SPEAKER_00: Okay, everybody, we got a great interview for today. I interview Cruise CEO, Kyle Vogt on the show for a second time.

SPEAKER_01: He just retook the reins as CEO of Cruise after a few years as

SPEAKER_02: CTO. But before we get to that, it's a there's a lot of car talk in today's episode, I'm gonna warn you, but there's also

Jason Calacanis: a lot of Apple talk, we're gonna recap the Apple event and talk about what they announced.

SPEAKER_05: And their legacy of making confusing names continues, we

SPEAKER_06: do our best to sort it all out for you. But essentially, they

SPEAKER_07: have released a Mac Mini that's twice as fat, and four times the price. So big innovations coming from Apple. And then we

SPEAKER_06: talk about a different robo taxi company called pony.ai. And we debate their $8.5 billion valuation and if when they will get

SPEAKER_10: these cars on the road, it's going to be a great episode. Stick with us.

SPEAKER_13: This Week in Startups is brought to you by Indochino. Indochino makes custom fitted suits, shirts and casual wear at affordable prices. Shop for your next best look or book a virtual style consultant at Indochino.com. Right now you can get $50 off any purchase of $399 or more by using code twist at checkout. Vanta. Compliance and security shouldn't be a deal breaker for startups to win new business. Vanta makes it easy for companies to get a SOC 2 report fast. Twist listeners can get $1,000 off for a limited time at vanta.com slash twist. And Odoo. Odoo is a fully customizable and fully integrated suite of business apps that let you build and scale your stack as you build and scale your business. Your first app is free forever. And right now Odoo is offering $1,000 off your first implementation pack at odoo.com slash twist. That's odoo.com slash twist.

SPEAKER_16: Apple had a keynote and as we like to do, we're going to spend a little bit of time criticizing Apple for their absurd decisions and branding and then give them their flowers for when they do something correct. It's going to be a mixed bag today. But Molly, I think we both agree the most exciting

SPEAKER_00: thing to come out of today is the Mac Studio. Explain to the audience in your perception

SPEAKER_18: what the Mac Studio is. I am trying to decide at this moment if exciting is the right word. Okay, but it is interesting. It's definitely the biggest announcement of all of them, right?

SPEAKER_20: Because it was because I would say what we were really hoping for was the VR goggles, which like nobody really thought we were going to get but hope springs eternal. I think with Apple at the point

SPEAKER_18: at which they start leaking, they're probably still two years out. Anyway, the Mac Studio is actually a very interesting new desktop Apple's. They announced their most powerful chip the M1 Ultra and this new desktop computer, which basically looks like if you just stack two Mac minis together and then fuse them perfectly. It's a big fat desktop that does seem and this is important, right? Because it's Apple sticking with the bread and butter like it is aimed at actual creatives, designers, video editors, musicians, the people who have historically made up the Apple brand.

SPEAKER_00: These are the core users who have felt left out in the cold for the last decade because Mac Apple, I'm sorry, would not listen to them. And their complaints were you make computers that are not designed for us. You took our ports away. You don't listen to us in terms of extendability. We want to connect our own monitors. We want a customizable computer because we do things like make movies or do animation or photos. And we need a very powerful computer and we don't want to replace it with an iPad. So you have this consumerization of Apple that occurred because of the iPhone and the iPad

SPEAKER_06: and the iMacs. And then you have the original people who felt left out in the cold. It does seem that Apple has turned a corner in addressing this group of people and not pissing them off

SPEAKER_00: by releasing the MacBook Pro with the M one chip in it, right, and putting the ports back and putting

SPEAKER_02: MagSafe back. And they don't have to live. Right. I mean, they still had to wait such a long time for a desktop. It is interesting because like, it's one or the either. And I'll be interested to hear your thoughts and the audience's thoughts. Either it is Apple finally, you know, giving the right product to its most core group of people, or it's Apple doing what it's done for the last 10 years, which is like

SPEAKER_18: occasionally trot out an updated desktop that is really expensive and has an all new form factor from the last one isn't expandable comes a year and a half after they've upgraded their most powerful laptop

Jason Calacanis: and still feels like, you know, you got to run Final Cut X on it, or Final Cut 10 or whatever,

SPEAKER_32: which is like crappy software. So like, I'm not sure which one of this, these this is,

SPEAKER_33: this is going to be super confusing for people. But there is the cheese grater tower called the

SPEAKER_00: Apple Mac Pro, right? Now, that is the one that you can put a bunch of hard drives in. And it's got power supplies and a ton of ports on the back. And that thing is like $7,000. So this is supposed to be for that same group of people. But a little bit more affordable, I guess. And now, because of this new M1 chip, which is called the M1 Ultra, if you've been paying attention, there's an M1 that

SPEAKER_16: was in your phone, this replaces like the Intel and all these other chips. So they make their own silicon M1, M1 Pro, M1 Max, which is what we got in our last Apple Mac minis, and in our MacBook

SPEAKER_37: Pros, correct, Molly? I have the not max. I think you have max in your

SPEAKER_39: laptop. So I guess we have the pros, which is incredible. Yes, there's the M1, the M1 Pro,

SPEAKER_18: the M1 Max, and now the M1 Ultra. I guess I believe this is the family. Now, I mean, guys, this is absurd. Like, luckily, Apple makes a crap ton of money and always has, but they're just a

SPEAKER_46: mess. It's just like, and also, why not just say M1234? Would that not have worked? Right? Like,

SPEAKER_16: or are they just so different architecturally that they need to be a different category? And

SPEAKER_49: are we going to get an M2 Max now and M2 Pro and M2 and an M2 Ultra clear, like we're dunking on the names and we're right to and and I will never stop like marveling at the fact that I have three

SPEAKER_18: different Apple devices that charge in three different ways. And like some of them are

SPEAKER_02: lightning and some of them are USB C, but then some of them are wireless, but not the same wireless like the AirPods can't charge on the Apple Watch charger, because they're not that kind of like,

SPEAKER_18: that is ridiculous. God, however, as a matter of computer engineering, these chips are incredible. And I have no doubt that these machines are going to be ridiculous. I mean, that Mac Pro MacBook Pro with the M1 in it is like almost scary fast. And I think what they have designed these for

SPEAKER_00: is power to watts. In other words, how powerful can they be? And how little power can they draw so

SPEAKER_06: that your battery lasts an obscene amount of time, which is what everybody says when they get one of these machines? Oh my god, I'm not getting the spinning wheel of death as much. And you know,

SPEAKER_00: my machine is super fast. And the battery life is just bonkers. Like, sometimes I forget to charge it. And I'm like, I'm, I'm, you know, on our 10, 11, 12, and I still have battery power left. Yeah. And on the old ones with the Intel, you know, they're like, yeah, you can get eight hours

SPEAKER_16: all day. And then like you use it, you get like six or five. So for those of you who are not watching youtube.com slash this weekend, subscribe there. Spotify has a feature now, I guess we're in the video beta. When you're listening on Spotify, there's some way to click a button and watch video. And then if you type this week in startups video, you'll find the video feed in Apple. Those are, I guess,

SPEAKER_55: the three ways to watch us on video. So let's play the video and we'll talk over it.

