I Tested Tesla Robotaxi & Waymo So You Don’t Have To — Uber’s In Trouble
39:30 Share

Alright, this is day 3 of the VrboTaxi launch. Alright, we'll get into this VrboTaxi.

Alright, so before getting into this Robotaxi, I discovered in the app you can actually control a few things before getting in. So you tap a button and the Robotaxi will whistle. So it helps you find where the Robotaxi is. You tap another button and it will flash the headlights. You can find your Robotaxi and then there's

There's another button to open the trunk and another button to lock and unlock the car. I'm guessing probably some of those things will adapt and change over time. And yeah, you might have it where over time, like you have to unlock the car from the app or it unlocks it automatically as you get near something so that strangers, other people don't just like sabotage it.

take over or take your car you know before you can get it. So one of the questions people have is exactly what does the Tesla safety monitor do inside the car? I mean what are the controls available to him? We know at least on the front screen you've got a few buttons here. One says pull over, one says like stop in lane and then call support.

but the question is like are they able to do other things besides those buttons some people have speculated that there's something like on the right handle and maybe they pull a trigger or something maybe that stops the car but i don't know i mean i don't know if just breaking the car would be you know enough of a kind of a safety provision or if you know

that alerts like some type of extra monitoring from support or yeah it's unclear to me exactly what the Tesla monitor has access to and Tesla isn't sharing much and the Tesla monitors aren't allowed to speak regarding any questions so I think in the coming days we'll find out my guess is probably Elon or someone from the Tesla AIT might share some more details

But yeah, it's gonna be interesting. Alright, we're almost at our destination here. This is our first destination of the day. Let's see where RoboTaxi drops me off. Okay. Alright, so my second ride of the day is gonna be a Waymo. Let's check it out.

Heading to 1701 South Congress Avenue. Please make sure your seatbelt is fastened. For any questions, press the call support button to speak with a rider support agent.

Alright, so yeah, I wanted to just mix up kind of the rides today. Tessa Robotaxi, Waymo, back and forth. We'll see how I feel. But yeah, yesterday I took my first Waymo. This is a second Waymo, so I feel much more comfortable in it, much more used to it.

experience Waymo and Tesla the experience is very similar you know you just kind of hail the car and just get in the car and sit back you know and both are very comfortable experiences so far I find that the Waymo

I mean, I need to ride it more, but I feel like sometimes it speeds a little bit faster in some situations than a human would. There's some little hesitation, but I mean, it's impressive though. You know, I mean, I don't want to discount it because it rides well, very well. So yeah, I mean, it is impressive what Waymo has been able to do with

their sensor suite and compute and their approach. I would liken kind of, yeah right now it's just kind of going a little too fast for my comfort, you know? It was going up a hill but it didn't really need to go that fast. It was just like five miles per hour too fast. And plus look, wait, wait, Waymo, Waymo, there's like all these cones. There was a pedestrian, a human would go five or ten miles slower.

than that yeah it just kind of goes a little too fast for my liking i mean the tesla robotaxi in my opinion would would have treated that much more cautiously and much more human-like all right so one question people have is like about teleoperation where is waymo taking me to okay yeah now yeah okay there's a small road that's taking me on i'll show you what road this is okay i guess it's um

is it a shortcut or something um interesting why i took this road yeah we could have just took a right but it did this interesting shortcut in a way yeah do you see it kind of yeah bypassed the right turn at the signal light so the question is some people are asking like so is Waymo teleoperated um no it's not teleoperated meaning Waymo is driving itself in the car um

Some people ask, "Can Waymo be tele-operated in difficult situations?" And the answer is actually not really.

really ever tele-operated in that sense where remotely another person is completely driving the car Waymo came out with a blog post last year describing their what they call remote assistance system it's basically because of latency issues and challenges like for example if you're in a place with spotty internet

It's too risky to completely take over for a remote driver. So what Raymo does is they will obviously send all the video feeds that they have as real-time as possible and then they'll get feedback from a human in difficult situations. So basically the car will ping a human and the human will then give feedback

suggestions on where to go, what to do, or what the situation is. And the car will use that feedback, not necessarily take it as a command because maybe the situation has already changed a half a second later. So it'll just take that into consideration to make its decisions. So that's kind of, you know, I'm sure we might have a few other options, like for example, could force it to pull over or do some other things. But generally speaking, yeah, the

what Waymo calls the Waymo driver is the car on board or the AI on board and that's what's making the decisions on driving and when it needs help

