Uncapped #55 | Tony Xu from DoorDash

0:00There are battles for attention, you know, bits and, you know, battles for what's happening in the physical world, you know, for the atoms and that we largely occupy ourselves in the second category. We are in the war for atoms because, you know, one day ultimately, like, what's the point of having a personal assistant if it can't actually do real things?

0:24Tony, thanks so much for doing this. One of the things I wanted to start with was this concept that, if you'll indulge me on it because it's something I've been thinking about a bunch, kind of around the idea of like AI spend at companies. And I think the trigger sort of anecdote that I saw that I'd like to get your take on was Uber was basically saying, you know, in the first few months of the year, they blew through their whole AI budget. They were just, you know, rampantly consuming tokens. and what did they get out of it? Did they get more rides? Did they get more drivers? Did anything, their margins improved?

0:56What happened? And I think they were kind of saying it wasn't obvious and you're obviously the leader of a company that's super metrics oriented and has executed super well. So I assume you must think about this, but I'm just curious kind of at a high level, how you think about what is a large rising line item? And as it continues to rise, what do you care about? Yeah, I think every company right now is trying to figure that out where, you know, the ultimate goal of any technology is actually to hopefully solve a problem and actually make things cheaper. I mean, that's the that's the goal of technology.

1:32But as you know, especially when you have constrained resources like compute in the case of these large LLM companies, you know, sometimes you don't get the timing of these things always exactly when you can deliver the outcomes. or you don't get the cost profiles or the efficiency profiles to serve the technology to match exactly when you can deliver the customer outcomes. And I think that's the world in which we live today. So I think every company is trying to figure this out. In terms of how we're thinking about it, first and foremost is what customer jobs can we actually solve? And so I do think when you have new technologies, there's always going to be this period of inefficiency and, you know, almost like discovery where you got this new toy.

2:22You don't know exactly what you need it for. You probably kind of know that you don't really need a, you know, frontier model to know what the weather is perhaps. But on the flip side, you don't also know the limits or the ceiling of what the technology can do. And you kind of want to know that. And so the way, at least I think about how you do this somewhat efficiently, even though by definition, you're going to accept inefficiency in the discovery process or in the invention process. You want to do it in the most contained set of ways that deliver customer outcomes. So you actually want to put customer outcomes and start there and then actually give teams, you know, as many shots on goal towards those outcomes as possible.

3:07It doesn't mean you get there, but at least it's directed and it's intentional and it's not, you know, entirely just YOLO. There's obviously some of that. But I think if it can be directed towards, you know, in our case, consumers, merchants, dashers, we've actually found some success. Well, it's funny because like as a software company, which is like where so much of, you know, the AI productivity is happening right now. It's actually harder in some ways for them to measure the like end outcome of the work that their engineers have done than maybe yours where you can say, hey, I can measure. Did we complete more deliveries?

3:43Did we get more restaurants? Do we have new users or all the things that you're tracking? I was reflecting that over the last few years, you've kind of watched this egg move through the snake in the pipeline from first, there's the build out and there's chips and then you get data centers. And then it's like, oh man, are people ever going to use this? And now they are using it and the token spend is ramping. But now it's like, we've got to turn the token spend into burritos at people's homes. And like you're in like as good of a position to see that as anybody. So like are you like, you know, we're going to have certain teams try to move the metrics in a market with AI or is it kind of just like bottoms up, let the teams explore whatever they want with AI?

4:22Yeah, it depends on the team. So if you're product teams, you're exactly right. You can direct teams, I think, towards metrics that matter to customers. If your merchants on DoorDash today, they're onboarding 35 % to 50 % faster because we can actually have AI produce all of their catalog or their menu in the case of a restaurant, help edit your photos, help set up what might be the best way to describe yourself, especially if you're a new retailer coming online into a city. So there's metrics like that that are very trackable, measurable, have immediate positive benefit in customers. Similarly with dashers, you know, we've had AI help detect, you know, both fraud as well as safety incidences before they occur.