David Friedberg: The combination of M1 Ultra and Mac OS cranks up the performance of Mac yet again. Okay. So now let's talk about where we're going to use this incredible new chip. Okay. Today, we're going to focus on the place where so many people create their life's best work, the studio. A studio is where creators like designers, scientists, and developers change the world. Whether it's in a home or an office, each studio is unique, customized with the tools that complete

SPEAKER_62: the user's workflow. Yes. And for many, the Mac, we have students like I'm feeling it's working. This is me. I'm important. I'm a genius. I have a studio. I'm creating my life's best work. It's

SPEAKER_64: not an office. It's not a closet. It's not a garage. It's a studio. Yet there are some users

David Friedberg: who want even more so they can push the limits of their creativity. First, they want breakthrough performance and capability to turn their studio into a creative powerhouse. Next, they want a wide range of connectivity. Of course. For peripherals that are key to their studio workflow. Yeah, we want to save files fast. And finally, many want a modular system and display so they can create their

SPEAKER_75: perfect setup. Kind of like saying you don't want planned obsolescence. Something totally new

SPEAKER_64: that gives our users exactly what they need to build the studio of their dreams. And here it is.

SPEAKER_77: The studio of their dreams. Yes.

SPEAKER_79: My studio is going to have this lighting.

SPEAKER_82: Whoa. Back it up. Whoa. This is such a letdown though. I'm expecting something so epic and they're

SPEAKER_88: like, and here is a Mac mini. And I was like, I have a Mac mini. It's not that awesome.

SPEAKER_89: It's like a, it's a double smash burger of a Mac mini is what it is. Basically made a double, double. It's a double, double. I'll take it animal style.

SPEAKER_91: Studio and studio display. And studio display. Mac studio is an entirely new Mac with the

David Friedberg: unbelievable performance of our most powerful Apple Silicon. The like lack of symmetry here is killing it. And M1 ultra. And the studio display is the perfect complement to Mac studio with a phenomenal set of features and that. They're so precious. Experience Mac users love. You mean you can plug them in?

SPEAKER_02: Okay. So you can plug the thing into the thing and then they will work together like monitors and

SPEAKER_25: desktops have for time in memoriam. Okay.

SPEAKER_07: Here's my take on this. Like craziness Molly. Yeah. Here's how you could have said it. A lot of you have wanted a more powerful Mac mini with more

SPEAKER_16: ports. We made one that's twice as fast with a lot more memory. Let me show it to you. Also,

SPEAKER_55: you could, we made you an external monitor in case you don't, you won't want the Mac logo instead of one for half the price with a Dell. Like that's the translation. Well, I mean, it all is on the back of this thing.

SPEAKER_33: As the world starts to open up and we all start to travel again, you gotta look good. And how are you going to do that? You're going to use Indochino like me. When I go to the all in summit in Miami, when I go to Australia again, I can't wait for them to open up. Maybe Tokyo will be open up again soon. And I'm going to go to Italy this summer for a week or two. I can't wait to go to Indochino.com and get all of my new suits dialed in with Indochino. You can customize everything from suits and shirts to chinos and bomber jackets at surprisingly affordable prices. And I had an amazing experience going there, getting custom shirts, custom suits. Every piece is made to your exact measurements and you can customize every detail, the fabric, the lapel, the monogram, the statement linings that I love so much. I was amazed at the process. It was so easy and it was so detailed. So here's your call to action. Get $50 off any purchase of $399 or more by using the promo code twist at Indochino.com. That's $50 off any purchase of $399 or more at Indochino.com. And then use that promo

SPEAKER_07: code TWIST to get the 50 the $50 off great suits, a great process so easy, so stylish, you're gonna look

SPEAKER_46: great. I think our real question is why are these big presentations still happening? Like why is that

SPEAKER_18: the way to announce this? Here's a new computer it like we put out a press release and people tweeted

SPEAKER_02: about it and then it went up on the website. Yeah, I don't like every three years or so they put out a

SPEAKER_18: new high end desktop or tower for pros. Yay, it's out now we didn't it's just like it doesn't need a party

SPEAKER_112: anymore. It just doesn't. It's when it's incremental. Here's the thing I think what you're getting at just the ship and let people put that in their tower. Sorry, jumped on. Well, if it's

SPEAKER_113: incremental. Mm hmm. Which this is like you stack two Mac minis on and you put two extra USB C ports.

SPEAKER_117: It's incremental. Yeah, when you do the Steve Jobs Mac book air reveal. Remember he had the manila envelope and he pulls it out of the manila envelope that was like mind blowing. But they're using that

SPEAKER_113: for this, right? And like that would be the same as him doing that same trick.

SPEAKER_16: We put that out of the manila folder and he's like now it's 16 gigs of ram not eight. Yep. It's

SPEAKER_18: like, you know, kind of. And I'm sure the chip is amazing. But there's no way to demonstrate that other than to let people benchmark it like it's just not. I mean, I'm sure that sales spike, I get it. It's a good bump. But the thing is, like people are going to buy Apple no matter what, you don't have a real problem there. And it's just, I think at some point, we have now seen this so many times and it's almost like the boy who cried wolf. So that even if the chip inside, even if the m1 ultra is in fact, like a generation defining technological achievement, they would have called

SPEAKER_02: it that no matter what. And so we don't know. And so it's just gotten a little boring.

SPEAKER_00: I like the boy who cried wolf situation here because just to remind people, if you're watching the video, we'll pull up in sequence here. Here's the Mac mini. Here's the Mac cube. And here's the Mac cylinder. And then here is the Mac studio. And Mac studio sounds like a software product,

SPEAKER_113: not a hardware product. So it's kind of silly. This is like the Mac mini pro. That's actually

SPEAKER_00: what this is. It's the Mac mini pro. It's a small footprint, not as small as the mini, but it's for pro users. So I'm going to call this the Mac mini pro. You want to call it Mac studio. That's fine. I'm calling Mac mini pro. Or you can call it the double mini. It's the double double mini. I don't know. There's a better name for this. But pull up for a second here.

