If robots are twice as good a driver or three times as good as drivers as humans, that's good for society going forward.
But I honestly don't know if society is ready to accept that.
So it might need to be a hundred times better or something.
It might.
Welcome to The Logan Bartlett Show.
On this episode.
What you're going to hear is a conversation I had with CEO of Uber, Dara Khosrowshahi.
If this is your first time tuning into The Logan Bartlett Show, I have discussions every week with different founders and executives, as well as investors in technology and beyond, about their scaling of businesses or investing in different companies and what it's like growing into the leaders that they are today.
In this discussion with Dara, we talk about a number of different things, including coming off of their earnings, what he's seeing from a macroeconomic environment, as well as any potential data related to recession that might exist and show up in Uber's business today.
We also talk about the state of autonomous vehicles, as well as Dara's decision to shut down the Advanced Technologies Group and the decision to partner with Waymo.
I think when it gets to those big strategic decisions, those are decisions that are lonely decisions.
You have to be willing to make the un-average decision, so to speak.
During those times you spend a lot of time alone, you talk to a lot of people, but ultimately you've got to go forward with the fact patterns that you see, and then your own instinct.
We discuss where long-term value might accrue in the stack between Uber and autonomous driving fleets.
We also talk about Tesla's initiatives in this market.
Finally, we discuss lessons that he learned from his longtime mentor, Barry Diller from IAC.
A really fun conversation that you'll hear with Dara now.
Dara, thanks for doing this.
Happy to do it.
Thanks for having me.
Welcome to my humble apartment here.
It's pretty sweet.
Yeah, it's cool.
Not so humble.
Yeah, not so humble.
This is a new apartment.
We're supposed to grow into this apartment over time.
But no, I appreciate you doing this.
So earnings a little bit ago, and congratulations.
Seems like it went well.
Thank you.
I guess there's a lot of recession fear permeating the markets right now and you guys are kind of a bellwether for economic activity, at least in some ways.
Was there anything that that came out of the earnings or what you're seeing in the data that maybe foots with concerns or optimism?
I think we're not seeing any signs in terms of the weakening of the consumer.
And you know we're pretty big service that about 40 billion in gross bookings and the growth rate of bookings has been about 21 for the last this quarter, last quarter as well.
Profitable.
Profitability continues to improve and we've been looking honestly, based on everything that we read about looking for signs of weakness, and we just haven't found it yet.
Now I personally have a hypothesis which is coming out of COVID spend on services continues to grow faster than spend on retail.
So, if you think about personal consumer expenditures on services versus retail service spend is still not where it was pre-COVID.
So I do think that there is some kind of tailwind as it relates to people going out, people traveling places, still going to concerts, et cetera that we're benefiting from.
And then I do think also the Uber customer tends to be a bit of higher earning consumer, is urban younger, etc.
And I think that's definitely benefiting us.
And our driver supply now is we've got 74 million people earning on the platform, up over 20 on a year-on-year basis.
So, in a market where the job market gets weaker, our supply improves.
As our supply improves, ETAs come down, surge comes down, service level actually improves.
So there's this counter-cyclical element in our business which is helping us.
So I think, put it all together teams executing well, And I think for us, we're definitely continuing to benefit from the spend on services that should continue for some period of time.
Yeah.
Well, it's great to see.
And hopefully that's a indication for the broader economy, although maybe just a discrete data point, of a well-run business.
Yeah, and listen.
We also look like what about the higher income consumer versus lower income consumer?
And both for a mobility business and our delivery business.
The lower income consumer is actually increasing spend faster And when I listen to some of our partners, when you listen to the news, it seems like the bigger worry is a lower end consumer.
We're not seeing it as of yet, and hopefully we won't see it for a while.
The other topic du jour, I guess, as it relates to your business is autonomous vehicles.
And it's an interesting sort of philosophical debate of where AI is going to go in society.
But this has been one we've been talking about for a while.
I guess one question at a more philosophical level.
What percentage of or how much more safe do you think autonomous vehicles need to be for people to accept them in society versus a human driver uh thank you for asking the most difficult question too it's hard it's it's a human name I mean how many people die in the U.S by car accidents so so that that's actually where I was going so about 40 000 over 40 000 people die by car accident per year and and mostly healthy Yeah, I have a measure whether they're healthy or not, but yeah, you would assume that they are.
