Uber 对阵 Tesla、Robotaxi 时间表与人类驾驶的终结 | Uber CEO Dara Khosrowshahi | #243

The robotaxi race is heating up. How are you going to win that race? Well, when you joined and took the mantle of Uber, I think Uber was losing $4.5 billion a year. That was quite the buzzsaw to walk into. Today, it's earning over $10 billion a year. You really focused the organization.

Autonomous technology and autonomous vehicles are increasingly a reality on the streets as we speak. It's been 15 or 16 years of development, with lots of long tails, et cetera. Waymo is a leader and a big partner of ours in Austin and Atlanta.

What we view is that there isn't going to be this kind of binary outcome—either all autonomous or no autonomous. As autonomous technology develops, it's going to develop in a hybrid way. You're going to have fleets in cities that consist of some autonomous vehicles and many human-driven vehicles as well. Ultimately, because of the incredible demand that we bring to both human drivers and robot drivers, we will have a big part to play in that transition as well.

A hundred years ago, we had this massive shift from horse and buggy to cars. It took about 15 years to make the transition, where 50% of the vehicles were cars. How long do you expect that transition to take, where 50% of the cars are self-driving cars?

Now, that's a moon shot, ladies and gentlemen. Quite the greeting. Wow.

Yeah, the energy is awesome, and you deserve it, pal. First of all, you look amazing.

Thank you very much.

Stay in good health.

Yes, trying.

Excellent. So far, so good. I mean, the biggest challenge we have is stress levels.

Indeed.

And way too much travel. So, I don't want to take it for granted. Thank you for coming out.

Happy to do it.

Great to have you. I'm grateful for your friendship. We've known each other since you were at Expedia, and you had an incredible success story there—one of the highest-paid, if not the highest-paid, CEOs.

You actually walked away from a $200 million stock package from Expedia to take the mantle of Uber. I remember—I just looked this up the other day—when you joined and took the mantle of Uber, I think Uber was losing $4.5 billion a year. That was quite the buzzsaw to walk into. Today, this is the hallmark of an extraordinary CEO: it's earning over $10 billion a year. What a turnaround. Glory.

Fantastic. Really extraordinary, and just the scope of what it's done. I remember you really focused the organization. I think one of the things that a lot of companies fail at is doing too much at once.

Yeah, and I think we were definitely guilty of that as well. One of the great pieces of advice that I got from a good friend of mine, Barney Harford, was that strategy is just as much about what you're doing and what you choose not to do. It's very easy to say yes to another idea. It's very difficult to say no, but sometimes the noes amplify the yeses.

It's about pointing that company—having a point of view, pointing the company in a singular direction, and then putting everything you've got behind it.

Yeah. Well, you've done that, with huge success. And now you're going back to the stuff that you left on the table before.

Yeah, which is—we'll talk about flying cars and all of that.

The robotaxi race is heating up, definitely. I'm going to start with the hard question: how are you going to win that race?

We are going to win that race with the autonomous communities. Autonomous technology and autonomous vehicles are increasingly a reality on the streets as we speak. It's been 15 or 16 years of development, with lots of long tails, et cetera.

Waymo is a leader and a big partner of ours in Austin and Atlanta. What we view is that there isn't going to be this kind of binary outcome, which is either all autonomous or no autonomous. As autonomous technology develops, it's going to develop in a hybrid way. You're going to have fleets in cities that consist of some autonomous vehicles and many human-driven vehicles as well.

Ultimately, the transition—and the part that we play in that transition, because of the incredible demand that we bring to both human drivers and robot drivers—we will have a big part to play in that transition as well.

We partner with Waymo, but we've got over 20 partners now in the autonomous space, including NVIDIA, WeRide, Pony, and Wabi, amongst others. Just like we want every terrific licensed human driver on the platform, we're going to want every terrific licensed robot driver on the platform as well.

We'll be in 15 cities by the end of the year with our partners. By 2029, we think that we will facilitate more autonomous and robotaxi rides than anyone else in the world.