SPEAKER_17: The cube. Now the cube was made had to pull up the year here. But this is like 15 years old. Maybe the cube is this. I'm sorry about the swear. But it just is. I always liked the cube. I think it's

SPEAKER_07: like underappreciated. It was at the time revolutionary. And so let's pull that out and take a look at the Mac. It was, this is when they had that plastic they were using that was see

SPEAKER_17: through. That was like a big deal. Like Lucite. They were like the ones who like in levitated,

SPEAKER_101: like it levitated. Yeah. And I think it had a handle if I'm correct. It did. I believe have a handle.

SPEAKER_18: Yeah. Oh my God. I'm having a flashback to the CNET offices at 50 Francisco. This thing was sitting on my desk because I reviewed it for CNET and everybody who walked by my desk was like, what is that? I believe it also was so hot on the inside that it effectively almost melted. It had some problems perspective. There's the handle you yeah, you would pull it out. And you could actually I mean, this was like the rare victory over the Steve Jobs lock it up mentality because you could actually

SPEAKER_137: change this the pieces of it. It was in fact extendable.

SPEAKER_138: Oh, check out the printer port. What did they call those ports? The 48 pin port, the RGB connector at the top and it had a parallel port, a parallel, a parallel port. You can plug here. Remember the

SPEAKER_16: parallel port with the clickers on the side to snap it in so it wouldn't fall out. But then if somebody tripped on it, it would go flying the opposite of MagSafe. So this was dope when it came

SPEAKER_07: out. I mean, and the way it floated like these things were gorgeous. And then remember they had the cylinder. Yep. And I have two of these I bought these, you know, to make my video editors happy. I like,

SPEAKER_139: you know, I try to give them a little, you know, hardware gift. It was lovely. But you have to

SPEAKER_18: admit it got a bad reputation for looking exactly like a trash can where we're popping it up on screen here. But if you remember, it was like, it was a marvel of engineering and that it was a perfect, it was in fact a perfect cylinder, but it had this sort of odd dip at the top. There it is. It had a lip until you could carry it around by the lip, which was genius. But then it looked like a garbage trash can. It really did look like a garbage can.

SPEAKER_141: I thought it was dope on a desk. It looked futuristic on a desk. I got like Gattaca vibes,

SPEAKER_06: you know, or like Star Wars vibes. It looks like a droid of some type plugs on the back was kind of

SPEAKER_20: dope very easily accessible. But this is what they do, right? Every three years or so every three or four years, like you get a new tower. Here's our little like, here's our stop to you creatives.

SPEAKER_00: Yeah, but they can't seem to decide if this thing should be tall or not. So let's pull up the Apple Mac Pro cheese grater edition. This thing is like $7,000. And I think they were charging like $300 for the wheels. If you wanted it to have wheels on the bottom. So this was the obscene thing they made

SPEAKER_113: for pros. So now you have them saying this really was Yeah, this is for pros. We mean it this time.

SPEAKER_00: Here's the cylinder. Then they were like, you know what, you didn't like the cylinder, we're going to go back to the tower. Because you all want a giant tower. So here's the giant tower where you can put in all your hard drives inside, take it apart, put hard drives in. Yep. And then everybody's like, you know what, maybe I don't want the hard drives inside. Maybe I want to have a rate array and that'll be shared by a group of people or maybe I'm using the cloud. So I don't need a rate array and they'll just have it back up to that whatever rate array for people who don't know is when you put like, you know, five hard drives and they all are redundant and they're faster because you can write to a couple of them at the same time. This is what video is like and then

SPEAKER_07: they decide to go from this to now the double double the double and so up smash burger. I mean, it's just confusing as hell, you know, like just now here's the Mac studio, which is ugly. I mean, let's be honest, I think it's ugly. Yeah, it's not that pretty. It's not pretty at all. This feels like

SPEAKER_16: this feels like they punted on the design. It feels like a punt to me. Yeah, like, let's make something

SPEAKER_122: that is, uh, yeah, just double size. I mean, which I guess fine, like you have a form factor that works.

SPEAKER_18: But if your deal is that you're super creative and innovative and you know, as all of the Apple

Jason Calacanis: fanboys are going to continue to yell at us, they're innovating like this is a double double you put another burger on top of the burger you already had. It's like, so that's why I think if

SPEAKER_16: this was Mac mini pro that would have been a better branding. It describes it. So if this was framed as Mac mini pro, I'd be fine with this. That's really the design. Yeah. But if you're going to see this is

SPEAKER_171: a new revolution, it should be a new design. That just me. That's just me. Yeah. Now, am I

SPEAKER_16: going to buy it? Maybe I don't feel the need to buy this. I don't think my Mac mini with the M one is any better. And I'll tell you what's just crazy about this is the price it says from 1999. But if you put this new chip in it, which is why you would get it, right? Right. You know, then you're

SPEAKER_172: talking about a lot of money. I think it goes up to like 4000.

Jason Calacanis: We cited it starts at three or 39 99. There we go. Yeah. So if you actually make this the machine

SPEAKER_18: that you want, like an upgrade, right, a reason to go from the M one Mac mini, should you have that, then you're talking about and you are talking about a beefy machine. Like I don't want to downplay the fact that you've got a beast, you know, it's a it's a beefcake and you have a really nice port

SPEAKER_177: situation. But it's 39 99 starting. That's obscene 20 core CPU. So if you want the ultra,

SPEAKER_49: you're gonna pay $4,000. And honestly, you're gonna pay $4,000 for a terabyte of storage like a terabyte of storage is like what you get with a Dell right now. Like I'm not trying to be a jerk,

SPEAKER_32: but give me two for $4,000. They're so stingy. They're so stingy and so overpriced.

SPEAKER_182: You know, do I want it? I honestly don't care if I had this. I don't think it would change my life.

SPEAKER_06: I don't think it's life changing. If I was a video editor, I guess I would. But video editors are going to look at this and go and a lot of video editors now are just looking at this kind of parade

SPEAKER_07: of rug pulling. And they're just like, I'll just go with them. PC. Right? I think a lot of them are

SPEAKER_18: using Premiere now. Yeah, Adobe works editing on PCs. But at that point, then I think you would just like spend less for a more extendable machine. But I don't like listen, I would love to know, because before so long, the Mac was the standard the Mac and Final Cut full stop. And then it was the Mac plus premiere. And that was great. But I wonder, like, email us and let us know or tweet at us and let us know. Is this the machine that video editors have been dying for or that or podcasters or engineers

SPEAKER_113: or devs? Like, also, why don't they care about gaming? Yeah, they're making this insane chip. Yeah, that's another thing. Like, shouldn't this thing be able to run like PlayStation Xbox level games?

