Ford Autonomous Vehicles President and CEO Sherif Marakby

Big car companies like Ford have broken themselves up to enter the market of self-driving cars and compete with Silicon Valley companies like Tesla and Uber. Ford Autonomous Vehicles hopes to become the go-to supplier of autonomous driving software, and have even set the ambitious goal of shipping a self-driving car without a steering wheel or pedals in the next few years. For this week’s interview episode of The Vergecast, Nilay and Verge transportation reporter Sean O’Kane sit down with Ford Autonomous Vehicles President and CEO Sherif Marakby to discuss the past, present, and future of self-driving cars. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Nilay Patel?

  No one likes waiting, especially for their phone to finish downloading or uploading something. And it turns out that to get the fastest data speeds, you need the right phone. Ookla analyzed over a million real-world results from a speed test app taken by users like you on AT&T and T-Mobile in Q2 and found that Android phones with a Snapdragon 845 from Qualcomm head up to 192% faster data speeds than non-Android phones with Intel modems. See all the results for yourself at qualcomm.com/vox today. Hey, everybody. It's Neil from The Vergecast, here with another interview episode of the show, a little experiment we're trying that seems to be going really well. This week, Verge transportation reporter Sean O'Kane and I talked to Sharif Marakbi, who's the CEO of Ford Autonomous Vehicles. I don't know if you know this, but big car companies

• 0:42Nilay Patel?

  have been breaking themselves up recently to compete with companies like Tesla and Uber. So Ford has broken itself up into Ford, the car company, and Ford Autonomous Vehicles, which is basically a Silicon Valley startup for autonomous driving. Obviously, they're going to use Ford cars, but their goal is to be a supplier of autonomous driving software. We talked to Sharif about how autonomous cars will work, whether they're even going to really happen, and how society is going to need to change to deal with it. We also talked about what Ford's going to do, and if they're really going to deliver on his ambitious goal of shipping a car without a steering wheel or pedals in just a few years. Check it out.

• 1:20Nilay Patel?

  Hello. We're here with Sharif Marakbi, CEO of Ford Autonomous Vehicles. Thanks for joining us.

• 1:25Speaker 01

  Thank you.

• 1:25Nilay Patel?

  So we have a bunch of questions about Ford. Sean O'Kane, our transportation reporter from The Verge, is going to join me. Hello. Actually ask smart questions. No pressure. I want to start with a very dumb question, extremely dumb question. Are autonomous vehicles going to happen? Are self-driving cars actually going to happen?

• 1:42Speaker 01

  They will absolutely happen.

• 1:43Nilay Patel?

  Absolutely.

• 1:44Speaker 01

  Absolutely.

• 1:44Nilay Patel?

  On what timeline?

• 1:45Speaker 01

  Well, it depends. It's a complex question because autonomous vehicles are going to happen in certain places, on certain routes, in certain places in cities, and they're going to continue to expand from that. So when you say autonomous vehicles, there are some test cars on roads today. With safety drivers and at some point there won't be safety drivers, but it will still be limited to certain areas and that's how it's going to first be deployed.

• 2:10Nilay Patel?

  Do you feel like we're in that sort of classic Silicon Valley disillusionment where, you know, everyone had the hype and we saw how it was going to work and we all watched CGI renders of cars and now they're real and they, you know, they don't quite work as well as people thought and we're gonna, it's gonna take a long time to come out of it? Or do you think the cars we have now are slowly and steadily getting better?

• 2:31Speaker 01

  Well, there definitely has been a hype. There's been a hype in the last couple of years and you judge by some of the feedback and some things that are happening in the last six months. And reality is kicking in and there will be autonomous cars, but they will be responsible autonomous cars in certain places. It's not going to be the big hype.

• 2:50Nilay Patel?

  Yeah. Okay. So I ask my super dumb question.

• 2:52Speaker 01

  Oh, no.

• 2:53Nilay Patel?

  I want to start with the foundation. Let's talk about you for a second. So before you were at Ford, you were at Uber. What were you doing at Uber?

• 2:59Speaker 01

  I was at Ford for a long time before that and then I, a couple of years ago, I left and worked at Uber in their autonomy group. And I was integrating autonomy into OEM into cars. So I did that for a while and then I came back to Ford about a year ago.

• 3:15Nilay Patel?

  And you came back to Ford to start this group?

• 3:17Speaker 01

  Yeah. So we actually started about a year ago and started with autonomous vehicles and electric vehicles. Ford's made a lot of good progress in electric vehicles with 11 billion dollars of investment. A lot of electric vehicles are going to be hitting the road in the next few years. And then about a month ago, I actually moved into this role of just being dedicated to autonomous vehicles. It's an ex big thing and it needs a lot of attention.

• 3:42Nilay Patel?

  So Ford is kind of restructuring, right? This is a new group. You're the CEO of a new group. How much autonomy do you have as CEO of Ford Autonomous Vehicles?

• 3:51Speaker 01

  Well, it's, it's, it's only been a month and I'm pleased to say it's actually been quite inspiring and autonomous.

• 4:01Speaker 01

  You know, thinking about we're putting the structure in place, we're putting the operation in place, we've got leadership already in place, a combination of Ford leaders, people that have been in Ford that think differently in a Silicon Valley way, and some outsiders. So that leadership is in place and now we're running and we're putting a lot of more things in place to be more like Silicon Valley environment than an automotive environment.

• 4:25Nilay Patel?

  That's a change that's happening all through the auto industry, right? We're seeing the other big automakers kind of break up in the same way?

• 4:29Speaker 01

  Absolutely. And in a way, the automotive industry, the way we've known it for 100 years is changing. And it's, you could say it's being disrupted. I see it as more of an opportunity to get in. And there's actually a lot of strengths in automotive and there are different strengths in Silicon Valley thinking. And you got to bring those two together, which is the trick.