Ping remote assistance a human will give feedback and I would imagine probably Tesla's approach is similar right now I mean if Tesla wanted to do real-time Teleoperation they would need some big or some really, you know interesting Starlink gig on the on the on the car, right? They would have to have basically

internet all the time but then what if you're under like a bridge or what if you're like under trees and the internet you know is spotty and you can't really control the car remotely there's just too many question marks to successfully remove

remotely tele-operate a car. So I would imagine Tesla is doing something similar where they have different features, they can give feedback to the car, they have different functions they can operate, like let's say just pull over, whatever, but the car is making decisions. Alright, Waymo's gonna drop me off here. Alright, this will be the second robotaxi of the day. Alright, so...

I'm gonna turn this. Makes it a little quieter if I turn up the temperature. I'm gonna do actually 72. So I left my mic that connects to my phone in my car, my regional destination, so I need to go back. So it's about a 13 minute drive back.

And I'll get that and maybe continue with some more drives. But yeah, if you look at this compared to the Waymo, it's so much more cleaner, sleeker, modern interface. I mean, the Waymo has an edge right now in the sense that they don't have the driver or the monitor. So it's a more comfortable experience where you don't have anyone else in the car. But I give Tesla the edge in terms of its...

it's driving is just much more human-like you know it just slows down in the right spots doesn't you know go overly too fast i think tesla's really made a lot of strides in the past six months or so um it feels like with tesla's fsc every six months it's just you know a big jump up um and i think it's just gonna be like that for you know the next um

one to two years. We're going to see just a big, big jump. I think by the end of this year, you see another big jump and then you're going to see, yeah, probably next year we're going to be, I think people are going to be very, very

shocked and surprised to see how good FSD becomes. Also by next year sometime probably I don't know exactly when but we'll have hardware 5, we'll have much bigger neural nets, Tesla has much more data and this is going to be a completely different performance by Tesla's FSD. So I think that's something not to discount is the continual improvement Tesla is seeing. Now I've been

You know writing driving these FSD beta builds since like the early days I think it was version 9 or so Yeah, I was just talking or texting James down man He was like reflecting back on when I took him on a ride with the version I think v9 or something back in the day when I first got FSD beta City streets, but yeah Tesla has come a long way every again six months. It just takes a huge leap and

forward and it's quite impressive what Tesla has been able to do. See that was a pretty impressive move right there. Just went across the street through traffic. Yeah the way it handles these streets too, it just feels more human. Like a Waymo would just kind of go a little too fast through these streets. It just is not, I would say a little too robotic or just too like

I don't know, I think comparing Waymo to Tesla right after each other, you can see Tesla is much more nuanced. It has a better feel of driving like a human. Alright, here we are. This is the third Robotaxi of the day.

Alright, so I have my microphone back on my phone now so I got some better quality here. So yeah, I was just looking back at my old emails and reflecting back on Tesla's FSD program and it's been three years or four years actually. It was July 2021 when I got V9 version 9 of FSD Beta City Streets and so

It's been a quite a long journey testing Tesla's FSD beta builds since then and I tweeted or I posted on exit every six months or so I I've noticed like a big performance boost like a huge job in performance and I think There's several factors one is you know, Tesla is a

improving their neural nets and the data they input into those neural nets. So for example, the amount of data, the quality of data, all of that is rapidly increasing. And then the size of the neural nets is rapidly increasing too. So you've just got a lot of just ways to boost the performance of the cars. And I think initially when there's a boost, there's like a lot of things to improve.

to work out a lot of kinks and bugs and you might see two steps forward one step back and then by the time like you know they they spend three or four months on that build it gets really good and then they you know they're working on the next huge performance boost which is you know just a bunch of

data improvements, the quality, the quantity, the type of the neural nets, the size of the neural nets, all of that, the methods that they use. And they're constantly tweaking how much they can fit on the FSD computers, how large neural nets they can fit on the inference side. And so you've noticed this pattern over the years, right?