Dara, do you ever think Elon would come on our platform, or is he just saying, “I'm doing it all on my own”? He doesn't play well with others, usually.

Elon's approach typically has been a vertical approach. They even own their own refinery. When the day comes when those Teslas are safe with a camera-only approach, we'd love to have those Teslas on our platform as well.

We've got tens of thousands of Teslas on our platform now, and some of our drivers use FSD. We've got a lot of data. It's a great car. It's a safe car. We'd love to work with them. As to whether we will or will not, that's TBD, but there are plenty of other partners in the space, and we think there are going to be many, many winners.

When you talk about platforms—and when Peter and I wrote this book, Exponential Organizations, Uber was a poster child of an organization that scales as fast as technology scales—you're turning it into a platform. You have to have so many sockets and connectors to all of these different hybrid robotaxi, human, and other systems. How do you think about that, and what are the major obstacles in building that platform?

The platform is actually what we're finding. Humans are much more complicated than robots, right? There are many more unexpected behaviors. There are very significant differences in how you set up your platform to be operational in Bangalore and also operational in San Francisco.

With our autonomous partners, there are standard APIs that you can put out there. We are a very big partner in terms of data collection—the amount of data that we are collecting to help train the models—where you should pick up, where you should drop off, fleet-management services, and so on.

If you think about an Uber driver, they drive the car, take care of the car, own the car, repair the car, clean the car, et cetera. All of those services are going to have to be part of the autonomous ecosystem. We are essentially providing all of those services so that our partners can focus on the core, which is how you build a safe and affordable autonomous driver, while we take care of everything else.

I'm curious. I saw the announcement you had with Lucid and Nuro—I think it's Lucid and Nuro, actually—but we also had another announcement with NVIDIA as well. I guess both happened at the same time, at CES. Can you talk a bit about your tech ecosystem and your automotive ecosystem? Who are the major players that you're building relationships with?

There are a number of players. If you think about the ecosystem, there's the demand aggregator, which is Uber. Then there is the autonomous driver, and there are many players building out the autonomous driver, whether it's Waymo, WeRide, Pony.ai, NVIDIA, which is building out its own autonomous driver, Nuro, Avride, and many, many more. Wabi, for example.

There are many companies building out the driver, and we are helping them in different ways to build out the driver. Zoox is another one.

Are you going to play with Zoox as well?

We certainly would love to work with them. There are some that we're working with and some that we're in discussions with, at different places in terms of partnership discussions.

Then you actually need the car, right? The car needs to have very sophisticated compute, redundant steering, braking, et cetera. The needs of the car are different, and we're working with Lucid, for example. Lucid has a great car that already has a bunch of the redundancy and sensor systems built into it, which is one of the reasons why we partnered with Lucid.

You can expect many other partnership announcements as well—OEM announcements. Then you need essentially the fleet management and fleet ownership as well.

Early on, we will go out and buy cars. But then, over a long period of time, just like Marriott doesn't own any hotels—it's actually financial players, these kinds of real estate investment trusts, that own the hotels Marriott manages—we think that autonomy will also move into this asset-light model.

We'll be operating the fleets, repairing the cars, and cleaning the cars either ourselves or through partners. Then there will be financial players—the Blackstones of the world—that will own large fleets of cars that give a 9% yield, so to speak.

It's a big ecosystem that we're building, and we're essentially working with everyone in that ecosystem. We think it's going to be another trillion-dollar marketplace, and we're very, very excited for autonomous vehicles to become a reality.

Not to mention that the streets are going to be safer, right? These cars—or these drivers—don't get distracted. They don't get tired. We do think that regulation is coming into place that will make sure the safety case is one where autonomous driving is actually safer.

It's 10 times safer to be in a Tesla with FSD.

The early data certainly suggests that it is quite safe. Now, most of those Teslas have a driver as well as a backup, but there's no doubt that eventually the drivers are going to come out.