• 4:50Nilay Patel?

  Yeah. So you, Ford has said you will produce a car without a steering wheel or pedals by 2021. Is that still the plan?

• 4:58Speaker 01

  That is, that is still the plan. And we already got going on that. And we've been actually running with that for a while now, with Argo being the software developer.

• 5:08Nilay Patel?

  Yeah.

• 5:08Speaker 01

  And then Ford doing the vehicle.

• 5:10Nilay Patel?

  So are you, you're testing self-driving cars right now?

• 5:13Speaker 01

  We have been, and we have vehicles already building, we're building the third generation of self-driving cars as we speak, to be on the roads in the next few months. We already have the second generation that's being tested in different places.

• 5:26Nilay Patel?

  And do you subscribe to the idea that the race to self-driving is really a race about miles driven and data collected?

• 5:32Speaker 01

  No, actually. I don't think it's just miles, because you could, you could take the car on the interstate and put as many miles as you want. We actually don't subscribe to that. What we, what we do believe is the right miles, and not only the technical right miles that were, you know, the clearly big urban areas where there's a lot of pedestrians, they happen to also be the right places for deliveries and moving people and moving goods. So those are important miles. They're not just all miles created equal. And the one unique thing we've approached this, this problem, if you will, is We're running the technology and the mapping of autonomous cars on the same streets that we want to run the business and profitably.

• 6:15Nilay Patel?

  Mm hmm.

• 6:16Speaker 01

  So that is, we think, one of the unique things is having profitable, right miles in the right places and then mapping or running the technology to make sure it works 100% of the time in the same places.

• 6:28Nilay Patel?

  Yeah. So 2021 in car years is like tomorrow, right?

• 6:32Speaker 01

  It's already started.

• 6:33Nilay Patel?

  It's, yeah, it's, you're already in production. So your 2021 car without a steering wheel and pedals, is it done being designed? Is it a clay model? Is it, are you just going to build software in a computer and bolt it onto a Mustang? Like, what's the plan?

• 6:46Speaker 01

  The design is very mature. I don't want to say it's all completely done, but it's well underway, the design.

• 6:53Nilay Patel?

  Existing Ford, like, platform and chassis? Are you building something new?

• 6:56Speaker 01

  We're building a new vehicle that is custom-built for moving people and moving goods by 2021. So you can imagine, we have clays, we have designs, and it's all running to the automotive cycle of getting the scale and everything. The vehicle will be not only durable for hundreds of thousands of miles, it's going to be designed for the business of autonomous cars. So that means the business of revenue per mile that we want to generate, the utilization is really important. Having the car run all the time is very, very super important for this.

• 7:31Nilay Patel?

  This sounds like you're, you're talking about ride-hailing, you're talking about fleet delivery, you're not talking about, I'm going to buy the car.

• 7:38Speaker 01

  Correct, correct. What we're developing right now is what we call Level 4, which is a service. So we will manage those vehicles to move people and move goods, and they're not for sale yet, not the first generation. Ultimately, Level 5 would be, you know, being able to sell a car. We think that's kind of a next thing, but right now we're focused on Level 4, which is a step towards Level 5.

• 8:02Nilay Patel?

  So Level 4 is no steering wheel, no pedals?

• 8:04Speaker 01

  Correct.

• 8:05Nilay Patel?

  But Ford owns all the cars. That's a kind of a different definition of Level 4 than I've heard in the past.

• 8:10Speaker 01

  Not to get technical on the definitions, but it's providing a service for moving people and moving goods in a geo-fenced area. So if you think about New York City or San Francisco, we would define what area of the city the vehicles are going to run, and we're going to map that area, and we're going to run the business in that area. And that would be a service that we'd provide. That's the definition of what we call level four autonomy.

• 8:37Speaker 02

  I would imagine that one thing that this LLC that you guys have spun out does, and correct me if I'm wrong, but helps move you away from some of the pressures that Ford faces as being a publicly traded company. Obviously, you're still within Ford, so all that stuff matters. But is that one of the reasons for going this direction as opposed to just having sort of a self-driving department? You can operate in this way that you don't really have to worry as much about the stock price?

• 9:02Speaker 01

  Yeah, so a little bit of background over the last year since Argo was created. Argo is a software entity. It's already a separate company doing the software. And we felt, and we've already been going down the path of you have two different cultures coming together. You have the software development, agile development, the Silicon Valley moving fast. And then you have the automotive, build at scale, build with safety, build so that you can have tens of thousands of cars coming together. The benefit of the new structure that we're created is to actually bring all of that under one entity that we can make decisions and move fast. So you still need to maintain the two ways of thinking because you need them both.

• 9:46Speaker 01

  But you can make decisions fast that you have the entire, call it value chain, everything about it, whether it's the business or the tech. So it's really targeted towards speed of decision making and being able to be autonomous, if you will, to make the decisions with the capital that's injected by Ford. So Ford has provided the capital that we're going with, but we're also open to future investors and investment.

• 10:11Speaker 02

  So into the Ford autonomy.

• 10:13Speaker 01

  Into the, exactly, exactly. Now, that, of course, will change the structure in the future, but it's setting the entity up so that it can move fast, but at the same time attract other entities, other outside investors into it.

• 10:26Speaker 02

  Right, but what are the boundaries of that? Like, can you, right, so Waymo is putting their software into like Chrysler Pacifica vans. If you're like, hey, Jim Hackett, I have some bad news for you. The Ford Transit Connect is not the platform I need for self-driving. I'm going to start using Chrysler Pacifica vans. Can you do that?