SPEAKER_06: And then, you know, if I want to play Age of Empires, like, you know, definitive edition or

SPEAKER_07: whatever the new one is that I play, like I don't play it, I can only play it on my PC. And yeah, I don't get it. Let's just play this quick video here. This is the spec video. The exterior is

SPEAKER_194: machined from a single aluminum extrusion with the foot 7.7 inches square and height of only 3.7 inches. Inside, every element was designed to produce an unprecedented amount of performance in such a small form factor. The innovative thermal system begins with a unique double sided blower pulling air into the system across the entire circumference of the perforated aluminum base. Okay. The air moves over the custom circular power supply and through channels precisely placed to guide it. I mean, that's nice, but that doesn't change my life. Finally, the air is propelled through a low impedance rear exhaust containing over 2000 precisely machined perforations. On the back, there are four Thunderbolt 4 ports. So you figured out how to cool your computer. Congratulations. A 10 gigabit ethernet port. Great. Standard. Two USB-A ports. Standard.

SPEAKER_205: 10 years ago. Standard 10 years ago.

SPEAKER_194: And a pro audio jack for high impedance headphones. Standard 10 years ago. Or external amplified speakers. Okay. Nobody cares.

SPEAKER_206: At least they didn't get rid of that. Yeah.

SPEAKER_207: Listen, when you're the founder, it's fun to trade war stories with other founders.

SPEAKER_33: Recently, Balloon CEO Amanda Greenberg, one of my awesome portfolio founders, told me how Vanta's SOC 2 solution helped her save an important deal in the final hours. If you don't know, Balloon sells a SaaS productivity and collaboration software package. It's brilliant. And when they needed 10 documents in place within 48 hours in order to close this deal, well, Vanta saved the day by supplying customizable templates and helping them through the process to close. So if you don't have your SOC 2 tight, you can't close major customers. And you know what? A lot of startups wait. Well, the waiting has to end. You have to work with Vanta's compliance software to make it easier to get and renew your SOC 2. They continuously test against technical and non-technical SOC 2 requirements and they partner with over two dozen audit firms who have been trained to file SOC 2 reports directly within Vanta. On average, Vanta customers are SOC 2 compliant within just two to four weeks compared to three to five months without Vanta. And guess what? Vanta is such a great partner. They're going to give you $1,000 off your SOC 2. Thanks for that, Vanta. Here's your call to action. Get $1,000 off at Vanta.com slash twist. That's V-A-N-T-A.com slash twist. That's Vanta.com slash twist for $1,000 off.

SPEAKER_00: Well done, Vanta. So here's something interesting. They're doing this new double-double Mac mini Pro and it's faster than the Mac Pro that's $7,000. So does that mean the next time are they going to

SPEAKER_16: take the Mac Pro tower cheese grater and put the m1 ultra in it or are they going to do the m1 ultra plus plus in that and then charge 12,000 for that? Hope springs eternal for the plus plus. Yeah, who knows? Who knows? The what's the name of the next? What's next? So you have the m1 then you have

SPEAKER_37: the pro the ultra the max I think is the order or max pro ultra something like that. Max pro max

SPEAKER_17: ultra. I believe max ultra. What's the next one in this lineage? What's number five going to be called?

SPEAKER_18: It's got to be infinity, don't you think? M1 infinity. I mean, that was actually an awesome name for a tip.

SPEAKER_25: What? Yeah, search dog is almost certainly right here. It will be the m1 ultra max pro plus plus.

SPEAKER_229: Well done search dog. For those of you who don't know, we do this live every day at about 10 o'clock.

SPEAKER_230: And this is crazy, but bear with me. Maybe it could be the m2.

SPEAKER_16: Yes, if you could do these incrementally. I mean, I was kind of used to that. Yeah, you know, like iPhone max 13. I mean,

SPEAKER_17: they are better names than the Intel chips though, right? Like didn't that?

SPEAKER_16: Yeah, Intel was just like, well, what did the developer call it? Uh, the Pentium 5.28 B. Okay. Yeah, that's good enough. Who cares? Nobody cares. It's just a chip.

SPEAKER_234: Yeah, you don't need to brand it. Then they at least went to core and then it would be like the core i7 before i9 at least that was

SPEAKER_18: somewhat and the numbers went up, which is normal instead of having it be like super max ultra plus

SPEAKER_07: plus infinity. I also like the idea of putting the year on it. Like, you know, with cars, we say like, well, this is a Mustang, a 2022 Mustang. So you kind of get an idea of how old it is. And,

SPEAKER_240: you know, give it a vintage, give it a vintage kind of a kind of thing in other as just like

SPEAKER_18: for a quick lightning round of other things that Apple announced today. I cannot believe I'm back covering Apple events, by the way, it was so nice to not care about these for like six years. However, Apple did announce they don't only care about premium consumers. And I'm happy to see they announced the new iPhone SE, which is its most affordable phone, the a15 bionic chip. Okay, so we got the whatever I'm leaving it alone. Better camera, obviously, but the price starts at 429. Even though this is their affordable model, it's getting more expensive. This is $30 more expensive than the prior version. They also announced a new iPad air, who cares about iPads at this point, but whatever, it starts at 599. And then in in timing that I'm sure they wish had not happened just as Major League Baseball went on strike, Apple announced that they will be exclusively

Jason Calacanis: licensing Friday night baseball on Apple TV plus, you know, that might be the second most interesting

SPEAKER_33: thing. We knew they were going to come out with an iPhone SE coming out with a new iPad air. These

SPEAKER_07: are both incremental what SE is their low end phone. So the low end phone now has 5g the low end

SPEAKER_246: phone now has a decent camera and decent battery. That's the one for your kid. I mean, let's be

SPEAKER_248: honest, like that's the kids iPhone. Yeah, 429. It's great. That's their like accessible phone.

SPEAKER_07: And it's now got the a15 bionic chip in it. So that's great. It's going to be more powerful. It gets all the new I guess you can use iOS 15 on it now, which is kind of nice.