The human drivers are going to come out. How do you handle the liability side? When a Waymo gets in an accident, it's very clearly the fault of Waymo. When something happens on Uber's platform, who's liable there? How do you handle that? Do you have special insurance that navigates that?

As it relates to our platform, liability laws are different everywhere, but we provide essentially commercial insurance for human drivers who are on our platform. I think that with the autonomous model, the way it looks more is that Waymo, for example, is responsible for its own drivers. They are building a software driver, and they speak to the safety of those drivers. These cars are very expensive, but we are quite confident that the driver is going to be much safer than a human being. So ultimately, liability is going to come down in the industry, and it'll be savings that we can pass on to the consumer, essentially.

I'm curious: Waymo right now, the numbers I've seen come in at around $150,000 for a car, something on that order of magnitude. Then Cybercab is saying they're coming in at $30,000. How important is the asset cost going to be for a profitable business? Let me ask you a different question: How many different providers do you expect there to be on the streets of LA a decade from now? Put it that way.

I think if you think about 10 years from now, every single new car sold is going to have autonomous software associated with it. And just like we like GMs on the Uber network and we like Toyotas on the Uber network, we're essentially going to have every single autonomous model on the network as well. So I think it'll be 10+ different providers.

Now, the software space may consolidate to some extent. I don't know if there'll be 10+ software providers, but there's no reason to think that this won't be a very, very big, fragmented industry, just like the OEM, or car manufacturing, business is a very large, fragmented industry.

Wait till your original vision of Expedia overlaps with Uber, because my autonomous vehicle is going to become my hotel room, and it's going to transport me from Las Vegas to LA. All right, I'm going to sleep.

You'll be able to get a real hotel room as well. But yes, we are in some ways in the business of wiring up things that move. We've tiptoed into travel. For example, in the UK and Spain, you can take trains on Uber as well. You can take boats on Uber, etc.

For us, it's just reimagining the way that the world moves. Anything that moves, we're going to wire it up—or anything that you want to move, whether it's food, groceries, or your local drugstore. We're going to be there to wire it up to make your life just a little bit easier.

Are there certain services—like Uber Health and Uber Eats—or are there other services you've got planned?

We have our hands full right now in terms of moving people in every single way and then moving things in every single way. I'd say the focus right now is that Eats started, obviously, as online food delivery. What we're seeing now is that a friend of mine, a VC who was involved with Uber early on, said, "Never underestimate the power of human laziness." You can build huge businesses there.

We've debated ourselves as to whether Uber Eats should expand to non-food items. What we're seeing is that the expansion into grocery and into every single part of retail—whether it's Best Buy, makeup, or Sephora—has been growing much, much faster than we expected. So I'd say that's the focus for us: delivery of anything and everything to your home within 1 hour. That business is just exploding for us.

100 years ago, we had this massive shift from horse and buggy to cars. It took about 15 years to make the transition to where 50% of the vehicles were cars. How long do you expect that transition to take before 50% of the cars are self-driving—autonomous?

I do think that in the next 10 years, every single new car sold is going to have autonomous driving software on it, and it'll have a sensor stack. The cost of lidar is really coming down, and the cost of cameras is coming down. So within 10 years, every new car is going to be autonomous-ready.

Now, there's going to be a huge inventory of existing cars, and the average life of a car in the US is over 10 years. So it's going to take a very, very long time. The lifetime of cars, I guess, is similar to horses, too. Horses can live for a while, but I think it'll take a while for that fleet to turn over.

I do think that this product is a relatively expensive product today. The cost curves are going to come down, and you will see autonomous become a very, very big part of developed markets. But also keep in mind that we operate in 70+ countries, and in a lot of developing markets, autonomy is going to take a while to penetrate.

I remember a moment when I did some of the research years ago and looked at how an electric autonomous car could be 4 times cheaper than owning a car.

I think ultimately we think the promise of autonomy is exactly that: it's just not going to make sense for you to own your own car. As these cars proliferate, the cost per trip is going to come down. Safety per trip is absolutely going to come up. You'll have your own privacy. You'll be able to have your stuff essentially loaded up in the car in terms of your music or whatever you like.

