of the countries in which we operate. But I think those are just outcome metrics. And candidly, they're not things that I stare at all the time. I mean what I'm looking at instead is kind of what I was saying in some of the earlier questions around what are the improvements that we're actually shipping for our different audiences. If we're seeing logistics improvements, like how is that translating into lower wait times at different stores, higher accuracy of picking or faster delivery. And if we can continue executing the way that we have, I think the share performance, the reaccelerated growth, I think, is only going to continue. And again, it goes to the DoorDash story, which is how do you build what's best in breed. And if you can continue building what's best in breed, I think customers will continue voting with their wallets and they're voting DoorDash.

Ravi Inukonda: Youssef. On the first one, I'll just question your premise because if you look at the underlying consumer input metrics, whether it's users, order frequency, we talked a lot about subscription in the press release. Look, I mean, we are seeing accelerated growth in subscription. Users are growing. We're gaining share in the majority of the markets that we're operating in. The other thing I would offer is if you actually look at the overall MAU growth in the industry, majority of that is being driven by DoorDash. And that should tell you the business is doing really well, both from a demand as well as an underlying improvement in the customer metrics perspective, right? To your second point around the impact from a gas rewards perspective, look, roughly the impact of that is about $50 million in Q2. I'll say a couple of other things. We did have to find offsets in the business. We will push out some investments into the first half into the second half. Our goal is to make those investments in the second half of the year. To your second point around, are we going to extend it? Look, we've not made any decision. Obviously, we'll monitor the situation very closely, and we'll do what's right for the business. But that said, my broader view on EBITDA for the full year has not changed. If you look last couple of quarters, I talked about the fact that I expect overall EBITDA margins for '26 to be slightly higher compared to '25, excluding GOV and GOV to produce roughly about $200 million of EBITDA. That view has remained very consistent. That view has not changed. If we do decide to extend the gas rewards program, we'll find offsets in other parts of the business in order to make sure we still feel good from a top line as well as a bottom line perspective.

Operator: Your next question comes from the line of Nikhil Devnani with Bernstein.

Nikhil Devnani: Tony, in a world with, I guess, AI workloads and a more productive workforce, is your mental model for headcount growth and even organizational structure for DoorDash changing at all?

Tony Xu: Yes, it's a really good question. I mean, like, in short, I mean, the answer is yes. And the longer answer is we're trying to figure out what that really looks like because we're seeing, for instance, a lot of productivity gains right now from AI, about well north of half of our code, as an example, probably closer to 2/3 of our code is written by AI today. But that doesn't alone kind of articulate how workflows and team setup ought to change, right? It means that we're being more productive. We're shipping more code. But the ultimate question I have is, are we actually delivering better outcomes for customers because at the end of the day, that's the only thing that really matters. And so we're in that period where we're seeing productivity gains. We're trying to figure out how do productivity gains now translate to what team setup should look like. That's the phase where we're at. And I think if we -- the top priority for us right now is definitely making sure that we can get all teams onto a single tech stack. The second priority is to make sure that I think everyone in the company, not just the engineers, but everyone in the company, I think, is as AI capable as anyone else. I think then we can start thinking about what are the actual workflows that have to change to actually truly deliver things faster. Like right now, we're delivering features faster. We're delivering like projects faster, components faster. But I think the customer holds us to a higher bar than that, which is can you actually deliver outcomes much faster. And I think that's a tricky question that all companies, ourselves included, are wrestling with right now, and we'll figure it out.

Ravi Inukonda: Nikhil, it's Ravi. I'll just add to it like just very similar to what Tony talked about, we are using it across the board. We are seeing productivity improvements. Look, the goal for us from a productivity improvement perspective is just has always been, right? We want to do more with more. We want to try to drive more features. We want to do more for our audiences. We want to do more internally as well. Ultimately, the way we think about it is how do we channel the productivity improvements into ultimately developing more features. But if it's purely from a modeling perspective, Nikhil, that you're trying to think about it. Look, I would expect from a near-term model perspective for OpEx to roughly be in the 2% range that I've talked about before. We are being very judicious. We're being disciplined. Goal is to generate leverage on it just like any other part of the P&L over time.

