Episode 1: Driving The Future

Amen. Tighten my skates first. I love rollerblading. I've been doing it since I was seventeen years old. Back then, rollerblades were still only really used for off-season speed skating training. But to me, it saves time, it's fun, and it feels a bit like flying. So every day when I arrive at work, I pull on my skates. Hi, can I have a one-shot cappuccino? Our mission control is an old shopping mall that we've spruced up a bit. Thank you. Have a nice day.

And as I enter the space, there's one thing that really sticks out. You can't miss it. There's a line painted all the way across the floor. I have to skate right over it. Written next to the line is a message. You may never cross this line. It's a stupid rule, right? You have to break it to get in and do your job. But the reason it's there is to help my team practice breaking stupid rule. This is about getting people past limiting beliefs and bad assumptions about what is correct or proper.

You want to race up the stairs? Most of the rules we all follow every day are just habits. I mean, that's just how it's done. Rules like you can't have fun at work. That's part of the reason I show up to work on skate. It's because we're trying to do something a bit different here. We want to help solve problems that seem impossible.

This is why it's a great place to be, because that is not the easy way to do it. That is a thinking out-of-the-box way to do it. Because the thing is, you can't change an industry for the better without breaking some of its most closely held beliefs. I had every reason to believe this is fundamentally impossible. And I was the expert. This is X, the Moonshot Factory, Google's innovation laboratory. And here our mission is to take Moonshot. It's striving to achieve the near impossible.

Working at the frontier of technological innovation to come up with radical solutions to tackle some of the hardest problems facing the world. Problems like energy security. We are trying to build Google Maps for the electric grid. Feeding a growing population. We need to do this work to save you know a billion people from potential starvation.

Predicting disaster in our changing world. If we're successful, in five years, everyone will have had the information they needed to prepare for, respond to, recover from natural disasters. Bridging the digital divide. A truly universal access to the internet without any speed breaks. Preventing and dealing with waste. Literally 40% of the

that we put through our food system in the United States ends up in the trash can or compost bin. And a moonshot almost by definition resists first principles thinking. If it didn't, someone would have already done it. The problem would already be solved. People here literally manifest things out of nothing. The way I see it is, to come up with a really interesting idea, one that's in the blind spot of society, you have to venture into the unknown. We can't help solve

what seems impossible without getting comfortable, being a bit uncomfortable. And I thought, oh my God, this is a

You're listening to the Moonshot Podcast. I'm Astro Teller, Captain. It's 2025 and the X is turning 15 years old. To mark this milestone, we're throwing open the doors to the Moonshot Factory to take you behind the scenes and show you how these moonshots get made. And as your host, I'm going to share some of our biggest successes and also explore some of our greatest failures.

And I'll introduce you to some of the people who make this stuff happen, to share their advice so that, just maybe, you too can go out and make a moonshot. Something here is working. So come on, let's go make some good trouble. So, I want to begin with our very first moonshot, but before we hear about how it started, let's hear how it's going today. Heading to our market.

Because this moonshot is already out there in the world, changing something fundamental about how we live our lives, how we get around. And yeah, don't worry, I've ditched the rollerblades for now. I'm gonna catch a ride in a Waymo, the first and only fully autonomous ride hailing service in the United States, with my old friend and colleague Sebastian Thrunn. This is so great. You and I have been in many self-driving cars together. When it was called chauffeur.

But we've never been in a Waymo, nobody in the front seat as customers. It's amazing. It's just amazing. I'm so happy. I have no words for it. And it's safe. Waymo is now operational in four cities San Francisco, Phoenix, Los Angeles, and Austin, Texas. The system is completely unique. The cars work a bit like taxis. Anyone can use them and simply pay to be driven to their destination. With no human behind the steering wheel. Waymo cars now drive customers over a million miles per week.

That's more than the average American will drive in their lifetime. Yeah, and then look at it right now. It drives confidently. It drives better than human drivers. It's amazing. About to arrive. Sebastian started Google Apps. And he and I go way back. In fact, it was Sebastian who asked me to co-found and run Google X with him. After Waymo, he went on to work on things like the future of education and flying cars.

But let's rewind a bit further to 2007. Back then, Sebastian Thrun was a professor of computer science at Stanford University. He liked to give his students interesting projects, and that year had asked them to find a way to photograph all the streets in San Francisco and organize them on a smartphone. The project went on to become Google Street View, and was how Sebastian ended up working for Google.

So we took the Street View asset and massaged it into a navigable street map, allowing Android smartphones to give turn by turn directions, which back in the day was actually a big deal. This is two thousand seven, two thousand eight. the iPhone had just shown up and no one yet had understood that smartphones can give directions. Sebastian is being modest here.