• 10:42Speaker 01

  Well, I mean, realistically, Ford has really a great variety of platforms that we can build from.

• 10:50Speaker 01

  And manufacturing, design, it's actually one of the things, one of the advantages we think we have because it's transparent. So we can make changes to a vehicle, we can make design changes, we can customize the car to a service that we're providing. And it's so much easier to do it within Ford and within the Ford, you know, design and manufacturing system. So I think our plan is to use Ford vehicles.

• 11:13Nilay Patel?

  I would imagine.

• 11:15Speaker 02

  Looking at it another way, like, I think one of the things that people are maybe more thoughtful of, in some cases even concerned about these days, is looking at investment coming from China, because there are a lot of automotive companies that are trying to beat out the competition there. And autonomous is going to be sort of the, a big opportunity there, is a potential investment from China. We've seen investments from Japan, like Softbank's venture fund and everything into this space. But is that something, like, instead of just, like, drawing the line at, like, what car might you use, what would you be looking for in an investor? And would you be critical of, say, something coming from China?

• 11:52Speaker 01

  Well, I think as we're just getting going, we've been having discussions around the type of partners. It's more important that we have the right partner to help successful business than just the pure investment. And partners could mean, Argo, for example, is set up to actually attract other OEMs, which would make the platform for autonomy bigger and have a bigger market share and more of a bigger entity. We could also have partners around, we'll call it the transportation as a service, or the network, or the service itself. We could have partners that are tech companies that provide content in those autonomous cars. We can have partners that are just pure investors, that can bring other pieces

• 12:35Speaker 01

  together. So in all the discussions we're having, we'd be open to any of that. It's very important that it's strategic and we think about, it could actually make the business better from a partnership standpoint.

• 12:47Nilay Patel?

  So are you thinking of Ford autonomy as a supplier of autonomous driving services and computing to every other carmaker? Is that the goal?

• 12:55Speaker 01

  Our initial plan is to have, not just provide the autonomous car, is actually to run the service. Into multiple, being able to have multiple companies that we provide that service to. And we would share in the revenue. So what that would mean is we announced last year that we have partnership with Postmates, for example, which delivers goods. We announced Domino's, we've announced Lyft, and we continue to work with others. So because utilization and having the car run all the time and create revenue is fundamental and also help the environment that the car is in, utilization is key. So we will have the capability within this AV LLC to provide all of those services

• 13:39Speaker 01

  through APIs into the different companies. And we actually can even share the cars across different services to get that utilization.

• 13:47Speaker 02

  Does that mean that, so you know, someone listening to this, that's like, all right, you say there's a car coming in 2021, how am I going to use it? Does this mean that they might fire up the Lyft app and they might get a Ford self-driving car or the Uber app and get a Ford self-driving car that's using all these, you know, it's got your infrastructure, it's got your software, it's got your hardware, all the things that you've talked about that you've built? Or are you working on the sort of platform that someone who needs a ride somewhere will say, I know the Ford app will get me to wherever I'm going in this particular way that I want to travel?

• 14:18Speaker 01

  So that's really good. So the first, the first part is yes, we are making the service available through others' apps, like you're talking about Lyft or Uber, others. And then a Ford self-driving car would show up, would, you know, provide that service seamlessly. But we're also building the capability that we could, if you think about a very typical example, a local business that doesn't have the technology, doesn't have the network, we actually have that. We can provide that to them to deliver things for them. And by the way, we've actually been running tests with small businesses, local businesses in Miami. And we've been running tests also with with companies like Postmates in Miami since February of this year.

• 15:01Speaker 01

  Now, they're not autonomous cars, but they were testing the system. We're testing how that would work. But we're actually covering both both elements. We can API into someone's app and someone's network, or we can provide that service to a business. And we actually think that both are needed because the different businesses have different capabilities.

• 15:20Speaker 02

  You've actually said before in some interviews, heard you talking about, especially with the Domino's trial, that you've come across some interesting insights that you weren't expecting that are like a little bit weird and maybe not directly related to the technology of autonomous vehicles. But like the example I remember hearing you give is we weren't expecting that one problem we would run into with this autonomous pizza delivery service is that customers don't wear shoes in the summer, so they don't want to come all the way out to the street to get the car or to get the pizza from the car. Are there other things, now that you've been in Miami especially for a couple of months now and testing around the country, are there other things like that that you've come across that are not directly related to solving like the big problem of self-driving software

• 16:03Speaker 02

  and hardware and all the questions that like we think about as big challenges?

• 16:07Speaker 01

  Absolutely. We actually have a very long list of learnings since we went to Miami. And just to give you a couple of examples, when we ran the test in, it was in Ann Arbor, Michigan last year, more of a suburbia environment and subdivisions, we ran into some of these examples where people would come out in the summer and actually, one is a lot of people in Ann Arbor, Michigan were open, a high percent of people were open to having an autonomous vehicle come to them. They go punch their, you know, their code and get it. What we've learned in Miami is if you're in a, you know, 20-story building downtown, you're going to be not as likely willing to come down and pick up your pizza.

• 16:50Speaker 01

  So in urban, in high-rise urban condensed areas, we're seeing a lower percent of people that are willing to come down and grab their stuff. Also, curbside pickups and things like that are more difficult in that. The other example we learned in Miami, and this is more around the business of running service, is when we're going into a city and we're assuming that, okay, here are where all the businesses are downtown, and therefore the delivery we think is going to be in these particular areas. Well, we did well guessing where we're picking the stuff from, like restaurants and businesses and things like that, but we got it really wrong where the deliveries are going to.

• 17:32Nilay Patel?

  Really?

• 17:33Speaker 01

  Yeah, so we actually, a few months ago, started remapping different areas.

• 17:38Nilay Patel?