I looked at a lot of data during COVID and you were sort of looking at the driver fall off or people in car accidents versus others in the population.
And I think it was more younger, healthier individuals that didn't have comorbidity or something, right?
Yeah, and they, probably those are the people who get out of the house more often right, totally travel more, Etc.
So so, if you do the math like that's call it 1100, 1200 people uh dying uh sorry, probably on a daily basis.
100 to 120 people dying every day, right?
As a result of accidents.
So if you get into a situation where you have robot drivers that are 10 times better than human drivers, Right?
On a good day, robot drivers will be responsible for five fatalities, versus a bad day, when they'd be responsible for 20 fatalities.
And that's unacceptable, I think, to society, right?
And those are robot drivers that are 10 times better than human drivers.
Uh, and i just think that society accepts human fallibility quite readily, but definitely does not accept call it um robot uh, or companies making mistakes, etc.
There's there's a much, much higher bar.
So, you know, logic would dictate that if robots are twice as good a driver or three times as good as drivers as humans, that's good for society going forward.
But I honestly don't know if society is ready to accept that.
Yeah.
So it might need to be a hundred times better or something.
It might, but I think that's up to regulators.
And I think that kind of ground is yet to be uh developed and you know ultimately we are going to do whatever regulators think is the right thing but this safety question is logic would dictate society letting it happen sooner but i think that humans are emotional animals and aren't always logical Yeah, I think any time there's a major technological shift or transportation-related shift, this happened with trains way back in the day.
And it happened with airplanes.
And so it's some form of just social acceptance.
And it's terrible to say.
We're talking about death.
But it's human nature to feel uncomfortable about all this stuff.
100%.
Yeah.
Now is there anything interesting on a usage side or from a behavior standpoint that you've seen in the early data around people using autonomous vehicles?
I think that what we're seeing is, first of all, never underestimate the power of human laziness, which is we have a dynamic dispatch layer that will dispatch a particular ride to a human or a robot, depending on pickup, drop off points, et cetera.
And even in a situation where you have to walk two blocks to go get your car, a surprising number of consumers are like oh no, thank you.
I'll take a human, et cetera.
So people, one of the reasons why people come to Uber is they value convenience and they value their time at significant levels, which is why they're willing to pay for the service.
So I do think the pickiness of humans and wanting to get things just right is incredibly important.
Generally we're seeing that when someone is offered one of our customers is offered an autonomous ride, about half the time they say yeah, that'll be really cool.
Half the time they say, no, thank you.
I'd rather have a human.
I think that's gonna improve over a period of time.
But I think the experience is a delightful experience.
Robot drivers aren't quite as good at human drivers in terms of ratings right now.
It may be because people like to rate people at higher levels, or they're gonna be less uh, they're gonna be tougher on robots than humans, etc.
But i think people are trying to get from point a to b and for us it's about driving uh, perfection in every case uh, and you know, giving people the convenience of their time back.
The consideration of rolling out to new geos how do you sort of think about like, going from phoenix scottsdale, to a broader market, either within arizona or coming to new york, hopefully soon?
Well, I think for us, the good news is we're in discussions with multiple AV partners about bringing their content onto our platform, as long as that content is safe, et cetera.
What we bring is, we're already operating in all these markets.
So we understand what the pickup points, the drop-off points are when you should operate, et cetera.
We've built out all that infrastructure.
Then an AV player can come and essentially instantly monetize what demand already spoken for.
We're absolutely going to look to expand as quickly as possible.
You've got to do it the right way.
You've got to do it with regulators coming along for the ride, so to speak.
And I think the promise is very, very significant going forward.
There's this adage of people underestimate what can be achieved in a year, or overestimate what could be achieved in a year and underestimate what could be achieved in 10.
You're seven years in now.
I think when you took the job, there was a lot of the advanced technologies group was going and Uber was trying to develop these capabilities internally.
I think people probably projected there's a lot of commentary that I've seen to this effect there'd be some broader impact of autonomous driving at this point in 2024, seven years later.
As you look out, seven years from now or whatever point in the future you can look towards, is there any projections or considerations that you sort of have around autonomous usage.
Yeah, I hesitate to project when every person's projection has been incorrect here.
Yeah, wrong.