What he isn't saying is that it was his team at Google Street View that at breakneck speed gathered the underlying data to power Google's own turn-by-turn navigation tool. Securing the future of that tool for Google Maps. was the first step towards something else, something really big being possible.

because Larry Page, one of the co-founders of Google, had an idea. My memory is that Larry Page came to me and said, Sebastian, I thought about this heart. We should build a self-loving car that could navigate everywhere in California. Sebastian was not convinced.

For people who don't remember, let me just put this into context. Two thousand five, DAPA Grand Challenge, a robot raised organized by the US government, where robots were driven in the empty desert, and most robots made it maybe ten miles, fifteen, twenty miles in distance before they went up in flames. So for me, thinking about building a car that could drive down Lombard Street in San Francisco That sounded like homicide. Sounded like we're gonna kill people.

For those who don't know, Lombard Street is a famous street in San Francisco. It's so steep that in just a single block it has to make eight hairpin turns. And then Lee would insist on and said, Look, Sebastian, I thought about it's hard. Maybe in the beginning it's not quite safe yet, but as you progress, um you can make it safer and safer.

But I still felt, oh my god, that's so insanely hard, we can't do it. And eventually Larry convinced me and said, Look, even it can't be done, why didn't you want to try? And then it was more the technical challenge that kept me there than any kind of

system or or entity to build or reward, whatever. It was really the idea that we could try to build a truly self-driving car, which at the time was completely impossible. So the the technical challenge, I really get why that excited you but I know that you were also passionate about kind of the problem itself, like in the benefits to the world. Yeah, like everybody else, I'm a user of cars. Uh it's

probably the biggest invention in the twentieth century because it really ch transformed cities and and labor and independence. But it also kills a million two people every year. Okay. And one of those people were was my friend Harold when I was 18 who died in a traffic accident.

And there was a pivotal moment in my life where I realized, oh my god, we just kill people. Like we worry about a plane crash. This happens every morning for breakfast on American highways. And we just don't worry about it as much. So I felt like if you could make cars safer Yeah. And here we get to the problem to be solved.

Self-driving cars may sound like a whim, a piece of futuristic sounding technology, a nice-to-have rather than a must-have. But the truth is, they were invented to solve a problem. Driving is dangerous, riding a bike on the road is dangerous. Even walking in a space you share with cars is not without risk. Getting into your car and driving is probably the most dangerous thing you do every day. But the bet on Waymo was putting computers in control of cars would make them safer.

Our main goal is to build the world's most experienced, most trusted driver. Here's Andrew Chatham. He's a VP of engineering at Waymo for commercialization, scale, and infrastructure. But back when he joined Google's self-driving car project in 2009, then codenamed Chauffeur, he was a Google engineer looking for a new challenge. Someone asked Bill Corin, who was a VP at the time, what other secret projects do we have? He said, Well we've got self driven cars. I would like to work on that.

The first set of tests for chauffeur to pass was to drive ten 100 mile routes without the safety driver having to take control of the car. But these weren't just any old routes. That we're just things that Larry and Sergei thought would be difficult. And so a lot of them were the worst possible ways to drive between Mountain View and San Francisco. So one is to just drive up El Camino. There were I think a hundred and forty traffic light intersections you had to go through.

One was to do every possible bridge in a zigzag pattern. Like no one would actually drive to San Francisco this way, but they do demonstrate a lot of difficulty. And in those early days, the chauffeur team tried a bunch of different things. We did a brief experiment with golf carts to see if we could make this cost, you know, a hundred X less.

And so the next phase was Firefly, which was the cute bubble looking koala car. This was the first time people were gonna be seeing cars with nobody behind the wheel on public roads.

So we wanted to make it as safe as possible and non threatening visually, right? And so if it was like a giant black SUV, that would have been one thing, but if it's this cute little bubble car, then people would perceive it a different way and Ultimately we did not proceed with that, but it was a good experiment along the way. Then there was a interlude where we worked on freeways. And so this is similar to what you can find in commercial products from other car manufacturers now.

is we're just gonna help you drive on the freeway. So what happened is we gave some Googlers these cars to try them out. We told them you have to pay attention. You have to keep your hands by the steering wheel. You are the backup system, and we're watching you. There are cameras in the car. So the Googlers took our cars out to use for their commutes to work. And what we saw was kind of terrifying, people falling asleep.

Doing their makeup, using two phones at once, all sorts of terrible things. We ended up canceling the trial early because it just wasn't safe. These were people who knew they were being watched, doing exactly what they had been told was an unsafe thing to do. Imagine, we thought, what people would do if there weren't cameras in the car. It was a huge pivot moment for Chauffeur.

It's where we decided, you know what, this hybrid human computer thing, we we don't think that's the path. And so we're gonna aim for what's called level four autonomy, where it's just door to door, you can take a nap, just not pay attention at all, the car will drive. This was the moment the realization came to the chauffeur team that their real job was to transform mobility.

It was not gonna be making your car but smarter. It had to be a complete service, door to door, moving someone from point A to point B. True autonomy.