  Wait, what'd you get wrong about where the deliveries were going?

• 17:40Speaker 01

  Uh, well, I mean, you know, every city is different, so we need to work with the partners. This really emphasizes working with companies like Postmates to understand where the deliveries are going, and then start mapping the area. So we learned, for example, a lot of things are being delivered from a condensed area to a less condensed area, so like suburbs and things like that. So we look at where the profitable routes are, where the businesses is, and also working with the city is very fundamental to understand the patterns of traffic and things like that. So, understanding from the city traffic, understanding from the business where the deliveries are going to, led us to map different streets, if you will, because that's where the business really is. So that, we actually adjusted our plans in Miami in the last few months.

• 18:25Nilay Patel?

  To go into more suburban areas?

• 18:27Speaker 01

  That's, that's an example. And it's not just suburban, it's just different areas that we didn't expect the deliveries were going to go to.

• 18:33Nilay Patel?

  Yeah. Do your self-driving cars in Miami, like, wave at the very many Mustang convertibles that fill the streets?

• 18:39Speaker 01

  Oh, we love Mustang convertibles.

• 18:41Nilay Patel?

  That's what I want.

• 18:42Speaker 01

  We love them, yeah.

• 18:43Nilay Patel?

  I just want a self-driving Mustang convertible.

• 18:44Speaker 01

  Oh, yeah. No, we love, we love, especially in places like Miami. So, yeah, I mean, it's, it's, it's not a Mustang that we're, that we're running.

• 18:52Nilay Patel?

  That's all I'm trying to get. I don't know if you've noticed, the theme of this interview is when do I get a self-driving Mustang?

• 18:58Speaker 01

  Well, you know, it's, we'll see, we'll see where this goes. I mean, you never know.

• 19:02Nilay Patel?

  Actually, the Mustang brings me to a real question, which is that Ford in North America, you have a new CEO, Jim Hackett, Ford, the car company, not the autonomous company, has stopped selling all cars, all cars, except the Mustang, and now it's all basically trucks and SUVs and crossovers. Is that affect you? Do you, are you thinking about that? But maybe it won't be self-driving cars. Maybe it won't be this like beautiful sedan with screens all around that you have tech companies putting content in, but really still people just want F-150s. Is that affecting your plans?

• 19:33Speaker 01

  Not really, because the first self-driving car, the, the purpose-built self-driving car that we're planning is, is custom-built for ride, you know, moving people and moving goods, which happens to not be a sedan.

• 19:45Nilay Patel?

  Yeah.

• 19:45Speaker 01

  The actual car we're actually testing today is a Fusion Hybrid, because that's actually a lot of companies are using that vehicle for that. But that's not the, the, the vehicle we're going to be launching in 2021. So it really hasn't affected our, our plans. And we think that where the company is going is responding to market trends, where people really would want those types of vehicles, which is really pretty synergistic, actually, with where autonomous vehicles are going anyway. So it really hasn't affected our plans.

• 20:14Nilay Patel?

  Synergistic, how? Just because they're bigger and could fit more people? Like, yeah, also could fit bigger batteries when things go electric?

• 20:21Speaker 01

  Yeah, I mean, both. I mean, those are two good examples. So when you're talking about moving people or moving goods, you can envision more of a open, high-roof, easy-entry-exit type of vehicle, which actually happened to be more of those vehicles that, that, where the market is, is going. And also, like you said, our first autonomous vehicle is going to be a hybrid electric vehicle. And we chose a hybrid electric vehicle because we know that is a better choice for utilization, that it can run all the time and, and it's still good on gas, but it's, it can run close to full day. As we progress that, we know that battery electrics are going to be part of autonomy in the future.

• 21:02Speaker 01

  And when you have a high-riding vehicle, it's more synergistic with a battery electric. You can actually put a big battery there because we need a big battery to go many miles.

• 21:13Nilay Patel?

  This episode of Vergecast is brought to you by Microsoft Azure. They sent us a little clip about the difference between big data and intelligent data. Check out this advertising segment from Microsoft Azure.

• 21:24Speaker 03

  In a city, everything we do, from swiping a ticket at the subway turnstile to checking into the public wifi, is a piece of data.

• 21:32Speaker 04

  In just the last two years, we've produced more data in the world than the rest of the years combined. We're getting into multiple petabytes, even exabytes of data.

• 21:39Speaker 03

  That's Rohan Kumar. He's Vice President of Azure Data at Microsoft. And that's exabytes, as in one megabyte with 12 zeros at the end. It's a lot of data.

• 21:50Speaker 04

  And with this volume of data comes a lot of potential for intelligence. When we say, how do you make the data intelligent? It's really about getting into a single place where you can run machine learning models which derive artificial intelligence from that data to make it more meaningful.

• 22:07Speaker 03

  One application is making our cities smarter. Ayal Federlevi is the CEO of ZenCity, a startup that's trying to make the data-driven city a bit more intelligent by adding people to the mix. ZenCity takes a pulse of what's on the city's mind through social media. And then they analyze that data for insights using Azure services.

• 22:28Speaker 05

  Posts, comments, 311 calls to the call center. We use machine learning to turn all of that huge mess of unstructured data into actionable, quantifiable scores and measures that can actually support decisions in the local government.

• 22:43Speaker 03

  Those millions of data points can help city governments understand what their citizens need.

• 22:48Speaker 05

  When we had big snowstorms in the US this winter, we saw some of our cities actually use our platform to recognize areas where more people are talking about being stuck in the snow and prioritize snow removal.

• 23:01Speaker 05

  There's so much data being created around every action that we take in life. The real question becomes what we actually do with it.

• 23:09Speaker 03

  Learn more about the tools you can use to make sense of the world and your business with Azure. Start free at azure.com/trial. A-Z-U-R-E dot com slash trial.