Nikhil Devnani: That's helpful. And Ravi, if I could just follow up on, I guess, the order growth dynamics in Q1 as well. Could you just elaborate a bit on the deceleration there? Is that just weather? Or are there other things you want to call out? And how are you thinking about that as you think about the Q2 guidance you've given for GOV?

Ravi Inukonda: Yes. Look, I mean, Nikhil, the question broadly is around the consumer demand on the platform. I mean, look, demand continues to be quite strong. The impact purely from a winter storm perspective is roughly about 1% on a year-over-year growth perspective from a GOV basis. Look, when I look at the underlying demand, Nikhil, it continues to be very good, right? We've talked about MAUs reaching an all-time high. Order frequency is growing. Subscription had a record quarter, in fact, across the board across DoorDash, Deliveroo as well as Wolt. What we're seeing in the business is member growth has accelerated on a year-over-year basis, that follows last few quarters where the member growth has been quite healthy. We're seeing that both from sign-up as well as an overall retention perspective. When I look at the other parts of the business, we're gaining share. New verticals is continuing to do well. We were volume share leaders in Q4. We've continued to extend that. Even across the board international, we touched on Deliveroo acceleration as well as the rest of the international portfolio also growing. Q2 is off to a good start. We feel good about the demand patterns that we're seeing on the business.

Operator: Your next question comes from the line of Deepak Mathivanan with Cantor Fitzgerald.

Deepak Mathivanan: Tony, so on groceries, in the last few months, you've added a lot of new partners. Can you talk about the trends in the business broadly, maybe in terms of penetration, how use cases have evolved, potentially some color on growth and maybe also where the unit economics have seen the biggest gains in? And then similarly, DashMart Fulfillment Services is also another big area of focus this year. Where does it currently stand in terms of where you want the service to get to ultimately before it starts becoming another incremental key growth driver?

Tony Xu: Yes. I mean those are related questions. So I mean I'll start with just like where grocery is at today, which is -- I mean, you kind of said it, which is -- it's pretty much at record highs for us, right? We were -- we became the share leaders by volume last fall and/or last winter, and it's kind of continued to go in one direction amidst a lot of activity, I would say, in the field. And you're right, in part, it's because we have added a lot of grocers and we like the trajectory of the selection we're adding. But we're also improving the service experience. So it's not just adding more and more selection. But I always ask myself, why is grocery not a lot bigger? Why shouldn't it be even bigger than, say, restaurant? Well, it's because the online delivery experience is just not yet good enough compared to the offline experience, right, of buying it for yourself. And -- but we're really closing that gap. I mean the team deserves a ton of credit for making us a lot more accurate, making us a lot more affordable, making basket building a lot easier, making customer support better, making the experience easier for shoppers and a lot -- literally tens of thousands of little things over the last 6 or 7 years, I think, are accumulating. But it's still, I think, reasons for me that over time, if you truly want to marry the best possible selection, which is every store inside your neighborhood with the best possible quality, which means every -- you get exactly the item you order without any substitutions or any changes and obviously, certainly no out-of-stocks or canceled items or canceled orders, I think you do have to work the fulfillment problem, which is where DashMart Fulfillment Services comes in. And so there, we are trying to build an inventory management and fulfillment setup with all of the grocers and retail partners that we work with. And I think if we can do that, I think then finally, you can actually unlock what is truly a magical experience where it's more similar to restaurant delivery where, yes, there might be a small premium you pay, but at least you get exactly what you ordered, which is not the experience today. In terms of where DashMart Fulfillment Services is, we -- I mean, we're doing it with a handful of grocery and retail partners today. That's kind of where it is. And if you think about that journey, I mean, we're trying to work with grocers and retailers who for decades now are used to running their supply chains and their stores in one particular way. Now we're introducing a second way -- and so there's a lot of things that you have to figure out in terms of technology, people processes, the interaction of business models and kind of everything in between. We like what we see with a handful of partners we have, but kind of in the spirit of all things at DoorDash really, where we really want to make sure we nail the experience before we scale it because I think this is both quite disruptive in a positive way to the customer experience. It's also quite disruptive to how retailers are used to working and buying and running their own businesses for so many decades that we have to make sure that we get it fully right end-to-end and then we can replicate the playbook.