But it took another important development to make it happen. The revolution in machine learning. Yeah, so today we're using machine learning in basically every part of the system. We are deeply reliant on it. In a lot of ways, we were way too optimistic in 2009. Had the deep learning revolution not come around, I think starting around 2014, there is no way we would have a working service. out in real cities today.

In two thousand nine, we used almost zero machine learning in the system. I think there might have been one place that was machine learned and everything else was a pile of heuristics and handwritten code. It is amazing we got it to work as well as we did.

But I don't think there's any reasonable way that we could have gotten to the service we have today had that, you know, deep learning revolution not happened in the middle. There was so much work that you and your team have done over the last 15 years.

That's actually, even once we conquered the safety problems and we're like, oh, we're done-ish, like we weren't even close to done. At this point, it is very clear that the Waymo driver is better than a human in the same circumstances. Like there's independent verification of this.

that there I think eighty two percent fewer airbag deployment accidents if a Waymo is driving the exact same style of driving that a human would be doing. So that's great. Does that mean we have solved every problem? No. Pickups and drop offs would be a great example. And so there's a lot of work that goes into doing the right thing. We don't want to block traffic. We need to figure out where the passenger is. There's a lot of subtlety to this. We don't want to drop someone off next to a wall.

We could ignore these problems when it was just a handful of rides, but at this point we're doing over a hundred and fifty thousand rides per week. So these things are happening on an hourly basis and we need to pay a lot of attention to them. We put a lot of effort into user research about how people feel about this.

I've taken a lot of people on their first rides in Waymo's and pretty much the universal experience is the first five minutes is wonder. They're just amazed that technology can do this. They're looking at the screen. Oh, it saw that bicycle over there. And then after five minutes they're checking their phone, which is exactly what you want, right? This should not be exciting. It gets boring quickly. Yes, which is the goal, right? Which should be boring, right?

Over time, we've also had to get really creative about finding the right safe ways to test the car in unusual circumstances. Yeah, that's right. And and in the early days when the car barely works, you can just drive for five minutes and find plenty of things to work on. But as you approach the point where you are probably better than a human, you mostly don't see anything that interesting, right? Even driving, we're driving a million miles a week, driverless.

And even that doesn't give us everything we'd want to see. Uh and so we do some structured testing. We'll take a closed course and roll out a baby carriage or set a fire or blow snow at the car. Uh and that's very helpful, but the majority of this we're doing in simulation, right? And it really helps that you can make synthetic versions of a lot of these things. You know, one of the big things we're thinking about is what happens if a car spins out.

Not not us, but some other car, and how do we react to that? And it's obviously not something we want to go out to the freeway and simulate and just test, right? We make synthetic versions of this. The more we can rely on that, one, it's cheaper, it's also safer, right? And and we can generate exactly what we want. So we're leaning very heavily into simulation and synthetic data at this point.

Can you give us one or two examples of kind of core users or groups of people that were getting something more dramatically positive than you saw coming? It's a very broad category of people, but uh we've heard a lot of positive feedback from women who take rides alone and feel much more comfortable in a car that is driven by a computer, seen really positive feedback from people who are blind, um or partially blind and can't drive, and this is completely changing their lives.

We got into this trying to build something safe and save lives and, you know, transform mobility. And along the way. There have been other things that people call out. The car is a very consistent experience. It is a nice car. It's a Jaguar. We keep them very clean. Uh and and people just like knowing what they get. You were saying right at the beginning. that it wasn't clear to you or anyone on the team if this was gonna work at all. Was there one moment which was the most like

Oh yeah, this is really gonna happen kind of moment for you? There was. And for me, it was the first time I was riding in the backseat of a driverless car in San Francisco, and we went around a double park UPS. And in my mind that had always been the Rubicon. I was like, I don't know if we're ever gonna go around double park UBS trucks and we just did it and it was no big deal. And then in the next block we did it again. And I was like, Okay.

This is a product now. This is not a science project. This is a real thing. And I'm gonna be taking this for the rest of my life. Where do you think this is going in the longer run? Yeah, uh this question is near and dear to my heart. I'm very into urbanism and all forms of transportation. I take the train all the time, right? I'm not sold on just autonomous vehicles.

But I do think they're the future. And right now we're deploying them as a ride hailing service. And it is mostly one or two people taking a ride from point A to point B. But that's not all it needs to be in the end, right? I do think there are things we can do in the very long run that will be just transformative for how we develop cities and how we live our lives.

And if you look at any major city in the United States other than maybe New York and San Francisco, a very good fraction of the city is devoted to parking lots. Right. And a very large fraction of the streets are devoted to street parking. It doesn't have to be that way, right? And having cars that can drive themselves and that move between different passengers

And are used 10 times more efficiently than a personally owned vehicle. That is magical for how we build cities. It's going to take a while before we get there, but I am very excited about the promise.