• 23:26Nilay Patel?

  Want to thank Microsoft Azure for sponsoring the Vergecast and supporting the Vox Media Podcast Network. And now, back to the show.

• 23:33Nilay Patel?

  So, we've talked very confidently about self-driving cars happening. That's the first question. But there's been a spate of accidents and mistakes. Uber had a fatality last year in Arizona. How are you thinking about those things happening? Like, I know you guys just released your transparency report. I think every company that feels like they're confident in the space is starting to do that. But how are you thinking about keeping these things safe?

• 23:59Speaker 01

  It's actually, the fundamental vision for our team is safety and trust. And that was in the report that we published. And what that means, it could mean a lot of different things. Safety is non-negotiable. And especially, that's one of the strengths of an auto company. When you think about redundancies, and when something happens in the car, how does the car respond? If something fails in electronics, what do you need to do in steering, braking, acceleration, those types of things? So, we're doing all of that, and that is not negotiable. We're doing all this stuff. And that's one of the reasons this is developed for scale. Because when you talk tens of thousands of vehicles, and you have, everyone has to do

• 24:40Speaker 01

  that. The working with Argo, which has really been a pleasure, when I look at the industry itself, have an Argo mindset with a lot of robotics capabilities of the people that are there. Many of those folks, especially the leaders, have actually delivered autonomy robotics, maybe not in cars on streets, but in other industries. So, they understand the safety aspect of it. Tied to safety is trust. And the consumer has to trust the autonomous vehicle. And we know that's a challenge in the industry today. So, we're spending a lot of time thinking about what does trust mean? It could be very simple things, like the car knows you, that you know the car, it opens the door, and you get in, and you

• 25:23Speaker 01

  know everything is safe when you get in. It could be, as you're riding in the autonomous car, how does the car accelerate, decelerate? It gives you situational awareness of what's around the car. You see it, you know the car is seeing the same thing you're seeing. You become more confident in the car. And it could be the actual mechanics of safety and how the car responds in bad situations. So... We're actually tackling all of that and it's one of the fundamental principles of this team.

• 25:54Nilay Patel?

  Isn't the kind of the most important one other people, though? That you see an autonomous car on the street and you trust it's not going to hit you? Or you're driving your regular car and there's autonomous cars around you and they're not acting in alien ways? Right? Like, how are you thinking about that part of the relationship?

• 26:12Speaker 01

  Sure, sure. It's both. It's actually the consumer that's in the car, but you're right, the consumer outside the car, the pedestrians, the cyclists, the people around it. And we've actually done some testing in Virginia, for example, last year of what do, when there's an autonomous car and there's no driver, how does the car, usually if you have a driver, you have eye contact, you can actually wave, you can almost sense where the driver is going to go. So we've been doing a lot of testing with lighting and signals and seeing what is the natural reaction of people when the car does a certain thing. And I'm not talking about just turn signals. I'm talking about big light bars, colors, things like that on vehicles. And we've actually learned some things through that because we're entering a brand-new space here.

• 26:57Speaker 01

  There are no autonomous cars with no drivers on the road. And when you go to some of the places that have a lot of autonomous cars and people start getting used to them, but today you still have a safety driver. When you pull them, we're learning some things work, some things don't, and we're actually incorporating some of those changes in our autonomous car.

• 27:16Nilay Patel?

  Give me an example.

• 27:17Speaker 01

  Lighting. You know, as you have a big light bar and the car wants to stop and it's not sure, the pedestrian is not sure whether they're going to cross or not, the signal that the car gives could indicate, "Go ahead, I'm going to stop, I'll wait for you," or it could say, "I need to go." So that's an example of something we've learned and we've seen human reactions being very unique. Some examples people just don't get, and some things people do get.

• 27:44Nilay Patel?

  Do you think that has to be standardized across the industry, the way that headlights and taillights are standardized across the industry?

• 27:49Speaker 01

  I think eventually it would need to. All the good practices that we're all learning as an industry feel like it does need to become standard, especially things that people get. Like, if there's a behavior that the autonomous car does and people get, that should go across all vehicles. We're just starting and we're learning.

• 28:05Nilay Patel?

  Do you think that you're going to need more vehicle-to-vehicle communication, more vehicle-to-infrastructure communication to make this actually work?

• 28:12Speaker 01

  The first generation of autonomous cars are heavy, I'll call them heavy processing and heavy sensors, and as you are aware, there is no infrastructure today. There is no electronic infrastructure or V to V. As we get more infrastructure and more, especially when you think about a city, this is one of the advantages of working with the city. If we work with the city and they're putting transponders or signals on streets that help the car not need to guess what's happening, that would all help. So we see progression in the future of autonomous technology having less and less on vehicle and more and more around the vehicle, and that would all help.

• 28:54Nilay Patel?

  But this is, I mean, this is a point of debate right now, particularly as we start to see some of the accidents that other companies have had. Some of the response from Silicon Valley is, you know, actually, we all need to adapt to the cars. The city needs to adapt to the cars, pedestrian behaviors need to adapt to the cars, which seems like asking a lot, right? Like, that's a lot to ask, but it's certainly an idea that's out there. Do you think that you can build the fully self-contained, self-driving car that does not need the city or the pedestrian or the community to adapt to it? Or do you think that it's going to be this, like, long process where, you know, you build your first fleet of Level 4 cars that are delivering stuff, and that pushes the next

• 29:35Nilay Patel?

  city to say, okay, we're going to do V to I transponders everywhere so we can have more of these things? Or Waymo and GM are going to come to you and say, all right, we need a V to V standard so our cars can all talk to each other, and that's going to... Which path?