But I do think that the I think the adage is correct, which is if you think long-term autonomous is going to have a very, very significant positive impact on society, on safety and, we think, for our business.
It's gonna make transportation and mobility available for a lot more people in a safer way, should bring prices down pretty significantly.
Now I do think that these technologies and how quickly they get into prime time depend on hardware versus software cycles, right?
Software obviously can scale much more quickly.
Virtual goods usually are much more consistent and uniform versus our business, which is a hyper hyper, local business.
Every single market is different.
Often we're regulated not on a country basis but on a state basis, city basis, even airport basis, et cetera.
So it's much more idiosyncratic in nature, which makes the scaling harder to come by, and then this ultimately, is going to be quite hardware dependent for the business to be economically viable or um, as promising as we expect it to be.
So kind of one construct that i look at is evs.
Right evs, it was a technology actually.
I think tesla, for example, introduced their first model in 20.
2008, right fast forward to 2023, 15 years later and Evie's account for about less than about 20 of car sold, so that was a technology that was Kind of ready to go.
Technically, it was ready to go in 2008.
It took 15 years to get to 20% of the marketplace.
I'm hoping AV doesn't take as long, but I do think, because of hardware cycles and the idiosyncratic nature of the regulatory pie, it's not going to be seven years.
It's going to be more than that.
10 to 15 years from now, I think it's going to be a real part of our business.
Yeah, there's some value props outside of the societal benefit of lower death and stuff like that.
But I just remember Minority Report, if you ever saw that movie.
And there was a lane in the highway of people going 150 miles an hour because everything was, you know, programmed in
And so the cars didn't need to react to one another and all that.
So there are some of these like benefits that the individual user can feel even outside of the supply dynamics and the access and all of that stuff.
And so I'm hopeful that people will see the opportunity sooner rather than later.
Now, will governments do the same?
It's hard to say.
Governments are going to work at their pace.
And the safety issue is a big one.
The societal safety issue is something you can't take shortcuts to, so to speak.
Yeah.
Logan here interrupting this conversation.
This is not an ad.
As everyone, I think knows, we don't run ads on this platform, but we'd like to do a quick call to action for everyone to please share with anyone that you think might find this conversation potentially interesting.
We are always trying to continue to grow the podcast so that we have more people listening and we're able to bring more great guests onto the show.
We'll get back to the conversation now with Dara.
So when you took the job, how long after, until you sold Advanced Technology Groups to Aurora?
So it was after COVID hit us.
So my guess is it was three and a half, four years in.
Originally obviously, ATG was a part of the business that we were developing, but then, once COVID hit, We had to make some hard decisions as far as what businesses we stayed in and what businesses we got out of.
And we got out of ATG which, in hindsight, was absolutely the right decision, but it was a really tough decision.
Yeah.
I'm curious.
I mean, I want to talk about it being absolutely the right decision, but in making that decision it wasn't in a vacuum.
Obviously, COVID was happening and there was a couple other decisions that you guys made strategically to sort of streamline the business.
How do you go about making a decision like that where you spent, I mean, how long was that?
It was a six and a half year initiative, maybe something like that.
Yeah, I mean it was a very, very important initiative that the company undertook, you know, to some extent at the time, properly because there was a hypothesis that there was going to be a winner take all on this marketplace.
Didn't turn out to be, and I don't think there will be, but it was a big part of the company's efforts.
Yeah, and so how do you go about making that decision?
Something that's like so- So I don't think there's like a general construct to make those kinds of decisions because hopefully, you don't make many of those decisions in your career.
Like, if you're making those big decisions every single week or every single month, you're going to be in trouble.
Like, you know, you're not actually running a business.
But I do think for me, as it relates to those kinds of large strategic decisions, One construct that I really think about is that there are certain times when it makes sense for you to bring in your whole team and kind of manage by consensus, make sure you get everyone to agree, et cetera.
But I think when it gets to those big strategic decisions, those are decisions that are lonely decisions.
You have to be willing to make the un-average decision, so to speak, because usually by consensus you get to the average decision.
If you're going to strategically shift, you have to be willing to make decisions that are outside of, let's say, the average of everyone's thought process.
So I do think you know, during those times you spend a lot of time alone, you start, you talk to a lot of people, but ultimately you've got to go forward with the fact patterns that you see, and then your own instinct.
For us.