So look at it this way. Over 15 years, Waymo has gone from straight-up science fiction, something almost no one believed could be built, to a widely used service out there in the real world. But for that to happen, for that dream to become a reality, the team working on it needed a home, a place to incubate. Somewhere that sat just outside of Google, away from the day-to-day needs of running a business. Somewhere a bit like Bell Labs.

the innovation lab that back in the 40s came up with the transistor and basically revolutionized communication. Projects like this need a home designed to nurture far-out ideas that could change the world. And that home became X, the Moonshot Factory. The founders of Google, Larry and Serge, who I admire indefinitely, really believed that at the edge of technology society is underinvesting and that great new businesses could be built.

Self-driving cars being just one of them. Um so they asked me, Sebastian, can we build more businesses at the age of technology where we have an insight that something might actually work? It's not working yet. And then they asked me Sebastian, would he be interested in being involved, right, as running it. And they'd been looking for many years for the right person. I'm not sure I was the right person, but I raised my arm and said, uh give it a try and uh the rest is history.

And then you came in December of two thousand nine and asked if I would set it up with you. I was so delighted that you came. and became a partner in this. And we I think we have equal partners from the get go. I appreciate you saying we were partners. I reported to you and I remember That you only gave me one really concrete piece of advice, which you gave me on my very first day working at X. We hadn't even named it yet.

Which was ignore your manager which was you. And invoked it well. It did work out well. It was a good piece of advice. Tell me a a little bit about what you hoped that this place could be or what you hoped it wouldn't turn into. So I I studied labs in preparation of my conversation with the founders. And the one that appealed the most to me was Lockheed Mark and Skunkworks, where small groups of people were able to go out into this like smelly building. That's why it's called Skunkworks.

and work on an idea without being distracted by management, by finance, by reporting and all that kind of stuff. I wanted clear focus on let's solve a really hard problem that we can turn into a business. I mean obviously in the early days we had no finance people on board. We had a moratorium on PowerPoints, no reporting. We uh were ultra secret. Google didn't know we existed except for a a small number of people.

So we are really operating in complete secrecy. For the teams that I hired and and team leaders, I would tell them, Look, you might be inclined to hire many, many people, but managing many people is a lot of work. Don't do that, okay? Hire yourself few people you really trust. So you can work with them in a much more focused way. And that worked really well. So the Google Glass team for a long time was like five, six people.

Same for G Cam, the the team that did all the computational photography inside Android camera system. Same with like Vamo. The the initial team was about ten people for a year and a half. Yeah. And yes, we could have hired a hundred people, but the saying was if you had a hundred people you couldn't accomplish this. You need ten people. And there was an insight that I took from Skunkworks and other places that the smaller the team, the faster you move.

as like the father of self driving cars. It must be incredibly Prideful for you, like just like a father, to see them out on the road. It's kind of amazing. I if I open my uh door in my house. There's at least two or three self-driving cars in front of me every morning. It's amazing to see how transportation is changing so rapidly right now.

As Waymo's wheels were turning, revolutionizing how people move, we were busy dreaming about something else too, rethinking how goods move. What if the fastest, cleanest, and most efficient way to deliver what you need wasn't by road at all, but above it? Because here's the reality. We move unbelievable quantities of stuff by road, and more every year. Which means more traffic, more noise, less safety, more pollution, more expense.

Trucks clogging up cities just to deliver a single item. There had to be a better way. A lot of these early projects go back to a dinner that I had with Larry Page. And and Larry would just sit down with me in a spreadsheet and say, Look, I I worked it all out. Here's the cost of batteries and the the the drag of a vehicle and the straight line physics and so on. And you can bring a sandwich to people for half the cost in a motorcycle.

And I would sit there and say, I just wanna have dinner and Larry, can we have can we talk about art? Can we talk about music? But it Larry would always focus it back on unit economics of of these new technologies. And Wing is no exception. Wing was born with Larry and Sergei believing that transport should be airborne. Airborne, think about it. Think of all the times you've needed something in a pinch.

Imagine how differently you would shop if you could have any item you need as long as it could fit in a shoebox in just a few minutes. You wouldn't stockpile milk just in case. No more batteries slowly discharging in your drawer, medicines expiring. You'd buy what you need when you need it, without filling cities with even more vehicles.

And that's where Wing comes in. This moonshot set out to fix the last mile problem with an autonomous aerial delivery system. Delivery by small autonomous drones. But moonshots, by definition, are never smooth sailing. So when I started at Wing, we were not allowed to fly commercial deliveries anywhere in the world. Margaret Stuart Nagel is head of policy, regulatory, and community affairs for Wing.

a public policy expert who specializes in making things happen. We had to build this very small, highly reliable unobtrusive tiny aircraft. We had to create a regulatory framework to make that business possible. And then we had to build public trust. So all of those things had to come together. to be a successful business, a way to deliver packages directly to people's homes quickly with minimal impact on roads, do this safely and reduce emissions as we deliver these packages quickly.