• 29:49Speaker 01

  Yeah, so I think maybe there are two paths going on right now, is we can't wait for the first generation to have all that, because as you said, you need all cities to adopt certain things or standards, there's all kinds of stuff. So we're developing the first generation of self-driving cars to be self-contained but limited in operation, like we mentioned, on certain streets. And there will be some cases where you cannot deploy an autonomous car. It's just too complicated or for whatever reason. I do believe, we do believe that it's going to be a gradual thing where it's going to build up. Cities will realize that it could actually help by doing some of that infrastructure. And that's why we have a city solutions team that is fully focused on that, actually working with the cities to understand the traffic patterns, what are the issues, the

• 30:34Speaker 01

  cities, and every city is different. And then figuring out where is the journey of autonomy in that city. And then ideally, if we were to go into a perfect world from an auto standpoint, yes, we'd like to have all cities have the same standards and V to V and V I and all the stuff, but that's unrealistic to think about that in the next few years. So we think it's going to be a gradual thing after the first generation.

• 30:58Nilay Patel?

  You don't think V to V could be faster than that? I mean, like Wi-Fi exists, right? Like, it's like this thing that happened and like Apple's like, here's an iBook, it's got Wi-Fi and then everyone else used Wi-Fi, right? You don't think that you can have the cars broadcast, here's my destination, here's, here are the move, the five moves I'm planning to make next, here's my speed, and just have that immediately available as a standard?

• 31:19Speaker 01

  Oh, we could. I mean, it's just what it takes to make some of these things happen in the real world is more the time. I agree with you. I don't think it's necessarily the technology because the technology is available. It's a matter of how do you get all the parties involved to actually put that in place? And then also the infrastructure and the cost. I mean, we're talking about an industry, an auto industry, a Silicon Valley industry, and cities. So you're talking three different entities that have to come together to make that happen.

• 31:48Speaker 02

  Here's a dumb question from me. Why? This sounds like a lot of work and you guys are making a lot of progress and other companies are making a lot of progress, but why do we need self-driving cars?

• 31:57Speaker 01

  Well, we need self-driving cars for a couple of just high-level reasons. One is there are transportation people and city needs. If you think about, there's a segment of the population, elderly, that don't drive and don't have access to transportation. We definitely see self-driving cars solving people issues in cities. We see them solving city issues. Now, you can argue that it could make it worse in some place. And better in other places, and I agree with that. You just have to really thoughtfully think about it with the city. How self-driving cars can help traffic patterns, congestion, everything else. But the other piece and why, why a lot of these companies are investing billions of dollars in this technology

• 32:42Speaker 01

  is because there's a real viable business for autonomous cars. So when you think about autonomous cars delivering or moving people and moving goods, you can provide that service at a fraction of the cost as today's services. And that, from a societal standpoint, that helps. But also from a corporate standpoint, company standpoint, it's a viable business that you can create revenue, you can create profits based on that. And in a way, you can call that a disruption to car ownership. I'd like to see it as more of an opportunity that the industry has to get into this space.

• 33:18Speaker 02

  Is the profit wholly dependent on being able to run these 24 hours?

• 33:23Speaker 01

  Not 24 hours. Actually, when you look at the business opportunity where you can get the cost of running a service closer to, let's say, a dollar a mile, which is pretty low compared to most transportation today, that realizes a certain level of utilization that is pretty realistic. It isn't 24/7. I mean, the more, the better, of course. The more you can run the vehicle and the service with people in it or goods in it, the better. But with modest, realistic assumptions of a good portion of the day, let's just call it more than half the day, you can actually create a pretty viable business at those prices.

• 33:58Speaker 02

  I noticed you didn't say as one of the reasons for why is the 37,000 deaths a year on the road. Is that part of the Ford mission statement? Is that a different priority compared to the other two things?

• 34:09Speaker 01

  No, it absolutely did, and thanks for bringing this up. I mean, when we talk about safety and we talk about-

• 34:15Speaker 02

  That's so smart now, Sean.

• 34:15Speaker 01

  Yeah, that's a smart question. No, absolutely. I mean, the fundamental driver for autonomous vehicles for society is safety. Today, like you said, 37,000 deaths and 100,000s or more globally, and 95% of these accidents are human-caused, so driver-caused. Driver or the driver and the other car. And less than 5% is vehicle, which we still have to keep working on. The self-driving cars, we believe, is going to substantially reduce the human-caused portion of accidents, which is the majority of the accidents that are happening today.

• 34:56Speaker 02

  So presumably, that means that the cars, the self-driving cars, to reduce those deaths are going to be safer than human drivers because they're reducing those human errors. How do you measure that safety at Ford?

• 35:08Speaker 01

  It's a really good question. So a couple of things. One is the self-driving cars, you can talk about very basic things, like you're probably not going to see a self-driving car texting while driving, right? You know, stuff like that.

• 35:19Speaker 02

  Okay.

• 35:19Speaker 01

  But those are, those are actually many of the reasons why things happen on the road today. The other thing is that the substantial amount of sensing that are happening in the first generation, at least, where you have LIDARs going around the vehicle, you have full coverage in 360 around the car as it's driving. We believe that's going to be a big factor in why it would reduce that. In terms of the, how much and how we measure it, that really depends because you could, you could argue, if you're running on certain streets in certain cities, you have or you can get existing statistics of what happens, right? What, you know, what are the accidents in this particular area of the city? And then

• 36:00Speaker 01

  you can measure how self-driving cars, over time, is going to be better. But it's a really hard thing. Obviously, there's driving in tough places in urban societies that we're doing today, measuring interventions and takeovers, manual takeovers, running simulation to make sure that we can project any, you know, why did this takeover happen? How do you simulate it so it never happens again? All of these are the real things that go into reducing that accident rate. And then the actual comparison has to be done compared to what's happening today, which, there are a lot of statistics, but it's also different for different places and cities.