Ravi Inukonda: And Deepak, on the unit economic side, I think that was part of your question. Look, we made a lot of good progress. Last call, I made the point that we expect the overall new vertical portfolio to be gross profit positive in the second half. We're trending well towards that. Look, we're not worried about what the profitability profile of this business looks like. It's something we understand quite well and what we need to do. It's not that there's any structural change that we need to make happen. It's just continued execution on a number of lines up and down the P&L. What we're truly focused on is how do we scale the business. Q4, we talked about the fact that about 30% of our monthly active users order from categories outside of the restaurant. We truly think that could be 100% over time. And that's going to come with a lot of improvements in selection, quality, improving the underlying product. And when I look at the underlying consumer metrics, I mean, look, order frequency has improved. Basket sizes for mature cohorts have continued to improve, which means that people are using us for more use cases. And over time, what we're seeing is the underlying order rate also continues to improve. These are all good signs, which is driving both the growth as well as the improvement in scale, which will ultimately drive the unit economics in the business as well.

Operator: Your next question comes from the line of Josh Beck with Ray J.

Josh Beck: Maybe more on the cost side. Ravi, you kind of mentioned the $50 million gross cost as you kind of look to find relief for those investments, maybe what are some of the big topics that you're kind of looking to uncover there? And then going to your -- some of your points on new verticals, certainly a very nice watermark to achieve gross profit breakeven. To get to the next milestone, what are going to be some of the really important elements? I mean, generically, it seems like within new verticals, advertising is a bit more of a weighting factor there. So just kind of curious how we think about maybe some of the real important drivers beyond where we're scaling into the second half.

Ravi Inukonda: Josh, I'll take the first one. Tony, why don't you take the second one. Look, I think, Josh, your broader question is around costs, gas rewards and impact on the model for the rest of the year, if I understood it correctly. Look, the impact -- I'll talk more broadly, right? Q1, we obviously had the impact from both winter storms as well as the introduction of gas rewards. Q2, we did extend the gas rewards program. Rough impact of that in Q1 was about $50 million. The projection for the impact in Q2 is also going to be about $50 million. Like I said earlier, we did find offsets in the business. Look, for us, it's a very dynamically managed business, right? We take our plans very seriously. We look at input metrics to make sure we're doing the right investments. We did have to push out some investments in H1 in order to make room for this. And my expectation, and we are fully convicted that we are going to make these investments in the second half of the year. If we do decide to extend the program, our goal is to find offsets just like we did in H1. But my view on the full year EBITDA has not changed. Look, what we've said about a couple of quarters ago is overall, 2026 EBITDA margin is going to be higher slightly compared to '25, excluding GOV. That view still stands. When I look at the trajectory of the business, I would expect second half EBITDA dollars to be higher than first half, second half EBITDA margins to be higher than the first half, largely similar to what I had expected at the beginning of the year. Overall, right here, when we sit, like we look pretty good from a bottom line perspective for the rest of the year as well as the demand on the platform continues to be strong as well.