SPEAKER_06: iPad air, I always love those. Great for kids as well. But I do think the Friday night, you know, doing the sports leagues is interesting, because that's something that TNT or ABC or, you know,

SPEAKER_07: whoever would have gotten ESPN would have gotten previously. And so that's, um, that's a reason to download the Apple TV plus app to get more people to use it. That's got to be expensive.

SPEAKER_234: Plus, of course, being the service, not the device.

SPEAKER_07: Yes, there's Apple TV, the device, there's Apple TV plus the free content and there's Apple TV plus.

SPEAKER_02: I think TV plus is the subscription service. You gotta pay for that. Yeah, you're not going to get this

SPEAKER_16: for free. Friday night baseball is not for free. You have to have the paid service. So there's a way

SPEAKER_18: to get subscribers, but it continues. I mean, it shows that Apple is apparently continuing to bet to invest in Apple TV plus. I mean, they've obviously had some really expensive original content. Yeah, I would argue that this is also a way to say, we're looking for reasons to keep people coming back after maybe they watch foundation, and then they cancel.

SPEAKER_147: Yeah, they had Ted Lasso. That's like American apple pie. Now they have baseball that's as

SPEAKER_55: American as apple pie and baseball, literally. And I want to see this new one that the Ben Stiller, dystopian like your work life and your home life is separate. I forgot the name of that.

Jason Calacanis: Oh, yeah, severance. I just started watching that. It's um, yeah, with Adam Scott. Yes.

SPEAKER_55: Yeah, just follow me on Twitter. So I'm that could be another celebrity bromance for me.

SPEAKER_32: Yeah, there's a bug in here. It is a super upsetting. I watched you. I mean, in a great way, like it's really good to think about. I later saw that they were releasing it weekly. So I

SPEAKER_55: specifically said to myself, I love Ben Stiller. Um, I like that Dannemore he did. That was good. I love Adam Scott from Party Down. If you don't know that show, it's pretty great. Oh yeah. And Parks and Rec. They're doing a new party. I never watched Parks and Rec. So maybe I got to get into that.

SPEAKER_49: Oh, really? That's such a good one for you to start because it's like got a nice big catalog.

SPEAKER_46: Oh, cool. That's a fun one.

SPEAKER_55: Is it fun? Yeah. Uh, is it okay for a 12 year old? Maybe I'll watch for my daughter. Yeah, perfect. Yeah. So I think this is, uh, I think Apple TV plus good content. I give them some credit for this. I think this is kind of a good announcement for them. It'd be interesting to see if, you know, for these sports leagues, I think their future of the sports leagues is having a direct relationship with customers. Like I am an NBA league pass um, subscriber for six or seven years. Cause I'm out of market and my team is in New York. It's just perfect for watching the next games. And I pay 250 a year, maybe 300 to get that.

SPEAKER_09: To watch the Knicks win 30 games. So I'm paying 10 bucks for every Nick win.

SPEAKER_07: It's kind of brutal, but, uh, they have a direct relationship with me, which is the most important thing. And do you pay for any sports leagues directly?

Jason Calacanis: No, it's been, it's been a while since I was that big of a sports. Like I'm a, I like sports a

SPEAKER_18: lot, but not enough to, I just don't, I mean, I know that on this show, it may appear that I watch a lot of TV because I'm like, oh yeah, I watched that. And what really happens is I watched two episodes of everything and then run out of time to ever watch anything ever again.

SPEAKER_06: What, but would you pay? Cause you like the warriors. If you could just pay to have warriors on demand, commercial free, what would you pay for that? Like if it was like five bucks a month.

SPEAKER_207: Okay, great. So you pay for the season, 50 bucks for the season or something.

SPEAKER_102: Yeah, I could do that. I could pay 50 bucks for a season.

SPEAKER_07: Yeah. See, I think that's going to be the future is like, I just don't want to deal with it. I just want to go to the warriors website and just get it all, you know, or go to the NBA site and just get it all. And so it's kind of interesting. These intermediaries like buying a slice of it. Cause you, even if they have the slice of this, I think you're still going to be able to get those games on MLB. I wonder, or if those games will be blocked out and then people.

SPEAKER_282: No, it's exclusive. Yeah. That's like their big announcement, I think.

SPEAKER_55: So then in the local markets, that means like, if you're a Yankee fan and this is a Friday night Yankee game, you don't get to watch it unless you pay for Apple TV. That's going to be super annoying

SPEAKER_18: for those three games for Yankee fans. I mean, those exclusives already are annoying, but yeah, I mean, I think it's, it is, yeah, it's a tough one. This is when people go to Reddit and they type in

SPEAKER_283: Yankee stream Reddit into Google to find the pirated streams on Russia. I mean, you know how we've

SPEAKER_18: talked about this so much. Like, this is how I feel about exclusive content. It's it. I don't want my entertainment ecosystem or any of my ecosystems to be so cut off. Like, I hate it. I hate the idea that I can't just plug and play whatever I want into another thing. And like, I'm sitting there in front of a TV that has 19 apps on it. But like this one, I can only get over here. And, and then the idea that you would pay for MLB, but also have to pay for Apple plus just to get some exclusive games. It's just like, it's messy. And honestly, it's just gonna make people want cable.

SPEAKER_287: It's gonna make people want to pirate too. Empire. So I think that's the thing is like,

SPEAKER_07: when you make this too hard for people with too many services and too many carve outs,

SPEAKER_06: they're just gonna go try to steal it. Like, I noticed like on tick tock, YouTube, and Twitter,

SPEAKER_00: whenever there's a UFC fight, I see UFC trending. I'm like, I don't like the UFC enough to,

SPEAKER_16: you know, pay for it. But like, I if I see it trending, I might be interested in seeing the knockout blow, you know, like the last 30 seconds of the fight is kind of interesting to me just to see how the person got knocked out. Yeah. And inevitably, you know, just even if you're not

SPEAKER_07: looking for it, you're gonna see the pirated stuff. And so you kind of get, you know, these fights

SPEAKER_55: only last 10 minutes or less. So you're gonna find it very quickly being stream somewhere,

SPEAKER_07: it seems like each new social network gets a free shot at it. So it was like periscope had it for a while on Twitter, Facebook, YouTube, and now it's tick tock. So tick tock is like the ultimate

SPEAKER_55: place to watch a UFC fight. If you want to watch a pirated UFC fight, people just take out their tick tock, they live stream and they pointed out their TV. And they can't keep up with it. It's a live

SPEAKER_25: streaming. Yeah, totally. I do wonder what what piracy numbers are at. Yeah. Anyway, yes.