• 36:38Nilay Patel?

  But how do you take all of those numbers? And we, you know, we read your transparency report, we read Waymo's, we read everybody's numbers, and then one autonomous vehicle hits one person in the street, right? And there's no way that those numbers abstractly can make people feel safe. The cars don't behave like human drivers. And I doubt they ever will because human drivers are unsafe, right? So like the driver, the cars are more cautious. They be, they, they look a little bit alien. They might have these giant light bars. How do you turn the reality of how safe you want them to be into the perception of trust that you've talked about?

• 37:13Speaker 01

  Well, that's, that's a really good question. And we're going through every piece of autonomy to say, how do you build trust? Because you're right, you can have, you know, people are talking about the one car that got hit by something else or whatever. And people aren't doing that on normal, you know, you know, normal vehicles today. We are focused on how do you build trust all the way from when you order your vehicle, you get into the car. If you're a pedestrian outside the car. And those examples that we talked about, like light bars or, or, you know, the vehicle knows you, but they're also looking at what's happening around you. And there's a call center that's monitoring things, you know?

• 37:57Speaker 01

  That's how you build trust slowly over time. And people will get comfortable that I am comfortable getting into this vehicle. I'm comfortable getting a ride in this vehicle. How the vehicle rides, you're right, most autonomous cars today are very cautious and they'll probably continue to be cautious for a while. But in some cases, you can make them, we believe you can actually make them more human. Not human that make mistakes, but more human like the car accelerates normally. It's, it doesn't just get confused at a light and just stops, doesn't know what to do. Those are the problems we're working on because that actually takes away trust. And that's what we're working on.

• 38:35Nilay Patel?

  So you were at Uber. Uber was famously sort of a winner take all business model, right? We're going to go everywhere we can as fast as we can. Everyone will use Uber and then the lawyers will come later, right? And that's basically what happened. Is autonomy a winner take all business? If, if you get there first and you've got level four ride hailing services, do you foresee that as the winning move? Because you have competitors who I think Waymo says are going to do taxis this year. Cruise and GM are going to do something next year. Are you worried about their timelines? Do you think this is winner take all and you're coming late? Or are you saying this market's big enough for everybody?

• 39:11Speaker 01

  Well, we definitely don't think it's a winner takes all. There are three trillion miles driven every year. And when you look at the opportunity for running a service, whether it's people or goods moving in a specific city, there's plenty of opportunities. So we feel that coming in, building trust, building safety, coming in at scale with a purpose-built vehicle that people are gonna want to get into and want to transport goods and is actually a very good formula. And we're in a race. There's no question that we're, you know, we're running this to 2021, three years from now. And but we're focusing on those elements more importantly than just being there first with a few vehicles. That's something you could do.

• 39:55Speaker 01

  I mean, we're not building prototypes for demos. We're building prototypes to test the technology and test the business in real cities. So we believe there's plenty of opportunities for a handful of players, of course, that will be in the market. And we definitely view that we're gonna be one of them.

• 40:11Nilay Patel?

  Do you think you're gonna sell a car to people, to consumers?

• 40:15Speaker 01

  Not in the timeframe of what we're talking about, the level four, the next three years. But the good news is everything we're doing applies to getting to that point. Because to be able to sell a self-driving car to people, there are a couple of big challenges. The first one, the obvious one, is you have to map the world and mapping the world precisely. Like, it cannot be, you know, you cannot make mistakes. And that's a huge effort in making sure that the technology works. The other thing is cost. The amount of sensing and the amount of computers and work that goes into these vehicles, we think is gonna be prohibitive in the next few years for somebody to just say, I want this feature. I want a self-driving car. It's gonna help me. It's gonna be very expensive.

• 40:56Nilay Patel?

  But you have competitors who are doing like level two ish things.

• 41:01Speaker 01

  We are.

• 41:01Nilay Patel?

  Yeah, right? Like they're there, right? You know, Tesla is out there, GM is out there, Super Cruise. Like, do you think that that consumer self-driving car, that path is going to supersede what you're doing? Or do you think it is you push the button, the robot shows up and that's the future of?

• 41:16Speaker 01

  There's a, there's a big step, just to get into the levels real quick. Level one, two, three is more driver assist and you're still behind the wheels and the car is helping you or doing a lot of work for you. And the whole industry is progressing that. There's a step function when you pull the driver, if I can use that word. Basically, you don't have a driver in the car. At that point, you have to have the redundancies, you have to have all of these sensors, you have to have all of that stuff. And the first one is level four, heading into level five. We, at least I haven't heard of anybody announcing they're going to sell level five. You know, that you can, you don't need to be in the seat at all.

• 41:51Nilay Patel?

  Look, Elon's going to tweet today. You have no idea.

• 41:56Speaker 01

  But realistically, what it takes to put all of these elements, the redundancy and the sensing and everything else, we are doing and we're doing in, in level four. As we map areas and the costs come down, I could see the path to level five.

• 42:10Nilay Patel?

  Because Ford is working on sort of some of those driver assistance stuff. You know, it's got this Copilot 360 that's being developed by their autonomous division. Or are you guys totally separate from the driver assistance stuff that's happening at the automaker level?

• 42:21Speaker 01

  No, we're, we're not separate. As a matter of fact, a lot of the technology, whether it's radars, ultrasonics and, you know, maybe small versions of LiDars start getting into the complex autonomy that we're talking about. We're very close to that and expect more and more from Ford along those lines. The Copilot 360 and more of lane centering, adaptive cruise and all of these features that are really useful to people, that people love to continue to get on many, many, many more cars. We're looking at those technologies as, hey, a piece of this technology can work in a driver assist and in autonomy. So in several cases, we're actually using the same technology that will migrate towards that.