We'll come back to Margaret a bit later, because before you ask regulators to approve drone delivery, you've got to get your system working first. When the idea for Wing was taking shape in 2012, drones that captured images and video were on the cusp of entering the mass market. But getting a drone to pick up a package, fly safely to its destination, and deliver that package to a customer without needing a human to operate it, that was many years and many technical hurdles away.

I've always tinkered and built stuff. I want you to meet one of the most creative engineers I've ever worked with. I would say like ten, twelve years old. I have memories of me just being in the shop of my grandpa. He was a carpenter, and I grew up in Germany. The young Andre Prager had an inventor's imagination. I was into skateboarding and found a large DC electric motor and I found a way to attach it to the skateboard.

It was powered by a car battery that was like taped to the center between my feet. And this was probably 91 90 something like that? I might have invented the first electro skateboard. Now we can't vouch for that. There's another claim on that title. But Andre certainly had never seen an electric skateboard anywhere before, and he got a taste for zipping along at speed.

This thing went like twenty-five miles an hour. I I still have the scar on my knee, so I Could you break or you just had to jump off? And it wasn't just making things to zip along the sidewalks, but also in the sky. He was an avid builder of remote controlled planes. eight or nine when I started building my first R C planes. My dad was a flight instructor in Germany and so on. So always like a big aviation connection.

Andre spent a lot of his childhood down at the local airfield pouring over the designs of aircraft. flying the remote control planes he'd been tinkering with, and going up in the air with his dad. As he grew up, he trained as an engineer, but his passion for planes never faded. And in two thousand and fourteen he saw a video from X pitching a bizarre idea, delivery by drone. And I said, This is so cool. Like this is so crazy, right? Like drone delivery? Like

I think that could actually work. I looked at the drones and the video and said like some of the stuff I'm I'm using on my RC planes, right? Andre was itching to get involved in the project, but despite all that experience building little planes, he didn't really believe he could do it. It was a friend who talked him into sending in his resume.

I never saw myself as like the kind of engineer or kind of person that would work like in Silicon Valley. Right. Like I on on my own, I probably uh wouldn't have done it to be honest. It's so interesting how you hold yourself back. Right? He got the job and went on to become Wing's chief engineer. But when he first rolled up at X, he was pretty nervous.

And had big time imposter syndrome, right? Like I sat in the meetings and everybody was talking and what they were working on and I'm like, oh my gosh, it will be like sixty seconds before they figure out they made a mistake. Hiring me you, right? But it was just like almost overwhelmingly creative this space. It was that creative energy that lit Andre up. For me, it's more art than engineering, to be honest. I actually read a very interesting book.

a while ago, Visual Thinking was the title. And I suck at math. Like I just barely made it through engineering school because of math. And she talks in that book about how math is used to test engineer applicants, but only for one reason, because it's so easy to measure, because it's so predictable.

Whereas the creativity behind engineering is very hard to measure. Like how do you test that? Right, because if I was interviewing you, I would want all those pictures of the things that you ripped apart and rebuilt when you were a kid. That would tell me way more than the math scores. Exactly. We need all kinds of engineers, right? Like there's so many different types of engineers.

And we need all of them. To make a moonshot, we like to say, lovingly, of course, that you need a diverse bunch of weirdos. Otherwise, how can you find really surprising and non-obvious solutions to a really difficult problem?

Andre got to work. I entered the door, went automatically into flow state and stayed there for eight hours or twelve and then came home and couldn't wait to go back in the next morning. But you had to figure out The fundamentals of drone delivery or and the flying part isn't so much the moonshot. The interesting part is everything else. How do I get the payload onto the drone and off the drone? Right? How do I charge the drone? What kind of like

traffic management do you do if you have a bunch of drones in the air at the same time? Where do the drones sleep at night? And you also had to solve all of this with a real eye towards cost. Because we could have made very complicated, very expensive systems that would have looked awesome in demos.

But when we want to be doing a billion flights a year, the cost would have just never gotten down to the place where everyone in the world could afford it. That's very, very important. It's the simplicity part. Like the less stuff you have. Andre started where we always started X, by prototyping. Our philosophy is to do rapid prototyping, to quickly and scrappily figure out the answer to whatever question you're grappling with.

to get to the next step of your moonshot. Forget the saying measure twice, cut once. If you don't really understand the problem yet, let alone the solution, why measure? Just start cutting. And all this happens in a special place at the heart of our moonshot factory. We call it the design kitchen. My name is Joe Sargent. I lead the design kitchen at X, which is a team and a place completely focused on prototyping, aka bringing crazy ideas into the real world.

The design kitchen is a combination of a studio, a shop, and a laboratory. I wish I had a better name for it, but it's really all of those things smushed together. It is also, I would argue, controlled chaos and looks a little bit like a playground in a good and bad way. Picture a brightly lit warehouse. Stocked with every tool or material you can think of and humming with activity. So we have everything from chemicals and biological materials to bolts, nuts, plywood, paint, mills, 3D printers.