SPEAKER_07: Mm hmm. It's I think it's generally kids who do it. And just people who are not the core users,

SPEAKER_290: because you view the core user, you want to watch it in real time, you want to see the knockout blow before everybody else does in real time.

SPEAKER_292: Listen, when you start scaling revenue quickly, your company needs to be run professionally. And Odoo is the software that helps you maintain control of your fast running business. Odoo is a suite of business apps where you can run your entire company from just one platform. This means you don't need to keep adding siloed SaaS products. Everything you need is there waiting for you to turn on when you're ready. Sales, accounting, HR, website builders, and so much more. You're going to streamline everything by bringing your apps onto one platform. No more issues transferring data between platforms. And you'll have one customer support contact across all of your app. Plus, if you only need two or three apps to optimize your workflow, that's all you're going to pay for. Odoo has over 30 main apps and over 16,000 apps from their open source community. And the best part? Well, here's your call to action. Your first app is free forever. And Odoo is offering a $1,000 credit on your first implementation pack. Just go to odoo.com slash twist for $1,000 off. That's odoo.com

SPEAKER_02: slash twist. All right. We got our startup of the day because we cannot get enough

SPEAKER_18: A startups, but B robo taxis. Pony.ai announced the first close of their series D at an $8.5 billion valuation to continue building its fleet in the United States and China. The funding amount and the investors interestingly were both undisclosed. Pony CEO James Pang was on episode, uh, 949 way back in June, 2019 in the before times before the pandemic, when people did want to take it before Molly. Yeah. Um, Pony said it has a billion dollars on its balance sheet. The company has over a thousand employees and is testing in China and us markets, both Beijing, Shanghai, Guangzhou, Shenzhen, Irvine, and Fremont. And it's on its sixth version of their robo taxi kit. What do you think about an eight and a half billion dollar valuation for building like a fleet

Jason Calacanis: of robo taxis that actually from a capital perspective doesn't seem like crazy. Okay.

SPEAKER_06: So 8 billion. If it was, uh, 10 times, uh, your top line revenue, your sales would be 800 million in sales. If you had 800 million in sales and it was $8 a ride, uh, it would be a hundred million rides.

SPEAKER_102: Oh, okay. So yeah, that's a really high valuation.

SPEAKER_25: Probably because there's no way this car I've never seen before is doing a hundred million rides

SPEAKER_146: around Fremont. Yeah. So a hundred million rides. Um, although those are really big Chinese cities.

SPEAKER_02: I should maybe walk that back a bit. Like those are gigantic, you know, we don't always remember

SPEAKER_00: how big those cities in China are. It's 2 million rides a day. Yeah. You know, ballpark 2 million rides a day, 2 million rides a day in 200 cities, 20,000 rides in a city. It's not ridiculous if they were at scale, but now getting to scale, as we all saw with Uber and Lyft, Uber is in a lot of cities,

SPEAKER_06: like over a thousand and Lyft is in a couple of hundred. Like it takes a decade to deploy to that

SPEAKER_07: many cities. So this is a future looking, uh, valuation. If you were to do it based on revenue, if they were to sell each car for a hundred thousand, they would need to sell 8,000 cars, no 80,000 cars or something, but that's not their profit. That would just be the top line. That'd be a low margin business. So, you know, yeah, it's probably a very rich balance. It's probably a very rich uh, valuation and it's based on them becoming a winner and it's the same value. I don't know what Lyft's valuation right now is, but that's comparable to Lyft's valuation and it said whatever

SPEAKER_55: the market's really crushed right now, a sixth or an eighth of, right, probably a seventh of

SPEAKER_07: Uber's valuation and Uber is doing billions of rides. So yeah, it doesn't make a lot of sense.

Jason Calacanis: Although an Uber is not no longer in China, right? There's DD and I don't know if DD is

SPEAKER_18: actually doing autonomous in China. So as a robo taxi play that simultaneously in these big Chinese cities and the U S I could imagine them. It's just sort of an interesting reminder that there are more companies in this and that some of those companies are definitely being incubated and built in the Chinese market so that they might get to scale without us even noticing.

SPEAKER_55: And we have, you know, cruise is doing a limited pilot in San Francisco in a limited area, basically the areas without a lot of pedestrians like pack heights and it's working. And so, you know, we're probably five years out. I think most people would say from having this work in, you know, the cities that don't have snow and ice and then maybe 10 years for the cities with snow and ice and, you know, the edge cases. So, uh, lift is at 12 billion. So this thing's almost as valuable as lift. That doesn't make a lot of sense if it's two thirds of a lift. So the valuation probably should be three, four or 5 billion, but you know, investors in these early stages are betting that it will 10 X. And so, you know, that's not completely unreasonable if you wanted to place a bet if it had

SPEAKER_05: 10% chance of 10 xing. So yeah, I do think, uh, you're going to see, uh, Uber, Lyft and DoorDash

SPEAKER_06: be the beneficiaries of this technology because there's a group of people who want to compete with those startups. Then you have a group of people who want to sell to those startups. So which is a better combination? The fact that Uber owns a piece of Aurora, is that the one that spun out? So they have that as a possibility, then they have everybody else who wants to sell to them. So you'll have some people, I think Pony AI wants to do their own Uber, Lyft competitor and DD killer in China. And then you'll have people who want to just sell to, you know, the ride sharing companies and let the ride sharing companies do everything else, build a brand, do a super app.

SPEAKER_18: That's what I'm trying to figure out about Pony AI is because they look, it looks like they make a kit. I'm realizing the more I look into this, they are on top of cars. Yes, goes on top of cars. And so maybe they don't care. Maybe they want Uber or Lyft to be a customer.

SPEAKER_06: Yeah, it's not clear from their website, if they're going to compete, or they're going to be

SPEAKER_00: customers. And you know what, they might. If I was running one of these companies, I would leave the

SPEAKER_25: optionality there, right? I'm trying to be aftermarket autonomous kit. Great, you don't have to make the

SPEAKER_16: cars, right? You just make you just focus on the kit, you add the kit to the top of cars now. And the fact that they can put this on five different types of cars, I see they have a Hyundai, they have a Lexus. So if they can put this on five types of cars, then they have,

SPEAKER_07: they get to draft off of the car innovations. They don't have supply chain problems. But they don't get to do a full stack experience like the new cruise that lets you have no steering wheel and passengers facing each other. So it's pluses and minuses, I think they'll be able to deploy these quicker. And then they can either sell them,

SPEAKER_16: or they can compete, and they can just pick a market, maybe in one market, they want to compete

SPEAKER_55: in another market they want to sell. So keep your options open is a good idea.