What really shaped my decision was twofold, which is, You know, this was a kind of a side project for us.
The main line business was building a network and algorithms, pricing, matching algorithms, etc.
It was, I think, Tobias Shopify called it, you know, a side quest, so to speak.
So, running a side quest for a big company is hard enough.
But running a side quest in a type of business which is hardware, which is so different from software for us, makes it doubly hard.
Like running a side quest.
That's very similar to your base business.
Maybe you can succeed in, but We at a core are a software company.
We run hard, we go fast, we make mistakes all the time, test and learn, et cetera.
The pace at which a company goes is something that I love.
Hardware, the cost of making mistakes are huge.
So you almost have to like you have two completely different businesses coexisting within one culture, one way of doing things.
And it became very, very difficult to be successful at both.
There are very few companies, like in Apple, who is both good at hardware and software.
But even there, their pace of software development, you can argue, is slower than, let's say, a pace of a Google or a Meta or some of the other players.
So ultimately, the SideQuest nature and the different pacing of those businesses led me to say hey, it's time to cut loose.
We're not going to do ATG.
We're going to partner in this area uh, with companies who all they're doing is hardware, so to speak, and all they care about is av.
That's their passion.
We want to work with them rather than trying to compete with them.
What um, i assume covid forced uh prioritization and considerations like this did you?
Was this in the back of your mind before COVID?
That hey, this might not be the long-term strategy for Uber?
Yeah, it was an open question for a long time.
The cultures of the two businesses were very, very different, right.
The mainline business at the time was fighting lots of battles out there.
We were competing with Didi.
We were competing.
You know there's a lot of capital in the marketplace.
Lyft was a very, very big and fundamental threat for us.
And at the same time, we were working on the AV on this other side.
And those two teams didn't work together that much, partially because we also wanted to partner with other AV players.
So we almost had to treat ATG to some extent as a third party within the same company.
And that was really tough to pull off.
Yeah, got it.
And so you said it in hindsight is proven to be the right decision, I guess.
Can you walk through?
If there was a skeptic thinking about how this market could play out and we'll get to a uh, we can talk about tesla in a second.
But if, if someone were a skeptic and saying hey, at its end, state autonomous vehicles will accrue a lot of the value and the network that sits in between might just be an intermediary that should be disintermediated at some point.
What would you say to to the skeptic around that?
I guess to some extent i.
I would Tell the skeptic that they're right, which is why I think we've made the right decision Which, if you think about our mobility business, you know, on average, our take rate is about 20.
80% of the value goes to the driver, pays for insurance costs, et cetera.
So that 80%, our mobility business did about $20 billion in bookings this last quarter.
So that's $80 billion on an annual basis.
80 of that uh 65 billion 64 billion dollars is available to our av partners and or drivers.
So this you know we want to run this marketplace at a low take rate.
Um, if your take rate is 20, you have to be able to drive about 20.
25% higher utilization of an asset.
And, based on the data that we've seen, we're able to drive well in excess of that 25 utilization premium, so to speak.
So ultimately, that's why it worked.
If the majority wasn't going to the AV provider or the driver, we might have undertaken a different model.
But at this point the AV player can take the majority of the economics.
It will expand the marketplace.
And for the 20 take rate.
We built a global marketplace pricing matching routing technology, unlike anyone else.
And an AV player can just plug in instantly and get instant demand from all of our marketplace, which we think works for both players.
It kind of dovetails into some of the stuff with regard to Elon Musk and Tesla.
And I think after Q1 of this year, they said they were going to release a preview in August, as we see here in August, of a summon which I think they showed a screenshot at some point.
What complexities do you think are not fully appreciated by someone that's very bright but making a car company and all the network considerations you've had to factor in?
Yeah listen, to some extent it's a repeat of people say history may not repeat, but it rhymes like Dimer BMW.
They tried to set up their own networks as well, but it's a really, really different business.
It's what I was talking about hardware to build a 20000 or a 50000 dollar piece of hardware from driving over 30 million transactions every day that, on a revenue basis, you make 2 off of.
It's just a very, very different business.
Everything that you have to build in terms of the matching stack, the pricing stack, all the things that can go wrong.
In a lot of markets, people want to pay with cash.
The accidents that happen in a car, people getting sick, people losing items.
We talked about this just during our earnings.