We do think that autonomy is an enormous opportunity in terms of the expansion of the TAM for both mobility and delivery of all kinds.

Let me give you my vision of what it looks like 5 years from now. You're sitting at breakfast with your kids. You stand up from your breakfast table and start walking—

Breakfast delivered by Uber Eats, but, you know. Let's do that.

And you start walking toward the front door. Your AI knows your calendar. It has cameras in your home. It sees you walking toward the front door and, automatically—my favorite term—an autonomous Uber pulls in front. You didn't have to call it. You didn't have to ask it. You hop into it, and it knows that you didn't get a good night's sleep last night from your Uber data. So it decided to bring you a car with a lie-down bed in the back as well.

Sign me up, Peter.

Yeah, I'm just going to give you some product advice here. So, the international side you just brought up: Where are we going to see these roll out? Obviously, the US and China, but what countries are most pro-autonomous vehicles, do you think?

I'd say the Middle East is actually the most open and leaning forward in terms of innovation of all kinds. It's not just autonomous vehicles. You look, for example, at Joby's vertical takeoff and landing vehicles. Whether it's Abu Dhabi, Dubai, or KSA—Kingdom of Saudi Arabia—these are all countries that recognize the transformational power of technology.

They're run by governments and administrations that are very young and incredibly tech-savvy, and they want to lead. We are seeing the Middle East open up. For example, we are in Abu Dhabi as we speak with autonomous vehicles and our partner WeRide. You will see many more autonomous vehicles on Uber in the Middle East.

I'd say Europe is also opening up. Europe is leaning forward. We're pretty excited about the potential in the UK with Wave, who is a terrific partner. They have incredible AI talent, and they're essentially building a software stack that they will sell to any and all OEMs. So we're very excited about that model.

I remember when France made Uber illegal. I was like, "Okay, that's a message to entrepreneurs out there not to start your company, or to tech entrepreneurs—"

Fortunately, things in France are much better now. Yes, they are. But I do think there's real promise in the UK, and there'll be some exciting announcements coming from Wave, who's one of our great partners.

So people forget the secondary benefits, right? I remember there was this big issue where the finance minister in France was trying to shut down Uber. Around Paris, you have what are called the banlieues, the suburbs of radicalized folks and drugs, etc.—not the kind of place you usually start a company.

LA in the '80s, like that.

The French government had spent $40 billion trying to solve that problem with zero impact. It was worse than anything I've ever lived in. Then Uber comes in, and 18 months later the crime rate has dropped 10x because they're all out driving. They're trying to shut down Uber while you're saving them a $40 billion problem. It's just incredible to watch.

And, you know, the same thing happens with drunk driving. As we enter markets, drunk-driving incidents go down significantly. So there are many, many ways—it's one of the reasons, getting back to how we started the conversation, why I wanted to come to Uber. This is a company that is having real impact on society and how society moves. I think with the autonomous revolution, even more so.

Amazing. I remember I was one of the first speakers at the first Uber Elevate conference.

Back when it was a—

Yeah, yeah. When Uber was pioneering the idea of eVTOLs, which is a name that rolls off the tongue onto the floor.

[Laughter] You pull it off pretty well.

And when you came in as CEO, you ended up selling Uber Elevate to Joby and putting all of that on hold, which, honestly, I was really disappointed about. But it was the absolute right decision.

I have issues with that. Well, it's turned out all right.

No, it was absolutely the right decision because it was still a decade away, and it wasn't the right place to focus. Again, for all of you here, so many companies fail from a lack of focus. I've had companies where I served as chairman, and the CEO starts a whole bunch of different divisions. It just takes the focus off where you need to be earning your money.

So, you're now partnered with Joby. What's your vision for flying cars? I'm going to call them that instead of eVTOLs.