SPEAKER_25: That's pretty that is if indeed, that's what they're doing. And it's a little bit hard to tell.

SPEAKER_18: That is quite clever. Yeah, because I think in the future, don't you think we've talked a little we've danced around the idea of like, what's not dance around the idea we've like glanced on do electric conversion kits exist, there have been, I've heard of at least two companies, maybe that tried to make a run at autonomous kits basically, like plop this on and turn any car autonomous. Yeah. And I don't know that that's a very fully fledged. No, I think that I think that works.

SPEAKER_07: It would be a cool idea if it's doable. I think it is doable. I mean, I think that's what crews did originally, then they got bought. And that's essentially what they're doing. Yeah, is there, you know, attaching these things, and then they just take over the steering, the braking, all that stuff is easily controlled electronically. Yep. So, you know, it's coming. It's late.

SPEAKER_312: Producer Kushal says, one of the notice says comma AI is doing kit style in the US. So maybe this is

SPEAKER_321: that we have a comma AI founder on he's a unique individual. You know, he's like, George Hots.

SPEAKER_18: H.O.T. Z. Yeah, he's a he's a very, he's exactly who I was thinking of when I was like, I know, there's a company that's trying to do this. Yeah, I met him ages ago at a concert, and I was at a concert, a convention, a what do you call those? Yeah. Anyway, like a convention type thing, and was just like, you seem a little nuts, and maybe a genius.

SPEAKER_102: Um, he was just very vocal, like he wrote a lot of blog posts and was like trolling people

SPEAKER_18: and yada, yada. So there you have it startup of the day, the the ecosystem is still vibrant people, there are lots of people doing interesting things, super cool startups, we want to hear about all of

SPEAKER_16: them. Keep them coming. It would be good for us to do, at some point, a handicapping of the race. So I, I don't know that there's an expert in the space. But if we could line up, you know, these are the seven players, here's how many employees they have, you could rank them by employees, number of cars on the road, number of miles driven autonomously, etc, we might be able

SPEAKER_07: to handicap, you know, who was in the, you know, like we did for the goggles, and, you know, XR VR AR, we kind of handicapped who we thought would win, and who would win which percentages of market share, we could probably do a similar thing, Molly with this. So like that might be an interesting idea.

SPEAKER_18: Yeah, because people are still working on cars. Like, you know, I mean, it's not it's nowhere near settled, just because it's gone quiet for now does not mean it's no shaking out or consolidating in any

SPEAKER_16: real way. Like people, it's still a dogfight. It is a dogfight. I'm putting this at five to 10 years before, you know, anybody who's in our audience, you know, I would say five to 10 years, is for when the majority of our audience, you know, 51% of our audience in the United States

SPEAKER_55: is has this as an option available to them. Yeah, somewhere between five and 10. I think 10% of our

SPEAKER_07: audience might have this as an option in two or three years. But it's it's going to take time. Everybody keeps saying it's like next year or two years. And everybody says been saying that people thought this would be available two or three years ago. But the edge cases are just too hard. I mean, you've had your own challenges with autopilot. And you know, I don't I'm not I really want to get in the autopilot beta. Because people who I'm watching the videos of those betas, you know, but you have to have a 99 safety score. And I have like an 89 or 90 or something. Because two people drive the car,

SPEAKER_298: and one person cares about getting in the beta. And one person is not saying they're parceling out the

SPEAKER_06: autonomous based on your driving score. Yes. So they gave everybody who's got 100 gets it. So if you're not driving, like making fast turns driving too close to the car in front of you,

SPEAKER_07: they're like, Okay, this is a conscientious driver, who we can put in the beta. And then

SPEAKER_18: we can trust those betas control because I was gonna say, yeah, it's almost like counterintuitive. Like if you're a bad driver, you need autonomy. But I see what you're saying is you want somebody responsible who's not going to drive it and not going to lay in the backseat and take a video.

SPEAKER_06: That's the problem with this is like, we're basically deploying this based upon stupid risk

SPEAKER_246: takers stupidest behaviors, like the people who stand on top of motorcycles or do whatever those drag races are called the spin outs they do in Oakland and on freeways here in the Bay Area. Yeah,

SPEAKER_07: what do they call them? Sideshows. Yeah, these crazy sideshows where people wind up getting killed, like, we're basically, you know, if people who are doing illegal sideshows, like it's fast and furious, like we'll deploy the technology based on their behavior. It's really dumb, because it is definitely the stay in lane stuff is now in a lot of cars. If just if we had stay in lane, it was required on freeways, like seatbelts and airbags are required now, man, what a game changer that would be if when your car drifted into another lane, your steering wheel vibrated, and it made a beeping sound. Why is that not standard? Yeah. And it's surprising that it's

SPEAKER_18: not because it is available on a lot of cars that exists. I mean, it is super easy to turn it on. It's solved technology. But in some cases, you have to pay extra for it too. And that a extra

SPEAKER_16: and unconscionable. Yeah, it should be standard like this is if we had functional agencies.

SPEAKER_07: What percentage of car accidents on freeways are people drifting into another lane clipping another car, it's got to be half, right? Yeah. Like rear ending is half and whatever rear endings, a third drifting out of your lane is a third and a third is other. Yeah, if you like right now you get most cars have the ability to do an alarm or hit the brakes if you're going to hit another car, the alarm should be 100%. If you're not the proper length, that should be standard like a seatbelt,

SPEAKER_18: right? Mm hmm. Yeah, it really should be like, I cannot disagree with that. And that would just it would just improve so much. I was thinking about how on the 880 in Oakland, there are all these free all these like billboards that are basically there's some word for it, you know, like if you just sort of rub up against somebody, they're like, just pull over like it's, yeah, get off the freeway. It happens so often.

SPEAKER_06: There you have it folks. If you have a startup of the day email producers at thisweekandstartups.com. Speaking of robo taxi handicapping. Next up is our interview with Kyle Vogt from Cruise. Enjoy.