We actually find 25 million lost items in a car.
All of these little details.
I'm responsible for a disproportionate amount of those, by the way.
A lot of those were me, 25.
We try to help you out.
You've done a great job, actually.
A taxicab, I had that happen recently in New York.
Forgive me, but I called a taxicab because of- You can call a taxicab through Uber.
I hailed it, unfortunately.
If it had been through Uber, I would have been able to find my item.
Instead, I was going through white pages trying to get the medallion.
And finally, actually, I got a hold of them after two days of panic.
You're a determined fellow.
It was a jacket that I had just purchased, and so the consideration set was high for me to get it back.
But it is much nicer to be able to hail the Uber and pay the fee that goes to the driver.
100%.
So I think, all of those considerations we've had to learn to build out a system that's able to make everything work for both the rider and the driver, with economics at work.
It's taken us 15 years.
It's taken us tens of billions of dollars of capital.
And we can provide that instantly to a partner.
And, you know, hopefully, Tesla will be one of those partners.
You never know.
It's in there's there's all the different regulatory considerations as well at a service level.
Right.
In addition to figuring out all of the human behavior things.
But then there's all the government consideration.
Different pickups, drop offs, airport pickups, drop offs.
And listen.
It's also not clear to me that the average person you know, Tesla owner or owner of any other car is going to want to have that car be ridden in by a complete stranger.
Which, by the way, for people that don't know, that's the current plan.
And I think he said hey, in markets, that there aren't enough supply, they will also put cars on the road.
But the plan would be, hey, Logan owns a Tesla.
It sits out there idle 23 hours a day.
And so do you want to use it for a third of the eight hours a day just to go around and pick people up?
Exactly.
And it just so happens to that probably the times at which you're going to want your Tesla are probably going to be the same times that ridership is going to be at a peak right.
There are these peaks and valleys in terms of supply and demand.
And we are able to essentially balance supply and demand or position supply on a temporal basis when you need the supply, or in certain places if there's a concert, etc.
But because essentially, drivers who are part-time are part of this ecosystem, we only have to pay for that supply when we use it.
If you have a fleet that essentially you've gotta use 100 of the time or you've gotta drive utilization 100 of the time, you're not gonna be able to work the peaks and valleys the way that a hybrid network that's gonna have both humans and robots on it are going to be able to do.
So I think there's a lot that we bring to bear And, by the way, I don't think it's a zero sum world right.
There is a world in which 1P and 3P work together.
And if you look at food, for example, you got McDonald's Starbucks, Domino's.
Every single major player out there has a direct channel to consumers.
But as they try to maximize the utilization of their restaurants, et cetera, they have come to the conclusion that they should work with marketplaces.
I think the same is going to be true of cars.
If you want to drive utilization of that asset, you're going to want to engage definitely in 3P.
And if you want to develop your own 1P channel as well, that can be true.
It doesn't have to be either or.
You've successfully gotten Uber into a number of, I guess, adjacent revenue streams, advertising.
Was advertising in place when you?
No.
No, it wasn't.
Got it.
That was actually for me, it was pattern recognition, which is At Expedia.
We'd actually built a big advertising business.
Hotels could bid for placement to move themselves up in the stack.
A lot of tourism companies advertise as well.
And we saw the same thing eats, which is now restaurants can essentially bid for placement.
They can provide sales what we call merchant funded offerings that are increasing significantly year on year, to drive their business in kind of a variable way.
When they need demand, they can advertise or make offers.
When they don't need the demand, they don't have to.
And that's a billion dollar revenue right now?
It's a billion dollar revenue business.
It's a little more than 1% of gross bookings.
We think that it can be 2 plus percent.
We're also building an advertising business with our grocery retail business as well.
And then actually on rides too.
We're being very, very particular in terms of the brands that we work with, etc.
But we will put some brands in front of you when you're waiting for your driver, etc.
So it's a billion dollar business.
It should be multiples of that in the coming years.
And I assume pretty good profit margins advertising.
Advertising is a wonderful, wonderful margin business.
It's an amazing business.
It's magical compared to our base business.
Yes, yes, yes.
So we might talk 1% to 2% on a bookings basis.
But on a flow through basis, it's probably.
Definitely.