It is a little bit easier. We had founded Elevate because we wanted to catalyze the industry and get real investment into eVTOLs. With battery technology and battery density, for the first time, the math behind the mass of a battery and how much power the battery had made sense to actually build these eVTOLs, these electric flying cars.

Once we provided that catalyst and saw funding come into the marketplace, we thought, “Listen, we could then have specialized hardware makers fund this.” Joby was just the best that we had seen of the lot, obviously.

Joby is an amazing CEO. He's absolutely extraordinary.

The vision for us is end-to-end trip planning. A little bit like you said, what we're hoping is that in Abu Dhabi this year, toward the end of the year, you'll be able to push a button, get an Uber to the vertiport, get onto a Joby, and then take a Joby to your destination, followed by maybe another Uber at the end to wherever you're going, and save yourself hours of travel. The amount of time that you spend in traffic is extraordinary.

It cannot happen fast enough.

No, absolutely. It's happening again in the Middle East. It's happening now, and hopefully it'll come to the US sometime next year as well. The technology is absolutely real. Now it's about getting the technology licensed with the authorities, so to speak.

And getting production up and running, with enough of these vehicles, enough experience, and enough AI oversight.

Joby has a partnership with Toyota, so they have the best in the world in terms of manufacturing and safety.

Dara, one of the things I've been critical about in the past is that Uber has been extractive with the social contract, right? You're layering on existing society, but you have a plan to turn drivers into autonomous-vehicle owners. I think that's an amazing transition from shifting the focus on labor to capital. Can you talk about that and how quickly it's being rolled out? Because when you enable that, you give people so much agency. That's a really powerful direction for the future.

Absolutely. I'd say there are 2 angles here. One way of looking at Uber is that Uber is an on-demand transportation company, moving people and moving things as well. Another way of looking at Uber is that we're a platform for flexible work.

We've consistently expanded the kinds of work available, whether it's driving, delivering, or shopping. Now, we have a group called Uber AI Solutions, where we're doing data labeling, looking at models, and picking which model is the right model. We're expanding the kind of work available.

You've got drivers tagging things with the—

Exactly, tagging images and so forth. We're expanding the kinds of work available. At the same time, exactly as you said, as the world changes, to some extent you've got these capital assets, these cars, that we're going to augment and then, at some point, use to replace some of the labor, whether drivers.

The approach for us is: How do we make sure that there are other kinds of work available that lean into the technology, versus running away from the incoming technology wave? We also want to give drivers an opportunity to be their own fleet manager.

I love that. I love the idea that an Uber driver who gets displaced gets to buy a car or a couple of cars that work for him or her during the day, and can kick back and work out at the gym while it's out there.

Well, there's always some work involved in that, and you've got to maintain the fleet, right?

Yes, yes. But maybe you'll kick back and have some drivers for you.

I've thought about that.

For you, yeah. I've thought about that. Buy 100 cars, have them going around LA. What I'd want to do is put LEDs on the side and have my clawbot start experimenting with different marketing pitches as it's driving down the street—differential pricing.

I will make sure I give you a call when we open up in LA.

Besides flying cars, drone delivery: I've got Keller Rinaudo from Zipline here tomorrow. It's an amazing company, and I know you know Keller. I've had the CEO of Starship, which is the wheeled ground robot. Talk to me about how you're going to start incorporating autonomy into your delivery.

We're talking to Keller, and we work with a number of other drone companies—Sky Trax, Mana, etc. Delivery is actually a little bit harder because, as it relates to mobility, the passenger gets to the car and gets out of the car, right? Food doesn't do that. You need to make sure that the positioning of the pickup and drop-off is different.

You see the robots we had here earlier? They could jump out of your car.

That could be part of the solution. It's a first-mile, last-mile issue as well.

There are 2 modes that we're working on right now as it relates to delivery. One is drone delivery, which is for suburbs and markets that don't have high-rises. We're very excited about that because suburban markets are a new growth opportunity for us. I do think drone delivery is going to be one of those surprise-and-delight moments where you get your food in 10 minutes. Usually, prep time is 10 minutes, maybe it takes 2 more minutes, and you get it within 15 minutes, which will be absolutely spectacular.