Yeah, every six months or so, there's been like big boosts in performance. I think we're going to continue to see that. I don't see that slowing down. And for Tesla to already have gotten to this place where, you know, the FSD can pretty much drive around by itself. And we haven't seen, you know, a lot of bad drives since RoboTaxi launched. And we've had probably like hundreds of drives recorded. And I think this is a huge bullish trend

I think sign here because in the next six months, you're going to see Tesla's FSD just be radically that much better. And again, in another six months, another six months. And that's pretty much the scaling laws of AI where you throw more compute, more data, better data, all that stuff and bigger neural nets and you get better performance. And that's what Tesla has been

able to follow, especially recently when they went end to end. And so, yeah, I think this is just the beginning. You know, it's very exciting to see what Tesla's doing here. What we're seeing here, Tesla driving by itself through city streets in Austin, right, without a driver in the seat. This is like the early days. Within six to 12 months, I think Tesla's system is going to be radically better. And here's the thing, it's getting to the point where

Critical safety interventions are so rare that, you know, how do we really know that it's better? Right. And I think it's just going to be through thousands of miles, right, of driving that we're going to see. It's just interventions are super, super rare and they get more rare over time. So this is important because this also will determine how fast Tesla can scale. And I think if they're ready at this place where they're able to roll out in a

in a limited geo-fenced area already. This means in six months and in 12 months and 18 months, they're gonna be able to probably roll this out in a much bigger area. The question is kind of what is their rollout plans? And I think you've just kind of followed the kind of like the roadmap of the past into the future, meaning every six months, right, the system improves so much that there's a huge jump

And that allows you to do so many more things. And so I think in six months time, we're going to probably, I don't know exactly when it'll happen. It might be within two or three months, but it might be six months. But I would imagine in six months, we get rid of the safety monitor.

And we have expansion into some other cities without the safety monitor. But again, it's like Tesla will be ultra cautious because again, just wait six months and your system is like that much better and that much safer. And so Tesla is not necessarily in this huge rush to rush out something that isn't ready. They're going to put out something that they feel confident in and they're going to be extra cautious about it as they roll it out. But again,

they've got the scaling laws of AI on their side, right? Every six months, things are just gonna get that much better. They've got lots of stuff in their pipeline too, right? Tesla's going a lot into simulation data. We've got the AI5 computer coming up. So there's a big pipeline of improvements in the next couple of years for Tesla's FSD as well. All right, we're approaching my drop-off spot. This is a local park.

Alright, I've called a regular Uber. Wanted to ask the Uber driver what their opinion on Waymo here. Waymo and stuff like that? Yeah, so I'm comparing Waymo, Tesla Robo Taxi, and Uber. So do you see a lot of Waymos around here? Yeah, especially downtown. Like, do they make it harder for you to get rides? Because I know there's so many Waymos, right? I would feel like that, yeah, downtown. Yeah, because especially like these short rides, I feel like...

Over half the time I get a Waymo or something, you know? Really? I don't know. This is probably my second week of Ubering. Okay. I've ridden in Waymores more than I've noticed them. Okay. But yeah, I can see that. They're definitely everywhere I go downtown, one's either two cars in front of me or right in front of me. Yeah. And...

while they are smooth I do see how if someone was a full-time uber in how it could definitely hurt their income yeah I see makes sense all right thanks a lot appreciate it yeah David you have a good day all right I called Waymo

Alright, this is our second Waymo.

of the day i've done two robo taxis and a regular uber so yeah talking with that uber driver he's a new uber driver just like two weeks or so um he didn't really have strong feelings for or against Waymo but i i did have this oh my gosh there was this big stop there wow what was going on there huh i must have seen something but you know it's like going under a little bit of a bridge

Yeah, I think it's a little bit confused right now Waymo doesn't know exactly how to get out of here It's going backwards Okay, so it's gonna go back to where it came from Yeah, interesting There's a little bit of an abrupt stop, but I mean, that's what it is So anyways, yeah, the whole tension is this kind of like the innovators dilemma So the more you bring in Waymo, let's say into Uber app

it's competing against human drivers, right? And so the human drivers will get less and less enthusiastic and optimistic about the future of Uber for humans because it's like the writing is on the wall, right? I mean, they're just, it's obvious Waymos are gonna just increase, et cetera.

human drivers are going to be less in demand. So Uber drivers, are they really going to be excited? No. And then less and less people are going to sign up, right, to be Uber drivers. But that's like the big thing about Uber. They need recruiting. They need new drivers because they have like a lot of drivers always, you know,

quitting or moving on. So they need recruiting for new drivers. And it's not just in the city. It's like to get coverage in like the suburbs and places around outside of the city, they need, yeah, a good amount of people to join the network. But if humans are less optimistic about the future of Uber for humans, less people will drop off. That means their service for

outside the city will decrease. And so you might actually see a deterioration over time of the number of human drivers outside the city for Uber and the quality of the service. Wait times increase, et cetera, cost more.