Water jets, test chambers. Virtually every project at X has been formed in here in some way or another. The first prototype of a wing drone was about five feet wide, and it was designed to take off vertically. flop forward to fly horizontally and then to turn back into the vertical position to land on its tail to pick up or drop off a package.

This is one of the first units. This is a tail sitter, because it does exactly what it says. It sits on its tail. They designed this very logically to think. you would fly and you would come down and land, someone would come and they would retrieve what they need. Makes sense. But we went out in the real world with this and what happened, not surprisingly, is the thing would go land and it would kick up dust.

And it would scare animals and it would concern people. So we learned very quickly that that's not gonna work. This is what prototyping is all about. Trying and failing. And most importantly, learning and iterating quickly onto the next prototype. Not a tail sitter. It is this system and this is going to hover.

And it stays above and it drops the delivery. Then you start learning that what's super critical for this is. The hook that holds that package and drops it off. If that's not perfect. It will not work. Ah, the hook that holds the package on a string beneath the drone. One of the best examples X has ever had of something that looks easy and obvious afterwards. and that took a ton of creativity and experimentation to discover.

We have a saying here at X. If you want to put a monkey reciting Shakespeare on top of a 10-foot pedestal, don't start by building the pedestal. Start with the riskiest part of the problem first. Teaching the monkey to recite Shakespeare. In Wing's case, how to receive, carry, and deliver the package was one of the toughest technical monkeys the team had to tackle. Could a cleverly designed hook on a string work?

So it needs to hit the ground and fall off. And it needs to not come back and pick it back up if it swings just the right way. But it needs to be the minimal weight, but it needs to be heavy enough so it wants to drop. and it needs heft so that it doesn't flip back over if it gets a bounce. Everything about this was just obsessively thought about because it it can't screw up. It was a design challenge that Andre and his team spent months on, 3D printing dozens of prototypes.

When you see the end result, you think like, Oh yeah, I mean that's how you do it. And it probably took you two hours. No, this took two years. to figure out. And about like ninety-ish prototypes of that hook alone. When you got it right, how did you know? What what were the characteristics that made it so great? Because it was completely passive. There is no electronics involved. There is no uh complicated mechanics involved. It's like this hook has no moving parts.

It looks like a bobber. It looks like something you go fishing with. It About the size of a a ping pong ball. It has a hook, but it's almost like a mouth. That is a very specific angle and a very specific depth. And the wing package is built for this. Because it's not just the hook that had to be perfect. The packages themselves needed to be custom designed. The handle of the wing package fits very neatly into the mouth of this hook.

It's a simple, neat solution that was two years in the making. The winch system itself is also very simple. It's basically a motor with a spool. And the motor is able to sense basically force. We can we can sense how strong someone is pulling, we can sense when the package has touched down, when the hook uh caught or or stuck. Our approach there is keep the hardware really, really simple and then you can add more and more intelligence later with software updates.

This was the design philosophy central to Wing, keeping things simple, resilient, and affordable, so you can really scale up to offering a delivery service to everyone. And for that scale to happen, it can't be humans who have to load the packages onto the drone each time. I think this is one of these examples where

how can I distill this down to the smallest denominator, right? Like the absolute minimum of infrastructure. And I'm like, what if that tether could just like go through something which reorients the the hook into the right position and then Just grab the box. The autoloader looks a bit like a couple of umbrella spokes.

It's two thin metal arms protruding out of a stand. This thing has no moving parts. It has two arms that grab the tether, the hook gets caught, goes through a little channel, And then uh get attached to the tab of the package just by basically friction. And then the drone can pick up any time by itself. It's so elegant and simple. That's the beauty of this approach. If you think about something really complicated.

Because it's so complicated and expensive, you have to limit the amount of them, right? If this breaks down because a sensor failed or something like this, then you're down until the technician comes and fixes it. In in our case you y y you can either like have five of them from the get go, or you just grab another one out of the shed.

And swap it out. And because there's no electronics, no power, no moving parts, it's not gonna break either. No, I think the failure case that we were most concerned about were like birds like making a nest in there. Don't worry, that has never happened. So, like Andre said, the key to this moonshot was simple hardware and sophisticated software, and artificial intelligence became the secret sauce that empowered the drones to really know what they were doing.

The drones are doing more and more learning by themselves. In the early days when you were placing what we call a nest, this is our airport, right? we had to take a a a precision GPS like uh surveying uh tool and tell the drone exactly where these pads were and then program that into them and so on and so on. By now it's like the system just does it on its own. Like every month they're a little bit smarter. One time we land it on a power line, a month later, it won't do that anymore.

The software includes an unmanned traffic management system, which plans and reserves flight routes. Think of them like tubes of airspace that follow where the drones are or will be. And the system shares those reserved airspace tubes with other drone operators and aircraft to ensure they don't fly into each other. This was a crucial element to gradually gaining the confidence of the US aviation regulator.