5:17And you can stop those and take preventative action if you knew that was actually happening. So there's things like that where you can, you know, get immediate benefit. You know, so immediate customer benefit. But you said something earlier in your question, you know, well, what about like software teams, for example? And I think one of the things you notice there is that only certain parts of a software engineer's day is writing code. And it's great that like we have models today to write the code for us. but that helps maybe the whatever 25 30 50 percent of our time that's actually you know shipping code but what about everything else you know that has dependencies in product reviews design meetings you know alignment with business teams etc etc etc that also has to change if that doesn't change and come together you're not going to be able to just have perhaps the productivity gain that you hope to have.

6:15Yeah. And I think that's what companies like ourselves, but I'm sure a bunch of companies across the industry are trying to figure out in getting those workflows right so that you're not just AI native from code development, but you're AI native in how you actually operate. Have you done anything to track different engineering and product teams inside the company comparing how much more productive they become with AI usage, like ones that are using it more and less heavily and things like that? Yeah, I mean, you get all these facts, and I'm sure, you know, as with all things, there's always a distribution of outcomes.

6:46You know, maybe the average is 50 % more productive. But, you know, you have engineers who might be 35 times more productive. It's crazy. Right? And it is crazy. And it's always fun to study that distribution. But, again, like, I think that, you know, what's the value of that? The value of that is really knowing what the art of the possible is, especially in understanding your workflows. But what, you know, someone like I have to do is I have to think about, well, how do you create that working environment for everyone? You know, because it's great that we have, you know, engineers who are 50 times more productive.

7:21How do you actually get everyone to that new ceiling? Exactly. And the question is, is that an observation and somebody's going to be far out? Or is there a learning there that you can go replicate it? Yeah, exactly. And I believe it's the latter. Because with all things, especially if they're newer, you're always going to get a distribution of outcomes. Right. There's always, you know, in sports, in academics, like there's always a distribution of outcomes. But that doesn't mean there aren't classes and teams and, you know, good things that you can teach as a coach or as an instructor so that you can get the rest of the class, you know, to move to the higher plane.

7:59Yeah. You know, for a lot of companies, when AI came around, it was either like the most like, you know, kind of petrifying new technology ever, or it was like the, oh my God, thank you, Tailwind. And I feel like for DoorDash, it was probably like neither. And, you know, tell me if you object to this, but like, I would say like, you know, you're mostly operating in the world of atoms, not bits. you've got this two-sided, three-part marketplace with restaurants and dashers and consumers and all these things. And you're very related to cities and all of that stuff. And so I would think it's both insulated from AI in a good way.

8:37And then you also maybe didn't, it's not like your business started growing 400 % faster because this AI thing showed up and it wasn't their history. So how has that been? Is that right? Is that about what it's been for you? Has it been kind of just insulated and that's both good and bad. Yeah. I mean, we're not selling, you know, burritos or Nike shoes or groceries as fast as tokens, you know, at some other companies, but yeah, look, I mean, I think you're largely correct. I remember, um, I don't know. I think this was 2021. This is, yeah, this is 21 before the arrival of chat GPT. We were playing around with some of these models.

9:15I mean, I think it was like, it was like GPT 2.6, something like that. Yeah. Right. And just kind of getting a sense of what the world could look like. Um, and obviously, you know, the models back then are very, very different from the models of today. Um, but even then we kind of had this description of the world where there are, you know, battles for attention, you know, bits and, you know, battles for what's happening in the physical world, you know, for the atoms. And that we largely occupy ourselves in the second category, right? We are in the war for atoms and moving things around. And if we can do that and we can build a catalog, for example, for where every item exists inside of a city or every parking spot exists or all of this information, we can, as the kind of company's battle for attention, start maturing that we can really partner and work together in very productive ways.

10:16Because, you know, one day ultimately, like what's the point of having a personal assistant if it can't actually do real things for you? Right. Exactly. Right. And so, and that's kind of how we thought about it four or five years ago. It's actually funny on that point. Like, you know, I, it, it strikes me that like one of the, it's very cool, but you know, one of the things that I think there's room for improved is that so much of tech is just like typing on our screens and we're all living in the computer. But at some point to make lives better, you need to, you know, move stuff around. You need education.

10:46You need stuff that happens, you know, in a hospital. Yeah. Healthcare. Exactly. Yeah. You know, and so it's, you need, you need, you know, real estate to be developed. It's like all this kind of stuff that's, but it's all physical at the end. Yes. And so in some ways it's like all the software ultimately does need to be in service of stuff. Yes. Yeah. Yes. I mean, and this is, And this is exactly why, you know, what is I guess it would have been five years ago at this point. The kind of the strategy was very much, no, play the game that we're meant to play. The game we're meant to play is the game of atoms and, you know, be best in class at that.