In the urban markets, we've got these cute little sidewalk robots, whether it's Coco or a number of other sidewalk robots.

In Santa Monica, where I live, they're all over the place.

For those, they're about right for short deliveries within a mile or 2 because they don't go very fast and they use the sidewalk. Ultimately, you're going to have this multimodal delivery network that we're putting together, whether it's on the sidewalks or in the sky. Eventually, you'll have specialized delivery robots that go on bike lanes or on the road as well.

We're going to go to your questions next in a couple of minutes, so get them ready. Dara, there are roughly 1 million active drivers in the US.

More than that, but yeah.

And you've got about 10 million globally. As you see the rollout of autonomous vehicles, what's the curve in which you expect the number of drivers to drop?

Looking forward to 2030, we're going to have significantly more drivers on the platform, including in the US, than we have today, just from the growth of the business. The business is growing over 20%, and we need to add that many drivers, essentially, to keep up with the growth. Our audience is growing at almost 20% as well.

What we see in terms of autonomous is probably around 20% of drivers every year sloughing off. They find something else to do, or they find full-time work. As we introduce autonomous into a market, we just slow down the recruitment of new drivers so that the drivers who are on the platform have just as much work as they had before.

An interesting trend that we're seeing is that autonomy looks like it's bringing new customers into the marketplace. One of the hypotheses is that autonomy is not going to replace just human driving; it's going to expand the market. What we're seeing in Austin and Atlanta is that these markets are growing faster than the average market nationwide, which suggests that there's actually going to be more business to go around, especially if the price starts coming down.

Yes. It's like buying time. I've got 2 amazing cars, but I want to spend that 30 minutes working on a project, not behind a wheel.

That's what we do: essentially give your time back. That is the value that Uber represents, whether it's being able to work in your car or not having to go out to the restaurant.

Dara, you are, in my mind, one of the most extraordinary tech CEOs out there. I want to drill down for a second for all of our CEOs in the room. What are your philosophies about running the company?

You're not shy about telling your employees they've got to work their butts off. Can you talk a little bit about that? We came off 2020 and the whole COVID era, where people were out of the office and such. What's your philosophy for driving Uber's success as a CEO—by inspiring or directing your employees?

I didn't come to Uber because I thought it was going to be a layup.

Right? At the time, Uber was going through a very, very difficult time. I think anyone who joined Uber around my time—and there are a bunch of our teammates who joined around when I did—

2017.

Yeah, you know, there’s a certain personality at Uber who likes running to the fire, who doesn’t run away from the fire. Ultimately, the biggest reason why I joined Uber was one of the lessons I learned in life: go to a place where you can make an impact, and go to a place that is having an impact in the world.

If you want to have an impact at a company or if you want to have an impact in the world, it doesn’t come with a free lunch. You’re going to have to work your ass off.

And you’re not scared to say that to your team.

It’s not that I’m scared. I say it to the team. I think it’s a feature of the place. It’s an intense work culture. You work with super-smart people who are working really, really hard.

Yes, that comes with a trade-off. It doesn’t mean we don’t have flexibility around work, but it’s fun to work hard and succeed. It’s fun to succeed at things that are difficult and really, really hard. I think it gives us another level of satisfaction. That’s what Uber offers, and we’re going to be very plainspoken about it.

You recently introduced a Dara avatar for employees. Can I see a video of you?

Our employees have a Dara avatar. They get to practice their pitch in front of it and get feedback.

How’s that working out?

It drives me nuts. I know, it’s all these prep meetings and so on. It’s such a waste of time. I’m like, just come in here. Let’s talk. We’re both human beings. Just because I’ve got a fancy title doesn’t mean that you’ve got to prep endlessly.

But people tell me the avatar is amusing and it helps them prep, so good for that.

You should do a pitch in front of it and see how it goes.

I actually prepped for this meeting, you know.