That might be the cost of doing business with Waymo for Uber. But what do you do if you're Uber? Do you resist and not let Waymo on your network? But you know Waymo is the future, so you have to let Waymo in, but it's going to destroy your human driver network. So you've got this big innovator's dilemma. So, okay, let's play this out. You let Waymo into your Uber network.

your human drivers start to leave your system. Now you're more and more dependent on Waymo. What's in it for Waymo? Why can't they just do their own app like they have in some other cities? Why do they need Uber over time? Let's say they achieve a certain level of penetration. They just create their own app, leave Uber, and then Uber is left with nothing, left with barely any human drivers and barely any autonomous drivers. Now, there might be the case where...

You might think that a lot of other players will have autonomous cars and Uber can facilitate a network of various autonomous players, but I just don't see that. I think autonomous driving is extremely difficult. It can be extremely expensive. And because the data and network play, I think it's a winner-takes-most market. And so I just don't see a ton of players. That's where...

Uber just doesn't have, I think, the brightest future. They're really stuck in a place. I think they probably should have gone all in with autonomous driving early on, which they tried. But, you know, they gave up and sold it off. So, yeah, they're in a rough place. They're kind of in a place where technology is going to

hammer them, but they don't own the technology or have the technology themselves, right? So I just don't see what they have to offer going into the future. Some people might argue, oh, Uber's got the network, right? They got the network effects. They got the app and all the logistics and all this stuff. But no, it's about usage. For example, if Tesla RoboTaxi is able to deliver a cheaper ride at density and shorter wait times, let's say, and

let's say here in Austin, right? Downtown or Austin or this area, people are going to use it. They want better service, right? Cheaper, better service. They're going to use it. Tesla will get the network effects. They will get the usage. Yeah, Uber's advantage only exists when they operate.

have the most riders the shortest wait times and all of that. But if that's not the case, their advantage disappears. I think that's why it's a good move for Tesla not to have tried to buy Lyft or Uber. I think that's a ridiculous suggestion some people have made in the past. There's no need for Tesla to buy Lyft or Uber. They just create a better product that's like

basically, you know multiple times better than the existing product and the usage will come because they're the only ones that can deliver it at that cost and at that quality of service. Yeah, it's gonna be interesting how this all plays out. You're here. Please make sure it's clear before exiting. Alright, I just called another Tesla robotaxi. This is I think number three robotaxi today.

Alright, I'm adjusting the temperature so it's a little bit more quiet in here so you guys can hear me better. Yeah, this will be a short ride. I found this place here in South Austin. It's called Prototype Vintage Design. I have no idea what it is, but the name sounds intriguing. So we'll check it out. Yeah, there's something just refreshing about the Tesla Robotaxi compared to Waymo. If you look at this, check out the clean, modern, sleek, minimalist inside. Waymo just feels like an old...

I don't know, legacy car to me. Inside at least. I can't wait till a cyber cab comes out. I mean that is gonna be nice right without a steering wheel. Okay we've got this car coming up in the lane. Tesla pulls over perfectly so far. Let's see what happens. Yeah it goes on the right. Let's this other car pass by.

and goes around. Yeah, it's like that maneuver was perfect. It's just like what a human would do. No hesitation, no jerkiness. And here is my drop off spot. Oh, we got an interesting situation. Okay. The UPS car is coming very, very close to us. If I was Robotaxi, I would be going back a couple of feet. But I guess the UPS driver is okay with parking like that.

Yeah, decently well done. Yeah, Tesla road taxis off and about. All right, I've called the Waymo again. We'll see how it goes. Heading to Starbucks. Please make sure your seatbelt is fastened. For any questions, press the call support button to speak with a rider support agent. All right, so we've called another Waymo via Uber, right? So in Austin, you can't, there's no separate Waymo app.

you've got to call it through uber but you don't you can't really select a Waymo you just like it just select your destination and then if it matches you within with the Waymo it asks you are you okay with the Waymo and then you say yes and it comes to you so it's um yeah it's a little bit I don't know clunky where you can't really choose if you want like you know Waymo

I guess you could deny you don't want a Waymo, like reject it and then it'll pair you with a human driver. But anyways, we're headed to Starbucks here, pretty close by. Just another chance to ride. I think each ride I'm picking up a little bit, you know, more stuff. Yeah, I've been thinking this is a pretty busy part of Austin. It's not like as busy as downtown, but like South Congress,

you know, this street called South Lamar. And I think the potential is if there's a lot of density of robotaxis, there will be people who just say, forget it, I don't need a car anymore, right? Because if you can get a robotaxi within like three or four minutes all the time, pretty much, then you can avoid parking, avoid, you know, car ownership, all that stuff, and just go wherever you want.