11:20And not only is that good for our business, but I also think it's great for our relationship over time with these companies as they, you know, finally build out, you know, some of their assistants and build out super intelligence. and build out agents that can actually do things versus what they currently do because we naturally will need one another in order to be useful to the communities that we serve. Has there been any impact on recruiting or things like that when everybody's losing their mind on AI and you're in this different thing that you're like, I know this is really important and it's going to become even more important eventually, but have you had to work through any of that?

11:57Yeah, I mean, it's definitely a battle. We are an applied AI company, and we have been pretty much since day one of the company. Whether you want to call it math or machine learning or now LLMs, we can use whatever terminology we want. But as the years have gone by, we are an applied technology company that also solves real-world problems. So yes, it's certainly difficult when you're competing in places where activity is really hot in machine learning, in folks who are looking to do applied research. It's very competitive right now. Yeah. Are there things like as you think about your next frontier and where you want to go and you look at the AI stuff, it does seem like some of it is pretty direct.

12:41Like, for example, I could imagine like robotics maybe could matter at some point. I could see self-driving matters at some point. What are the next parts of AI outside of the LLM on the software side that you care about? Yeah, well, I mean, I would say first, before we step away from some of the LLMs, we do care about what LLMs can do. And I think they have gotten materially smarter and more powerful, not just helping with coding, which may be the most prevalent use case today, and certainly in the most prevalent spend consumption. Well, it's definitely a good input to you for building stuff.

13:17A hundred percent. A hundred percent. But even, you know, for instance, you know, one of the things that, you know, we always ask ourselves is like, how can a technology actually improve outcomes for customers? So if you take, as an example, the DoorDash app, you know, we've grown a tremendous amount over the last even five years. Forget like, you know, the 13 years we've been doing this. just in the last five years, kind of the same arc in which some of these LLMs have grown up. DoorDash has moved from one product category, restaurants, one market, US, into virtually every retail category, the leader now in deliveries of grocery, convenience items, alcohol items across 40 countries.

14:03There's lots of positives with that. One of the challenges for a consumer those, boy, it's a lot harder to use an app like DoorDash because you now have restaurant things in there, you now have grocery items in there, you have retail items in there, you now have the ability to make reservations or get deals inside of restaurants. It's getting more complicated to actually use our product. LLMs really can help with that. And just as I think there will be personal agents soon, the DoorDash app should be a personal agent. It should be a personal agent to help you do anything inside of your city, right?

14:39I can't think of any other product and any other utility, frankly, greater than the number of connections that you can have between you and the businesses inside of your city. And so those are the kinds of things that you'll see us launch just with LLMs, right? You should be able to have an easier time actually doing research on what items you want to buy or what might be great recommendations for you based on all the history that we have across tens of billions of orders on you. And you should pretty much be able to get anything to your house and not have to shop for it. Yeah. Or have the choice of getting it later when you're home or because you may be away or the next day because it's better for you.

15:19You're exactly right. And so I do think that there's still a lot of excitement just in the LLMs, but you're right. If you want to go beyond that, there is a lot of work around the physical space. So for example, we've been working on autonomous vehicles since 2019. And a lot of that started with kind of traditional systems, kind of like with machine learning, yet traditional ways of thinking about how to use those techniques to build things like recommendation systems. LLMs kind of put a complete new spin on it and didn't really require any of those techniques. Something similar is happening in the physical world where it used to be if you wanted to perhaps drive autonomously inside of a market, a lot of what you would do is actually build mapping systems and almost like heuristics and rules to create an engine in which you can make good decisions and kind of weigh them in real time.