• 43:06Speaker 01

  I was referring to more of to have a full up, no driver vehicle. You need to do a whole lot more than driver assistance and what you're seeing in, in the market. And it's going to cost a whole lot more. We're very optimistic, though, because we think the work we're doing to get to level four, plus the fact that if I if I look at the last hundred years of hardware and technology, the cost will come down. There's no question the cost will come down. It's going to be more and more affordable. And I'm very optimistic as we go forward to be able, the industry will be able to get to a point where you could sell a fully autonomous vehicle. We just don't think it's going to be in the next few years.

• 43:43Nilay Patel?

  So two more questions. You got to get out of here. One, we brought up Elon, and then you're talking about sensors. So I have to ask, Elon Musk has been out there saying LiDAR is a local maximum. It's not the future. We're going to do vision-based sensing of the rip. Do you agree? Do you think LiDAR is a local maximum?

• 43:59Speaker 01

  We have vision-based. We have cameras. We have many cameras around the vehicle, high-definition, you know, very, very complex cameras looking at the field of view. We do believe LiDARs are part of the first generation to make sure, again, that we can really get all use cases in the areas that we're running correctly. So we are utilizing LiDARs and we'll continue to utilize LiDARs. I'm not referring to the future beyond the next few years. As all these sensors get better, if the vision gets so good that it could capture all of the areas and, you know, surrounding the vehicle, we will use it more and we'll use LiDAR less. But today, we believe that LiDAR is part of the autonomy.

• 44:41Nilay Patel?

  Okay. And then last question, big and broad. You work at Ford. Ford's doing a lot of stuff. Ford sells more F-150s than anything, right? It's just like a huge part of that business. And Ford, new CEO, little bit of a turnaround going on. The company's changing, developing. Do you think Ford Autonomous Vehicles will one day be a bigger business for Ford than the F-150?

• 45:00Speaker 01

  When I look at autonomy as a business, and, you know, taking all of these things that we talked about, I do believe that autonomy is a massive opportunity for the auto industry and the Silicon Valley on the business side. I don't want to compare. We love our F-150s. We love our Mustangs. We love all the cars.

• 45:20Nilay Patel?

  I was hoping you were going to come out and take a shot.

• 45:22Speaker 01

  But I'll tell you, I know I love F-150s and I hope, you know.

• 45:28Nilay Patel?

  Jim Hackett's like texting you right now.

• 45:30Speaker 01

  I think I could see the world where autonomy will be on more and more vehicles as we talked about it, hopefully including F-150 someday. So I think that's going to be part of it. But we, autonomy is a massive business opportunity. There's no question.

• 45:46Nilay Patel?

  Well, let me, I'm going to try that question in another way.

• 45:49Speaker 01

  Okay.

• 45:50Nilay Patel?

  The F-150 is like, they sell it and they lease it to people, right? It's a consumer product in that way, and a fleet product in some other ways. But at the end of the day, there's a transaction, you take it, you walk away, unless you buy the 4G plan for the F-150. You don't have like the service relationship. You are describing a very big business model shift for Ford, that you're a service provider, that you're a consumer service provider, that you're potentially a computing vendor to the rest of the industry. Do you think all of that in total will be bigger than the consumer car business that Ford has represented for, you know, 150 years?

• 46:25Speaker 01

  Yeah, if I include, if, put autonomy aside for a minute, the entire service model versus selling a car and then, you know, when you think of connectivity, the vehicles are going to be connected, more and more connected. That provides through the app, through the service, through our mobility, we have a mobility division that has all of these different services that are running. And ultimately, there'll be, some of them will be autonomous and some won't. I do believe that the whole mobility in the automotive sector is going to be very big, whether it's going to be bigger than selling cars, maybe. It's a massive opportunity. Once you connect the cars and you provide the service, whether it's things from the car or things that people care about when you're sitting and, you know, watching what's happening with your vehicle, what's your tire pressure, that stuff is

• 47:09Speaker 01

  happening today. And that's a lot more than what was happening five years ago. That's going to continue. You think about the, any consumer product and that got connected, it's amazing what it transpires into. And I think that's going to happen in the automotive business.

• 47:24Nilay Patel?

  Very cool. Well, Sharif Marakby, thank you very much for being on the Vergecast.

• 47:27Speaker 01

  Thank you.

• 47:28Nilay Patel?

  I want that self-driving Mustang. So let me know.

• 47:31Speaker 01

  I will, I will, I will. Thank you so much for having me.

• 47:36Nilay Patel?

  Well, thank you so much to Sharif Marakby, CEO of Ford Autonomous Vehicles, for joining us on the Vergecast this week. I am really interested in knowing how you think these episodes are going. I think they're really fun. People have been tweeting me that they like them. But if you don't, I want to hear from you too. So tweet at me, I'm at Reckless. I would love to hear what you think.

• 47:54Nilay Patel?

  Here's an easy question. Do you want faster data speeds on your phone? Of course you do. Everyone wants more speeds. Why? Because waiting is annoying. Whether it's in line, in traffic, or when you have one bar of signal, it's taking forever to finish downloading your music. So how can you get faster data speeds without switching carriers? Turns out, all you need is the right phone. Ookla analyzed over a million real-world tests from the Speed Test App, taken by users like you on AT&T and T-Mobile in Q2, and found that Android phones with Snapdragon 845 from Qualcomm had up to 192% faster data speeds than non-Android phones with Intel modems. So check out all the data for yourself at qualcomm.com/vox, then upgrade your data speeds with a phone powered by the Snapdragon 845. That's qualcomm.com/vox.