And I think one of the tricks for commerce companies because advertising with the high margins, et cetera, can be quite intoxicating, if you want to use the term is you have to make sure that it doesn't hurt the base product and the base experience.
So we constantly have holdout groups.
We could be driving our advertising business and growing it much faster than we are, but we don't wanna succumb to that kind of temptation.
Once you do it, it's very difficult to pull back.
So we're going in a kind of determined, measured way to make sure the customer experience doesn't get hurt.
Yes.
You don't want to be the fourth search ad on the results when you're when you're going down the pages.
It's hard on mobile, and data labeling is another business that you guys have.
Have very small business that that we're working on now.
And for us, our hypothesis there is If you think about Uber to some extent, you could argue that we're a labor platform, a platform for flexible work, right.
There are 7.4 million earners earning flexibly on our platform.
They earn almost $18 billion just this last quarter. growing over 20%.
And you could argue that the first kind of flexible work that we've introduced has been transportation.
So are there other kinds of flexible work that we can introduce?
We first started doing it internally map labeling, data labeling, et cetera for machine learning algorithms.
We're now opening it up to third parties.
And in many cases it's their drivers who they may drive during the day, and at night they go home and then they label and they do some of this work.
The economics have turned out really good.
The quality has turned out to be excellent.
And it's more ways for our earners to make a living, which is very, very cool.
So very promising, but small part of our business right now.
Yeah, well, how does something like that come to be?
Does that bubble up at a grassroots level?
It sounds like advertising was maybe top-down consideration.
Very, very much grassroots.
And we started doing it internally.
There's a team that we have in India who does a bunch of this work internally.
And we often compare them to external benchmarks.
They just, this team is extraordinary.
They keep just beating the crap out of external benchmarks.
So we said well, if you beat external benchmarks, well then why don't we compete against these external benchmarks?
So we built a very, very small sales staff and they're doing pretty well and they've earned themselves some engineering bandwidth and some sales bandwidth.
But teams gotta perform as they deserve more.
Listen, if scale AI is any indication, I think there's a good big market to go after in some of these JCCs.
Yeah, I mean, it's a great company.
It's something to aspire to.
It's really impressive what they've been able to do.
So we touched on Uber Eats and I guess maybe, as you think about, Uber Eats and DoorDash is obviously a big competitor and other player in the market.
For people that don't know the evolution of Uber Eats and how it came to be it was an adjacency to the existing Uber business.
And so it sort of grew up in cities and a much more urban business.
Can you speak to, I guess, a little bit of the history and maybe contrast that with DoorDash and then where we are today in the eats market?
Yeah, definitely.
I mean this was before my time, but I actually think the first iteration of Uber Eats was drivers kind of loading up burritos in their trunks and driving around the kind of instant push a button get your food in five minutes.
And it did start in the cities.
So our strength has always been urban centers and we're growing outside of those urban centers into the suburbs.
DoorDash to some extent is opposite, which is they have.
They made a bet on the suburbs, and it's an effective bet.
And, as it turns out, in the US actually the suburban market for food is larger than the urban market for food.
So in the big cities we have a big advantage, which is we have a lot of liquidity in terms of drivers.
One of the advantages that we have is our mobility.
Business is a huge customer acquisition channel for our delivery business.
We have one membership program, Uber One, where you get benefits on both free delivery and then you get cash back on mobility as well.
So cities have been a real strength for us.
We are expanding into the suburbs.
And I think DoorDash is trying to defend their turf.
You know, they're a good, strong company.
And then when you look from a geographic basis, we are much stronger internationally.
I think one of the really good bets that Travis and team made early on was expanding the company very, very quickly internationally, when one could argue that the economics weren't proven out.
Turned out that the economics, you know, eventually did prove out.
So I think the international footprint that we have is an asset that is a very, very substantial asset that I think, puts us in a better position than our competitors, whether it's DoorDash or any other local player.
If you could wave a magic wand, and I realize this is a very US-centric question, but is...
Is what you're working on in Uber Eats to continue to up-level how much of it's brand versus supply versus I don't know, any other products related considerations.
I think that certainly selection is an area that we're working on our selection.
We've added, we've got about 1.1 million merchants on the platform.
It's growing 13%.
We're not even close to be fully penetrated there.
And I think our selection outside of urban centers can improve and needs to improve very, very quickly.
And then for us as well, a lot of the algos that we have built
