Lyft’s cofounder has his own vision for a self-driving taxi fleet

Lyft’s future is driverless, and mustacheless.
Lyft’s future is driverless, and mustacheless.
Image: Josh Edelson/AP Images
We may earn a commission from links on this page.

Laguna Beach, California

Just about every car company, Silicon Valley heavyweight, and startup is trying to build a self-driving car of some sort. Earlier this year, General Motors announced that it had acquired Cruise Automation, a startup aiming to build after-market self-driving “kits” for cars. It also invested roughly $500 million in Lyft, Uber’s largest ride-hailing competitor, in January. Last month, Lyft’s president and cofounder, John Zimmer, released a treatise on his belief that Lyft can bring self-driving cars to its service in five years.

On stage at The Wall Street Journal’s WSJDLive conference Tuesday (Oct. 25), Zimmer explained how he sees Lyft as a “network” that can be applied to any group of cars—self-driving or otherwise—and in the future, envisions a subscription model for autonomous Lyft cars that would resemble how we pay for streaming services like Spotify today.

Quartz sat down with Zimmer after his talk to discuss in more detail how he sees Lyft bringing autonomous cars to US roads, how that plan differs from competitors like Uber, and what will happen to human drivers when the robots arrive.

This interview has been lightly edited for length and clarity.

Quartz: You’re very bullish that in the next five years, the majority of Lyft rides will be self-driven. Logistically, how do you actually see that playing out?

Zimmer: People think an autonomous car is going to drop out of the sky and do everything—that’s not true. To solve for things like an airport ride, where there’s pick-up and drop-off, to solve for things like really bad weather, tough intersections, speeds over 25 mph, will take time. To provide that really perfect user experience, I think the analogy is very helpful of going from 3G to 4G [cellular technology]—if you don’t have 3G to fall back on, 4G doesn’t work, and 4G doesn’t cover everyone everywhere from day one.

In terms of how I think it’s going to happen: You’ll select a certain geofence within a city, and block out certain intersections and other things, and that’ll be complemented with the large driver network we have. And when you request a ride, the app will do the thinking—“Do you want an autonomous vehicle? Yes? Okay, great, because we can do it for that trip”—but we won’t surface you that option unless we can do it for the route you’ve told us you’re going on.

I think when we talk about an autonomous trip, no one’s really done that yet. Even like the experiments, like Tesla or Uber are doing, those aren’t autonomous—there are drivers there. So it will be fully autonomous, but we’ll do things along the way to build to that. What I’m talking about is where there’s no driver.

I think the best way to articulate it is, if a car company was to sell a car with no steering wheel to right now a consumer, it would take so long. If you sell a car without a steering wheel it has to be perfect, it has to cover everything. Whereas on a network, you can sell us a car with no steering wheel, and we can fix it to only do what it’s capable of doing to the highest safety degree.

Wouldn’t regulators still have issues with those being on US roads?

I don’t believe so. I think they’ll have different grades of service, different levels: Once you’ve shown for this many miles you can do sub-25 mph without crazy intersections, you get this level of approval. I’m making this up. This is just a way that they could do it. And then, okay, now we can take on these types of intersections or we can go between 25 and 35 mph, and we get this level of certification. I imagine it happening in that way. Part of it is, you’re going to put these things on the road when they’re safer than what we have today. You wouldn’t want to hold back that safer, more affordable transportation because it can’t do everything—you would just limit it to not do those things.

What do you think about other ride-hailing companies doing this sort of thing? Uber has cars on the road that may or may not be safe—we’ve reported they’ve already been in scrapes.

I think companies should be careful not to do publicity stunts with cars. That said, there will be stages of rolling this out. We won’t do it until we’re confident in the safety of what we’re doing, and I think by working with a car company in General Motors that’s been doing this for 100 years, we have that advantage and that perspective.

In your talk, you made a distinction in how your autonomous future works with GM, in that Lyft essentially is the network, and GM would provide the cars with Cruise. How does that work?

It’s not perfect, but I think of it like this: We are the AT&T, and with our technology, with our algorithms, with our scale of both driver and passenger, we can provide that 3-minute ETA (which is the equivalent of their three bars of coverage), but to try to do all three components in what could a multibillion-dollar business we think is, one, not necessary, and two, not focused. And it gives us the opportunity to partner in a way that our competitor can’t. GM is the first of the ways that we could find partners to do those elements, but the way you characterized it is correct.

Where will the capital for your self-driving cars actually come from?

It’ll be financed. People often ask, ‘Who’s going to own the vehicles?’ To me, the question is, well, it could be us, it could be the car companies, or it could be an intermediary bank. It just really matters like, what is the financing cost? The advantage of the network is that we can monetize these assets instantly. So you give me a $100,000 autonomous car, I can pay that back in a short period of time, because we have this existing demand. And it will be easy to see the existing demand, which has been there for many years, and create a financing vehicle for that, just as they do on real estate, or cars as they exist today. The most likely outcome, if we don’t want to have this large capital outlay on our balance sheet, is that there’s a finance percentage that we pay for this fleet that is again either leased by us from the bank or from the car manufacturers. We could finance it through debt, because it’ll be quite obvious how we can monetize it through the existing business.

I think aviation is a super interesting parallel, as you think about financing of fleets, liability insurance—all of the questions people have about our industry, there are close equivalents.

Has Lyft thought about how it would insure self-driving cars? 