Remember what Margaret said at the beginning, that at the start of this moonshot, not a single commercial delivery drone flight was permitted. We had to work with regulators to build a process. to approve each individual flight. Each individual flight. It's enough to put a lot of people off even trying. But if there are two strengths that Margaret and the wing team have in abundance, it's patience and tenacity.

While Andre and his team of engineers were obsessing over the technology side, she was working with regulators to carve a path for this moonshot to take off. We have conducted over 450,000 commercial deliveries safely to people's homes. Um, but we've done even more thousands and thousands and thousands of cycles of testing uh at our test site. We've moved from literally not being able to offer a drone delivery business because there was no path to approval.

to having that pathway not just in the United States and Australia, but also Europe, where we can move city by city. Wing's first test deliveries to the public happened in rural Australia. Gradually expanding to more and more urban areas. and with each new location, building trust with the public was a vital part of the journey. We take our drone.

into a local shopping center to the farmers market on the weekend. And we bring our pilots and our engineers and talk to the community. We answer their questions about how does it fly safely from your hub to my home. How loud is it gonna be? Am I gonna notice it? We assumed nobody wants the irritating buzzing of a drone zooming over their house.

Understandably, noise was one of the main concerns for the public as we did our early testing in Australia. But the root of the problem didn't turn out to be quite what the team expected. Let's go back to Andre Prager, former chief engineer at Wing. This is a wonderful example of like what you think is gonna happen and what the reality is, right? Wing drones have two types of propeller. Two propellers to fly forward and ten propellers to hover up and down towards the destination.

And at the beginning, there were a few complaints about their sound. We were thinking the hovering, the up and down in the delivery will be the thing that's most important, right? So this is what we need to make really, really quiet. We develop propellers that are turning much slower and they have blades that are designed to be a different length for each pair. So it's four blades.

And one pair is longer and the other one is a bit shorter, which spreads that frequency band out and makes the sound more pleasant to the ear. They fitted the new, much quieter propellers to the drone. Interestingly, the noise complaint like percentages didn't change. And we're like, I mean we're like half as loud now, like why why doesn't that work? And then we looked closer into it. never came from the people that got the delivery.

And not even the neighbors, because they were all like excited, like looking what what's going on. We looked into the data and so wait, it's the people we fly over. So it's at like two hundred feet. in the air, moving at sixty or seventy miles an hour, it's only going to be over their house for a second or two. Yeah, but but the propellers that we had on the drones back then, they had like the

It sounded a bit like a Formula One car. It wasn't loud, but it had this nice you know the sound when it came by, and people noticed that. Higher in the air, the cone of sound is wider, so it reaches more people. Just the sheer amount of ears. Is is really really high. Versus when you are low near the ground, you have like, you know, ten pairs. When you are up there, you have a couple thousand that you fly over. So someone will complain about it.

So that was a big epiphany. Andre's team applied the same changes to the flight propellers as they had to the hover propellers. changing their length to spread the sound frequency and reducing their rotational speed. The effect was a dramatic sound reduction. And literally like there weren't any complaints anymore. If you are in a normal conversation on the street and you talk, you will not look up

Another worry some had was that you'd see loads of drones zipping around, spoiling the view of the sky. I was so excited to go there and see what's going on there, and I also wanted to get a feel You know, how how does that feel when the sky is full of drones? I went there and I drove around in Brisbane. I knew how many drones were out there. I didn't see a single drone.

You have no idea how big the sky is. Like I literally like a moron, I was like in the car, like like looking like this the whole time, and I was out there an hour. I didn't see a single drone. And when I saw something, then it turned out to be a bird. You know, like so so even thousand flights a day from a single nest, you're hard pressed to even see them.

This made me so happy because the last thing I really wanted with this whole endeavor is to create some intrusive technology. I think you really get a net benefit, even from a noise perspective. compared to cars. And so this this picture of like darkened skies and whatnot, no. They they they could have done ten thousand deliveries from that nest. You still wouldn't see us finger troll in the city. Today, Wing is flying in every sense.

It graduated from X in 2018 to become its own company. So we're operating today across three continents and we're in cities like Melbourne and Australia delivering DoorDash to customers there. We're in Dallas, Fort Worth, where we're partnered with with Walmart as well as Doordash.

When the aircraft arrives at the customer's home, it hovers down to about twenty, twenty-five feet, gently lowers the package to the ground, and then flies back to the nest where it recharges and is ready for the next delivery. And the customer can just run out and pick up their package. I think our fastest delivery is about three minutes, two and a half minutes. And typically we're filling these orders in twenty, twenty-five minutes. And it's not just for grocery items that you've forgotten.

an operation in partnership with API and the NHS. The National Health Service in the UK. We are delivering medical samples between two hospitals. And we're excited to be able to offer this service in truly the middle of London, between the shard and the eye. We're delivering these medical samples uh and helping get people out of the hospital more quickly.