Tony Xu: Yes. And with respect to your second question about what else do we need to do to make -- to achieve higher levels of profitability within grocery. I mean the short answer is more of the same, which is there are -- we're not trying to rely on any one source of revenue like ads, for example, to make grocery profitable. We don't need to. We fundamentally have created, we believe, a lower cost structure that allows us to make grocery delivery profitable. But it's just not good enough yet, right? I mean if you think about it from the perspective of the customer, not the perspective of our P&L, we still need to be more accurate. We still need to have more items available even from existing stores, and we need to do it at better and better price points. And I think if we keep doing that, I mean, you already see in our cohort behavior. It's not true for all across the business because we're still gaining a lot of new customers. In fact, we gained about 1 in every 2 new customers that comes into the industry for grocery delivery for the first time. But you see with cohorts over time that they actually buy bigger and bigger baskets and achieve the profitability milestones without any unnatural, shall we say, or overreliance on any one cost or revenue driver. And so that tells me that at current course and speed, it will get there. And the question is like how do you get there faster? But perhaps most importantly, how do you actually unlock a much bigger industry? I mean grocery delivery fundamentally should be as large, if not larger, than restaurant delivery. It's just that the product isn't good enough yet. And so if we can -- we already are leading from what we've been told by some of the top grocers in the country in terms of quality, but we still think there's miles to go. Perhaps we brought some innovations to the market, but we think that we have to keep innovating on all things, accuracy, price points, and we have some interesting ideas on how to do that. But we don't have to do anything unnatural or over rely or perhaps even rely at all on any source of any one line item to make the math work.

Operator: Your next question comes from the line of Brian Nowak with Morgan Stanley.

Brian Nowak: I have 2. The first one, Tony, in the letter, you talked about making some new tools that helped design that streamline the merchant onboarding process. Can you just sort of talk to us about areas you've made the most progress in bringing on new merchants and sort of more inventory per merchant? And what are some of the technological advancements you're still looking to make to really make that easier to get more of those bananas and avocados that you talked about earlier and even carving knives. And then, Ravi, one for you just on the replatforming. Now that you say that you have live production traffic ramping up across all 3 of the global marketplace brands. Does that mean that you're sort of running all 3 tech stacks now, so we're burdening the P&L with the Max costs and then we should start to turn some of those off in the back half? Or how does this sort of triple platforming down to one platform time line work?

Tony Xu: Yes. I'll take the first one, Brian. Yes. So you're right. We continue to ship a lot of different tools to make it a lot easier to work with us or a lot easier for customers to find what they're looking for or Dashers to perform deliveries. On the specific question with respect to the merchant tools, a lot of what I found AI to be helpful, especially now with more powerful models that can reason in a multi-turn kind of fashion is that you can start looking at repetitive processes that kind of are stitched together and actually get them done with perfect quality every single time, and you can actually do that with just an agent. Whereas I would say even perhaps 6 to 7 -- or 6 to 8 months ago, this was even less true because you would kind of need to build a lot of backup or redundant systems to kind of make sure that agents don't just kind of go off the rails and actually can finish the task. And so that has happened with something like what you said around with onboarding as an example, where whether it's helping you with your menu or your catalog as a restaurant or a retailer or with your photos and your metadata and kind of the annotation of that data. All of these things are effectively repetitive tasks, if you will, in which you can create agents and stitch them together and to do that in a really, really productive way. And I think like with all things, the removal of friction increases activity and activity -- increased activity increases the business that we get to do together. And that's kind of what we're seeing with -- so we are already seeing benefits to the P&L from some of the AI work that we're doing. Some of it is on our own products like the AI ordering agent stuff I mentioned on a separate question and some of it is on tools that you're talking about, whether it's related to merchants or customer support or Dashers. So we're already seeing that. And with respect to like things we still have to do, you were mentioning about like all the inventory inside of a city. Well, we still are just a tiny fraction of all items sold or even represented, I would argue today, DoorDash is, the collective. And that's becoming more interesting now as some of those items are also different when it's an in-store shopping experience, for example, right? Some restaurants, as an example, offer different in-store products and experiences and services that they don't offer in the kind of takeaway or in the offline world. And so there's a lot of things we have to document. That's one thing we have to go and do. The second thing we have to go and do is build structure and cleanliness out of what is inherently very messy and constantly changing, which is a challenge, I would argue. But if we can do both of those things across every one of the categories, as we kind of march from restaurants to grocery to different categories within retail and to do that through the merchants channel online, the DoorDash channel online and the merchants channel offline or in-store, I think that just builds a really rich data set that's kind of, a; nonexistent anywhere, b; extremely valuable for the merchant to have a full view of all the different types of customers that they may want for different occasions, and c; I think, really interesting opportunities also for DoorDash to build both products as well as businesses.