We have this wonderful woman on our team, Kate Samson, who runs our insurance program and had been in the insurance industry for a long time. She created the first product for Airbnb, and for us. So we’ve already done some thinking, but we’re not doing those trips now, so I don’t have the specifics, but there’s different ways of estimating the risk based on all the data we have, taking on portions of the risk, as well as finding partners to do that as well.

You’ve said that you think manual cars will be “irrelevant” by 2025. How do you plan to upend over a century of automotive history in the US so quickly?

Well, I’d say, look four years back when we said we were going to get people to share rides in their personal vehicles. What did people say to us? They said, ‘You’re out of your mind. Like, do you know how tight control the taxi industry and the government has on this?’ And then we felt like, on the flip side, if you like, landed as an alien in our culture, in the country of freedom, and you can’t give your neighbor a ride and get compensated for it, that’s quite absurd. With that simple analysis of like landing as an alien, you can say that we are on the right side of history, and we need to negotiate a path forward and make sure we’re protecting consumer safety. And I’d say because we’ve been through that experience, I’d argue that maybe it’s a similar level of a hurdle to get over. I believe that hurdle was higher before than what this hurdle will be, namely because there was a large change of behavior and an existing regulatory structure that limited competition. That future obstacle, we’re actually aligned with car companies like GM in doing it the right way.

Can government regulations keep up with the technology? 

I think the fact that the US put out guidelines before these things are ready and out there, the answer should be yes. The onus is on us to prove that this is safer than what the current options are. And then when you have a safer, more affordable, more convenient solution, I believe that people that are voting will ensure that they are allowed. Which is what we had to do the first time around, in that you are dealing with a change of behavior, societal norms. For example, if I get into a yellow or black vehicle, it is okay not knowing that those vehicles in California didn’t do criminal background checks or that you can have a DUI in San Francisco and drive a yellow taxi. But there’s that societal norm that to me is harder to overcome than the regulatory battle that’s ahead.

That’s fair, but on the other hand, people are not always rational. If you look at how the media and the general public are reacting to the incidents with Tesla’s Autopilot, and the way that Elon Musk has responded, saying how much safer his cars are than human drivers. I wonder if you present all this data saying how much safer your cars are, if people are still going to be okay with a computer killing someone?

It will take time—we’re oversimplifying in a short conversation—definitely as a society, it’s something that we’re going to have to wrestle with. And I think that I want my children to be in the safest possible environment. It’s weird that we’re going to have computers making these decisions, but I think we can virtually eliminate fatalities, which is the goal, but I am 100% sure that we can reduce the number of fatalities. And I agree with others that it would be irresponsible not to want to push for that. But then we have to wrestle with that fact that there is a computer making these decisions.

What do you think will happen first—me being able to hail a self-driving Lyft, or a Lyft IPO?


Okay then. 

So, if you could go back to the beginning, would you keep the pink mustaches that you hinted you’re moving away from onstage today?

I would. Maybe we had it for a few extra months or something, but it was meant to be a launch tactic, but there’s some whimsy and humanity built into our brand. That’s who we are. I don’t want a future that is transactional; I want a future that is human, and that’s our constant reminder, and that was our token of that in the beginning. It’s part of who we are, and how we treat our drivers, and our riders, and I think you’ll see that in the brand work we’re doing in the fourth quarter. It was an amazing word-of-mouth marketing tool.

You say you want a future that is human—how does that fit in with your autonomy plans?

It’s a hard question. It’s a question I think about a lot. But like, driving everyday to work—I think we spend on average 70 minutes a day doing that—is not a needed human behavior. [In the future,] I think my daughter will be like, “You drove? You actually did that? What a waste of time.” And then on the way to work, or school, you’ll be able to interact with people, and we won’t have to position the car seats all facing one day—we can have a more community-oriented experience. The other side of that is car ownership itself, and the need to park, has impacted our cities in a very detrimental way. Los Angeles is like two-thirds paved. So humanity for us, we have to treat those impacted by that correctly and responsibly, and I think we’ve demonstrated as a company that those are our values. But the impacts to our cities—technologists spend so much time optimizing pixels in their apps, why aren’t we spending more time designing how our cities are so more people interact? New York is a great example of one of the most diverse groups of people in the world, and yet, people don’t talk to each other. I do think city design plays a huge role in that, and we are limited by the fact that car ownership is required, and that cars are parked everywhere.

I get that side of the humanity, but you’ve positioned yourselves as the company that cares about your drivers. How will they be affected as you move toward autonomy?

Well, it’s not going to happen overnight. So first, on the actual details, I don’t know 100% how it’s going to play out, because it hasn’t played out yet. Eighty percent of our drivers drive 15 hours or less a week, and therefore that more part-time driver uses the service for a year or two, and then they move onto something else. I think for the next couple years, the number of drivers will increase, because more and more people will move away from owning cars toward transportation as a service. At the point when you say that autonomous technology is ready, you wouldn’t be adding additional drivers, and there would be natural attrition of the existing drivers. That already happens because the majority of them are part-time. For the full-time drivers, there’s also attrition, just over a longer time. My point is that I think you can plan for this. It won’t be perfect, but I think you can plan for this to have the least impact on the driver community, and that the net number of jobs you can create as part of the multitrillion-dollar transportation-as-a-service business, you can provide opportunities for folks when the role of the driver changes. But you know, the details will continued to be worked out.