with the RoboTaxi. I think a big part of it is just the wait time and also the cost. If the wait time is high, over 10 minutes, and the cost is significant, then you're like, "Forget it." It's just better having a car. But I think RoboTaxi is the first real option where because the hardware is cheap enough and the solution is so streamlined and efficient with just basically neural nets and cameras,

and we have a chance of really bringing down the cost of rides super low and also the density of robotaxis very high which will decrease the wait time so you've got all of the elements for Tesla to make a pretty big splash in the ride-hailing market I mean Tesla has been working on this for a very very long time and yeah I think they're getting close very very close to be able to see the fruit of their labor

Another question is what will Waymo do if and when Tesla expands their robo taxi rollout and they say offers cheaper rides, will Waymo have the demand that they need to really roll out more cars or yeah, there's a lot of question marks here. Waymo's thing is it just costs a lot, right? Their cars just cost a lot.

Okay we're already here, we are already here at our destination. You're here, please make sure it's clear. Alright here's our last robotaxi of the day. Alrighty we've got what a 13 minute ride back to the parking lot where I parked to start the day. Yeah it's amazing how fast you get used to the robotaxi.

like the first day I rode it, it was like pretty thrilling and exciting and everything, which it is. The second day, you know, it's more calm and chill. And now third day, it's pretty much like commonplace. It's like kind of riding an Uber in a way, just to just get to one place to another. What's been interesting though, is comparing the Tesla Robotaxi with

Waymo and also with an uber a regular uber and thinking about like kind of what the ramifications going forward are and Yeah, it just seems like Tesla has all of the right pieces together. They've got you know, the hardware the software They've got I think the best product potential to scale they're writing the laws the scaling laws of AI They're going fully end-to-end

They just made all the right decisions and I think it's gonna pay off. We're gonna start to see I think Tesla shoot out in front eventually as they're able to scale. Alright if you notice the map, Tesla's not able to go on the freeway right now because that's not part of the geofenced area so it has to go through kind of residential neighborhoods.

take a little bit of a longer route to get to our destination. I wonder if Elon will disclose if the monitor has any extra controls and if so what they are. Yeah it's interesting. Alright guys we're at the end. I'm gonna see if I unbuckle a little too soon what happens. Oh it says pass the seat belts. Okay okay I'll buckle it back. Alright we're almost there. Okay here it is.

All right, so I'm back. That's the end of a pretty full day here. I did three or four robo taxi rides, I think three Waymo rides and one regular Uber ride and had a chance to really think about the differences, the challenges, the pros and cons of each.

Yeah, it's an interesting field going forward, this ride-hailing autonomous taxi field. I think it's lots of questions of who's going to scale, how are they going to scale, how fast, what are the challenges going forward?

And I think, yeah, it's interesting that Tesla has this out and about in Austin. Like I can ride this anytime I want from 6 a.m. till midnight any day of the week. And it's up and running actually. And I think Tesla is going to probably roll out more invitations for this. I think this is a super fascinating kind of rollout for Tesla.

And so far I've had like 13 or 14 rides in the past three days. Every single ride has been perfect basically. I haven't really seen any mistakes. I think that bodes pretty well. And there are questions like I still wonder like with the safety monitor, like what kind of controls or what options they have. And

Yeah, lots of questions like is the RoboTaxi software that you know similar is it version I heard from someone saying this version through 13.3 Which is sounds you know logical, but does it have extra features? I'm sure has you know different features obviously to enable RoboTaxi, but how far more advanced is it compared to what we have in our own cars and Yeah

a lot of interesting insights into Waymo as well, I think at Uber. And I think there's a lot of tension going on, I think in the future for Waymo to get into Uber and for Uber to invite Waymos on board because it creates this kind of a

a tense rivalry with the human drivers. And so, yeah, Uber, what is Uber going to do? They're kind of in a tough situation. But overall, every day I ride Tesla Robotaxi, I'm getting more and more confident and optimistic about its prospects. It just is a great experience, great product, end-to-end neural networks, super cheap hardware, and it works. And I think this is the scalable option that Uber

that Elon has been planning all along. So anyways, stay tuned for some more videos. The next couple of days, I'll have probably some special guests I might take along with me on some rides and discussing more about the implications of Tesla's Robotaxi network.