These two London hospitals are only a couple of miles apart, but in traffic it can take more than half an hour by road to move anything between them. By drone it takes two minutes. Wing drones are flying urgent blood samples from one hospital to the other, where they get analyzed, so that doctors can determine more quickly whether a patient is safe to go for surgery or to go home. And this is just the beginning. One of the things that I'm excited about is the potential to not just

care for patients in the hospital, but also when they go home, making sure that the patients have their medicine when they get home if they need a prescription delivered. And during the aftermath of Hurricane Helene, Walmart asked us if we could be part of their recovery efforts in North Carolina. And within just a few short days, we were able to stand up an operation where we were able to deliver from a Walmart to a Red Cross shelter.

And one of the first deliveries we did there was to a mother who was looking for an allergen-free supplement for her for her baby. And we were able to deliver that formula to her in minutes. The really exciting part I think is that we're hitting that point where the rollout and where people's access to this is going to grow.

very quickly, you can expect that over the coming years, more and more people are going to be able to experience drone delivery. Margaret has been advocating for Wing's mission since the early days. And now all that careful work is coming to fruition.

After all this time, what has she learned about what it takes to make a moonshot like this? Knowing where you want the endpoint to be, right? What's the problem you're trying to solve and what you're trying to accomplish? I think is really important. You can't get too stuck in any one moment. You have to kind of keep stepping back and seeing the big picture and and that endpoint. And sometimes you go sideways and then you take a big leap forward.

What about my old friend and ex founder Sebastian Thrun? Well, he believes the possibilities for moonshots are limitless. It's really important to realize how society is progressing today through technology. Societies or people are about 300,000 years old. And almost anything we care about today, from electric light switches to flushing toilets.

is really not older than 150 years. Cars, cell phones and so on. General anesthesia, like stuff that we take for granted today. That to me means that probably less than one percent of interesting things haven't been invented yet. And ninety-nine percent of someone else to invent. Okay. And if you don't believe it, just look at the last three or four years with AI and social networks and all that stuff. The world is massively changing. So pick something that you believe in that can be done.

Technologically, pretty much everything can be done. Uh almost nothing cannot be done. I love it the way you just said that. And as you just pointed out, almost anything that that could be made can be made. But not everything that can be made should be made. Any thought as we sort of rush into the future about the obligation of technologists to sort of help bend the world towards the positive. Like up to what point do the technologists have some sort of

ethical or moral responsibility to make sure that the things they're creating are actually gonna be good for humanity. It's very, very simple. First of all everything that can be made will be made. Whether you say it should or should not be made is immaterial because it will be made. Our job is not to stop making things. Our stop is to prevent the misuse of things that we make. Overall technology has been absolutely amazing. It has been a democratizing force in the world in every aspect.

in terms of longevity, health, mobility, education Uh whatever. Technology is a great force for the better, but every single person, including technologists have a responsibility. It's uh not just a technology question, it's a societal question. We have to involve politicians and and and and and lots of people from society because it transforms all of us and we jointly have to be transformed. But having said this, I'm an optimist.

Like just the technologies in my lifetime, street view, self-driving cars or stuff I didn't do myself, like I don't know, smartphones, have been so transformational that I'm super optimistic. Pick a problem you care about, like your grandmother cares about, like in my case the death of my friend Harold. And then ask yourself, can you envision a technology that can solve this problem?

And then work on it. In all likelihood it can be solved. In all likelihood the problem is not that we don't have technology. The likelihood is we don't try hard enough to solve these problems.

I love Sebastian's optimism. The way I see it, not everything can be made now. I love coming to work at the Moonshot Factory every day to discover as quickly and efficiently as possible what can be made. This is why we do what we do and this is why I want to share it with you. You've been listening to the Moonshot Podcast, a Blanchard House production for X the Moonshot Factory, with me, Astro Teller. And let me tell you, we're just getting started.

because we're going to be pulling back the curtain on our wildest ideas, biggest breakthroughs, and yes, some of the biggest failures too, because that's how moonshots get made. Okay, you know the drill. If you're enjoying the podcast, hit subscribe, drop us a rating, and leave a review wherever you're listening. It'll help more people find us. And if you want to see these moonshots in action, Check out our YouTube channel.

That's X the Moonshot Factory on YouTube. Stay with us, because we have so many incredible things to share with you over the course of this series. And in the next episode, we're tackling one of our most ambitious challenges yet, how to make the web truly worldwide. There's billions of people in the world that don't have access to the internet and we want to change that. The first time I heard about the possibility of transmitting extremely high speed data, literally over thin air,

My jaw just dropped. Everybody thought it was totally impossible. And then we did it. It's almost like you have fiber optics but in the sky. I'm really looking forward to watching you and Tara change the world. We're just getting started.