Ravi Inukonda: Brian, on your second question around the global tech side, right? I'll make 2 broad points, and I'll get into the P&L dynamic question. Look, I mean, I think the team has done an incredible job. I mean this is a massive project. I have to give kudos to the team. It's going according to plan. Really happy with the progress that we're making. Even on the cost side, my view on the overall cost has been very similar to what I talked about last 2 quarters ago. So both on progress as well as cost, I feel very good about where it is. I think to your point around the mechanics of the P&L, look, I think I talked about the fact that there is a portion of the spend, which is redundant in the sense that we are going to run all 3 tech stacks in parallel while we're working on the global new tech stack. That is going to phase in. It's going to phase out. My expectation is majority of that will run through '26. Maybe some portion will bleed into early 2027, but that will bleed out. And hopefully, that should give you a mechanic of how the rest of the P&L is going to work for the rest of the year.

Operator: Your next question comes from the line of Justin Patterson with KeyBanc.

Justin Patterson: I saw you recently launched workplace catering for DoorDash for Business. Can you talk more about how you're thinking about that opportunity and what you see as some of the key challenges towards scaling this?

Tony Xu: Sure. DoorDash for Business, I think, is off to a very, very great start. It's actually only something that we've really recently focused on in the last few years. But DoorDash for Business really is a set of or a suite of a few products. And there's really 3. You talked about one of them, which is catering. There's also Meal Manager and then there's also corporate solutions related to DashPass as well as group ordering. And the idea is when you're a company or an organization, it doesn't have to be a company. It could be a nonprofit or a government institution or a school. As long as you're serving multiple different use cases, sometimes it's a group meeting with just a few of us, perhaps sometimes you're hosting an event in which you need catering. Sometimes you need individual meals as kind of your sales teams perhaps travel to do different things or client demos, things like this, you're going to want to work with one place ideally, which you can see everything in one view and offer your organization the best-in-breed selection, price and quality. And because we offer what we believe is the best-in-breed in price, selection, quality and service, DoorDash for Business is kind of naturally growing really very, very quickly. So that's something that we're doing. But I think the biggest challenge to your question around something like catering is, it's really how do you solve the kind of perennial hard problem of cooking for a large group of people. It sounds really simple, but if you think about it, as you cook for yourself and then you start adding the number of guests around the table, I would argue that logistics problem is a bit of an exponentially difficult problem as you kind of increase the count of guests. And so the challenges are numerous. It starts with understanding, well, things around kitchen capacity, then understanding things around menu design, then understanding things around staffing, then understanding things around the logistics operations. And we kind of have to do all of that. And I think to actually truly create the industry because I think the industry by itself is perhaps somewhat limited because not every restaurant is built as a manufacturing facility that can actually cook up to the needs of a larger organization or even a team of people. It's really working hand-in-hand with merchants and Dashers together to co-create that solution and hopefully create a very large industry.

Operator: Your next question comes from the line of Lloyd Walmsley with Mizuho.

Lloyd Walmsley: Wondering if you can give us an update on what you're seeing in the ads business on both the 1P basis and the syndicating ads outside of Dash? And then second one, Tony, earlier, you talked about miles to go in terms of improving the user experience in grocery. Just wondering if you can elaborate on some of the things you're doing, if you found any big unlocks or anticipate any big unlocks to kind of drive a step function improvement in the grocery experience that can help you guys penetrate deeper with your customers?