16:22Or perhaps you can you know put all of this into a neural net and and take similar techniques that some of the LLM companies are using and actually make you know even faster and better decisions right and so there are things like that that are going on that allow us to do some of the work that we're doing with autonomous delivery you know faster yeah as an example you totally this is now like going way back in time but are you surprised how much people are willing to consume on apps you know obviously when you started the company in 2013 this was the plan but like has it gone like a lot further than you expected you know like people really will just like order you know a burrito for you know a lot of money you know and probably like and it's worth it like it's turned out that it's worth it to people but like was that obvious to you in 2013 that it would go this way no it was not obvious like you would get like 35 bucks for a burrito For example, when we launched our first partnership with a national brand that was July of 2015, I believe, with Taco Bell, I was actually quite skeptical whether or not it would work.

17:33On the one hand, you have this legendary brand that's never offered delivered before, being offered for the first time. That was the bull case version. On the flip side, you're right. You're gonna pay a premium you know to get that order yeah, and and and Which wouldn't have been intuitive because you said well I can get the you know I can I can get this order for four dollars in store. Yeah Yeah, exactly and so that was not obvious to me in 2015 of what would happen the bull or the bear You know version of that but clearly people love getting Taco Bell delivers Well, it's also I mean it now makes sense to me in a weird way because it's like, you know You save a lot of time going back and forth you can use that time to do all this other stuff You can use that time to work like, you know, and I think yeah people are familiar with those kind of calculations now and all that stuff Yeah, I think it's that I think I think there's a lot of things Jack I mean like yeah, um, one of the things that I remember even in 2013 looking at as just this Marvelous like fact that that that only goes in one direction that there's a few of them, you know one of them is that if you looked at If you looked at just food consumption, you know in the 1950s when the US government used to measure this or when they first started measuring this something like 70 to 80 cents on the dollar was spent on grocery.

18:52Okay, this is like in the 1950s if you looked at 2013 when DoorDash was founded that number was getting closer to 55 cents towards groceries, you know 45 cents towards restaurants today it's closer to 55 cents towards restaurants, 45 cents towards groceries. And so over a 75, 76 year, you know, arc, yes, there's ups and downs, but if you look at the trend line, it kind of goes in one direction, which is in the direction of, you know, uh, food prepared by somebody else. Do you know by chance if like the relative cost of a steak you made through your own groceries versus a steak prepared by somebody else, if that ratio of cost has changed, Like has one gotten more expensive relative to the other?

19:36It's a great question, but I think it depends a lot on how you value your time. So this gets me to another fact that I think is pretty interesting, which is if you looked at the percentage of dual income households, same time period, 1950 to 2020. So this, you know, this goes way beyond AI or COVID or any of this sort of stuff. you find that the percentage of household has gone from like a quarter, you know, dual income to almost like 70 percent. Right. You know, present day. If you look at the number of restaurants as another example, just just total number of restaurants, total count. OK. In the 60 to 70 years in which this is measured, there's maybe two years in which the total number of restaurants in the current year doesn't exceed the previous year.

20:19So even if new restaurants, which is a very tough business, as you know, having worked as a dishwasher at my mom's place, I definitely know how hard it is. Yes, there's a risk of going out of business, but there's always more than enough that replenishes the bucket every successive year. I think when you put together some of these facts over like 70 plus years or something, they kind of spell out that people, whether it's through dollars or time, are expressing the fact with their activity, not just their words, but their activity, that they much value if somebody else made them the stake. It's interesting because like, you know, the, the, uh, stat about, you know, dual income households, it's like that, that's true.

21:01But also like the, you know, the cost of a home has gone up crazily. And if you talk to, I think you talked to a lot of like young people today versus young people in the fifties. Sure. I think people probably feel like it's harder to get like the home and the car and everything today than they used to and all of that. Yes. So it's a little counterintuitive that people's willingness to spend on food has done what it's done. Well, I think a couple of things. So, you know, the first thing I would say is, you know, food happens 20 to 25 times a week. It's not like buying a house. Yeah, that's right.

21:31So that's the first point I make. So even if you're the most avid cook, right, that you love making, you know, food, it's very difficult. It's very, very difficult to do it 20 to 25 times. That's the first point. I think the second point, and I think you make a very good point on the affordability challenge that we have as a country and probably globally, actually, not just here in the United States. is that people always want to find a way to feel good. And I think purchases, especially on food, they don't just become like an indulgence anymore. They become like the ability to actually feel good.