Tony Xu: Sure. Maybe I can take both and feel free to add in, Ravi. I mean on the ads question, it's never gone better for us in the world of ads. Ads is at a record high, continuing to grow extremely fast compared to any previous year. And I think the kind of the recent trajectory or kind of continued strong trajectory really comes from the fact that our team deserves a ton of credit in cracking the code, not just in solving problems for SMBs, whether they be restaurants or retailers, but also larger advertisers, both in the restaurant world as well as in the retail world. And then the other unlock has been cracking the code on CPG advertisers, where I think just there's no one thing. It's like a relentless checklist of just making the product a lot better for advertisers and delivering upon 2. I mean, really competing objectives. One is you have to deliver the best return on ad spend for advertisers, which we do. And the second is you have to deliver the best consumer experience where you don't spam people. And I think we have a much lower ad load than some other platforms. And I think the teams have been just working really, really hard to balance those 2 competing objectives. And then -- so I think the opportunity besides kind of scaling some of the unlocks in ads as well as discovering some of the kind of off-site stuff that you're talking about, which also includes things around all of our in-store activities in addition to our work just buying on behalf of advertisers off of DoorDash. I think there's a very large runway for the ads business. I think your second question was about just the early innings of grocery. I mean, we've been at grocery for about 5 years now. And yes, I am, on the one hand, super proud of the team and becoming the volume leader in which consumers both shop on as well as new consumers find out about grocery delivery for the first time. But on the other hand, yes, I actually do stand by the statement that we are miles to go on building an experience that I think can outcompete you going into a grocery store and buying the items yourself because that is still the winning product if you just look at all of the data out there. It doesn't mean that we're not growing extremely fast. It does not mean that we're not making a lot of improvements. It does not mean that we're not gaining share in doing so and improving the profitability while we do it, but I think there's a lot of work to do. So a lot of the work has to do with just continuing to build a cost structure that allows you to offer items at around the same price as in store and delivering with perfect quality. I think the hardest problem to solve in grocery is that because consumers, all of us, when we go into grocery stores and move items around and because of how supply chains and inventory systems and payment systems don't necessarily always talk to each other and kind of how grocery stores are run and how they were built both historically and then as well as they've made the move into e-commerce, it's really hard for them to know where things are. And that's still the fundamental problem to solve. And so we've done lots of things already in that space that were -- that we've pioneered and are really proud about. There's a long ways to go on just scaling that work to all of the stores in which we work with, not just the ones in which we've tested. But we also have to kind of do the next hill climb, if you will, to make sure that we can, again, achieve perfect quality at the prices that you would expect and be happy to pay and do it for every single item every single time.

Ravi Inukonda: And Lloyd, on your first question on ads, right? Like, the question is around -- if you're thinking about it from a flow-through perspective, look, it's growing, it's obviously having an impact from a margin as well as a profitability perspective. But the way we think about ads is very similar to how we think about the rest of the business, right? An ad dollar is very similar to how we think about improvement we generate from unit economics. Ultimately, the goal for us as operators is to find opportunities to reinvest that back in the business, ultimately to drive long-term free cash flow production. And that's largely what we're doing with advertising or any other efficiency that we generate in the business.

Operator: Your next question comes from the line of Justin Post with Bank of America.

Justin Post: I just want to follow-up on advertising. How do you think about integrating that with agentic capabilities on your own platform? And is there any way you could generate ad revenues on agentic platforms on other platforms?

Tony Xu: I'll take that. Well, ads are just a means to connect consumers with merchants who are hoping to be discovered and making sure that you do that in the perfect possible way. So with respect to agentic commerce, I mean, that's just one way of shopping. And so I don't think it will change in terms of our ability to advertise. It may increase some of the opportunities in service areas, but I think a lot of that remains to be seen, just as I don't think the ideal agentic shopping experience is just going to be a chat assistant. I think it's going to take on various forms, and we're iterating on that. I think with respect to what happens with ads on third-party agentic sites, I think you'll have to ask them.

Operator: And this concludes today's Q&A session. This also concludes today's call. Thank you for attending. You may now disconnect.