22:11It feels great to hit a button and a pizza shows up. It's great. To feel good more than once and also do something that you know you need to do. Food consumption. That's right. How much do you think about like consumer trends in general outside of food? Like do you need to just be myopically focused on food or is it worth your time and headspace to care how people are spending their energy on social media or what's going on with Calci and Polymarket or other consumers? Like does that matter to you to understand? As a consumer business with hundreds of millions of customers now, absolutely you have to think way beyond food.

22:46And as a company that frankly doesn't just do food anymore. So I'm increasingly our orders are coming outside of food. We have to pay attention So what are like some of the other consumer trends that are like not about food, but that are important and interesting to you? Well, I think you named one of them, which is this affordability piece, which isn't just about food people are looking for a Affordability across every segment. Yes housing transportation eating groceries industries, healthcare, education. I mean, we can keep going. Banking. Every category, I would say affordability is a huge deal, huge premium.

23:28And so a lot of what we're thinking about is how do you continuously do two things? One, continuously bring down costs. And two, How do you bring more value? And I think those are very hard to do things. We, again, mainly try to stay focused in the world of atoms though, because one of the things that I think people on the software side perhaps don't appreciate is all this information that you can get in software kind of sometimes gives you this perception that you can structure information pretty easily and maybe control information and experiences pretty easily. end to end. That is the complete opposite in the game of Adams where everything is an edge case.

24:13Every day there's this thing called traffic and weather. And every day by definition, it's not perfectly predictable. And we can argue it's range bounded or not, but look, if there happens to be a traffic jam and something takes 20 minutes longer, that's a real problem for that one customer. A lot of what we're doing is actually just staying as expert and proficient as we can in that game. What's cool about it is by being so good at logistics and costs and coordination and all of that, it's obviously going to apply to stuff outside of food. But even just within food, it seems like to bring more value to the customer, I mean, the obvious way is you could just keep chipping away at the cost, which I assume over time, you ought to be able to get to an extremely low place, I would think.

25:00With food, but also with inventory of other types of products, right? For instance, one of the challenges in grocery or retail is, well, there's actually separate challenges. In the case of grocery, most grocers don't know what items are on shelves. And it's not because of bad technology or outdated systems or a lot of different systems. There are those challenges, but it's honestly also structurally because consumers who go inside the store move things around. Or CPG companies want certain items to be promoted or not promoted, and those things change quite often. And as a result of that, that becomes really messy.

25:41So how do you get great at that? You know why? Because if you don't get great at that and you make mistakes or you have to make a bunch of substitutions, that's extra costs. Back to your point around affordability, that's costs. even though it's not just about the price of the items, but it's about everything surrounding it to support fulfillment. In the case of retail, if you didn't know that, you know, the pair of shoes that you wanted might be a half size off, you know, because it doesn't fit great, that's extra cost. That's going to get returned somehow or refunded, you know. And those are all of the challenges that we try to obsess about.

26:15What are the like tempting adjacent things that kind of make sense for you to do that you've like said no to in the name of focus? Like, you know, as an example, as I was just listening to that, like, you know, do businesses ever want to use you to like, you know, work with their own suppliers or things like that? And then if you said, you know what, that's just too far afield from what we do, like, are there, are there close by things that you're constantly saying, that's a good idea, but it's not a great idea. We're just not going to do that. Yeah, it's a good question. I mean, a lot of the separation between good and great internally right now is around sequencing.

26:45And so, for instance, I had no idea, back to one of your earlier questions about, well, how big could food be? It's very hard as an entrepreneur, when you're working out of your apartment, to know what that size could be one day. Turns out a lot of people eat food. Turns out a lot of people eat food. It also turns out it's a lot harder than we thought. And we worked on it for seven years before we moved to category number two, which was groceries. It's funny. It's like I could see getting started like, oh, this could be a$5 billion company. And now you're like, this could be a$500 billion. Actually, in our Y Combinator application, there's a question.

27:23I don't know if they ask it anymore, which is how much revenue do you think this company can make one day? And I remember we were just operating in Palo Alto at the time. And so I counted up how many Palo Altos there could be and how many orders we could do in each one of those types of cities. and I estimate something like$100 million of revenue or something like that. Thankfully, we're a few orders of magnitude off. That's funny. But that's true. But to your point. No, but it's surprising. It's very surprising. And so I think a lot of times you kind of, as an entrepreneur, have to take the greedy algorithm.