SPEAKER_345: All right, everybody, I've been doing this show for a long time. So long, in fact, that we've been having a conversation about self driving cars from the inception of the

SPEAKER_349: commercialization of this industry. One of the first players in the space shortly after all of these amazing XPRIZE style contests were going on to do autonomy in deserts and figure out if this was even possible 20 years ago was Kyle Vogt. He is the co founder and CEO at Cruise. And he was the last time he was on and was back in 2014. And for those of you who are into the archive, just jump in and look for episode 459. And that was a year after he had founded the company. Cruise was sold to GM as many people know in 2016. A really prescient and smart move by GM to dive into automation. And he was the founding CEO from 2013 to 2019. Today, he is back in the CEO seat. He had stepped back to become CTO and

SPEAKER_210: president. And actually announced that the day we're taping this Kyle is now the CEO. Again, welcome back to

SPEAKER_350: the program. Gosh, how many years later is it now? 4556 years later, it's a few. Yeah, thanks for having me,

SPEAKER_353: Jason. Good to see you again. Good to see you. And I was just watching on Twitter a bunch of people talking about

SPEAKER_07: the mind blowing experience of getting into a cruise vehicle in San Francisco to kind of think is one of the hardest cities to navigate given topography pedestrians, uh, regulation, California, high regulation, uh, state, and certainly San Francisco amongst the highest regulated cities of all. So tell me how is the

SPEAKER_349: pilot going? I know there's, I think I heard there were two cars on the road. It's just cruise employees

SPEAKER_357: right now, but maybe you could tell us about what's going on. Uh, well, our, my first ride and the first ride, uh, in a driverless car in San Francisco was back in November, November 1st. So just a few months ago. And since then we opened up our wait list to the public. Uh, I think we're in the 10 to 20,000 people, um, waiting to try out the service now, and we've been adding people from it, but obviously like people are at, you know, signing up on the list faster than we can, we can deploy them. At this point we're in the, you know, tens of vehicles deployed driverless. But what that means is right now we're only operating from about 11 PM to 5 AM. And over time, we're going to expand the service area and the hours, but around 11 PM, all of these cars launch, they go out of a parking facility and one by one without a driver, they kind of take off and start dispersing within the city, waiting for rides. And then, uh, you know, we've got enough users on the platform already that they're

SPEAKER_360: kept pretty busy, uh, throughout the night. Wow. Um, and did you say 11 AM to 5 PM or 11 PM.

SPEAKER_361: So it's mostly overnight right now. And our approach is, you know, a gradual buildup. So we're going to start with the least busy traffic times, make sure everything is working,

SPEAKER_357: you know, well, and that people are responding to it well, and then really quickly open up the hours and, and, uh, geofence. And I'm glad we did that because we've already seen some things where it'd be hard to predict until you open up the service. And we don't want to like, you know, roll into a new city and shock the city by having a thousand driverless cars on day one. I think there'd be a lot of confused people out there. So we've been taking our time, but, uh, things are looking pretty good now. And I think we're, we're going to be gearing up, uh,

SPEAKER_350: to expand pretty soon. Uh, and so to be clear, this is a level of automation where there is no

SPEAKER_349: safety driver. We saw Uber was doing some, you know, work, I think in Arizona, it was, and other folks have been testing self-driving, but there's always been, um, uh, what do they call it? A backup

SPEAKER_361: pilot, a backup driver's got a name for him, but it's a person whose job is to, to watch and take over if they need to, but really like, you got to break down the world of self-driving cars really

SPEAKER_357: into two categories. There are systems where no human is responsible for anything like there, or there could be no one in the car, as is the case for ours completely does not rely on a human at all. And then there's the other category, which is work some of the time, but relies on a backup human. And in the case of, you know, ADAS systems by major car companies, they may do like steering while you're in the lane or do some other maneuvers, but they're counting on you to grab the wheel if the system makes a mistake. And the same is true for self-driving companies that are testing. They have people sitting behind the wheels, ready to grab the wheel and, and take over. But there's a certain level of performance or hurdle you need to cross before you can remove the human from the vehicle entirely. And that means all the, there's all these backup systems built in. You're kind of ready to handle any sort of scenario, whether it's getting pulled over by a police officer or a hardware or software failure, no matter what it is, the vehicle can basically handle that

SPEAKER_350: situation without someone in the car. You mentioned it was geofenced. Uh, so it's a subsection of San

SPEAKER_370: Francisco because we both know there are some areas of San Francisco and basically any city that's

SPEAKER_349: existed and wasn't pre-planned. You know, Europe comes to mind with tiny, tiny streets that are, you know, uh, you know, in some cases, hundreds, maybe even a thousand years old, which part of San Francisco did you decide to, to do the footprint in and what were the considerations

SPEAKER_352: there? In terms of area, it's about a third of the city. So it's big enough that you can actually use it to get around and, you know, go to a restaurant or something like that. And if you

SPEAKER_357: live in this area, take it home. Um, and it's, uh, in terms of neighborhoods, it's Pacific Heights, the Richmond, Golden Gate Park, sunset, uh, and a few other areas kind of near those neighborhoods.

SPEAKER_349: And those neighborhoods for people who don't know in San Francisco have less pedestrian density. I'm sure that was taken into account and also the overnight was taken into account. Pedestrians,

SPEAKER_07: are they still, especially in a city where people are enthusiastically, uh, out and about, I think would be a great way to describe San Francisco or New York.

SPEAKER_361: That was a major consideration, wasn't it? Yeah, but there are, there are, um, you know, concert venues and bars and restaurants where people are there late, you know, until two in the morning.

SPEAKER_357: And so it's not like this is a, a desert absent of any vehicle or pedestrian traffic. They're out there. Um, but we wanted to start somewhere. This is the, the best lowest risk latest place to do it. And we're, we're pretty confident in the performance of the system because we actually test with people behind the wheel and the entire city 24 seven. So we know what the most stressful situations look like, but our approach has been to test in the, in the difficult areas and then deploy starting in the, the least risk areas, just to build some confidence and make sure that all of our testing and all of our assumptions and all of the homework we've done matches up with our observed performance when there's no one behind the wheel, because until you do that, you don't truly know, um, you can, you can simulate and estimate and make all these assumptions about performance, but at some point you actually have to do it, measure the results and start that feedback

SPEAKER_370: loop of learning from real customers. Tell me what, what happens in these edge cases where

SPEAKER_349: a pedestrian decides to stand in front of the car and, or a double park car on both sides of the streets, not able to navigate that. Do you have people watching in some command center somewhere who can say, you know what, let, let's get a driver out there or somebody to navigate this? Because sometimes you, you, you run into that as a human, you, you, this, I don't know, police activity on a block, ambulance pulls up, some crazy person decides to leave their car and block the whole street. These things do happen in cities. So what happens then in these like edge cases? Because, you know, driving on a grid or I use autopilot basically every day, I go to Tahoe, you know, on the, on the 80, I can use my Tesla's autopilot the whole way, no problem, except for maybe some of the windy

SPEAKER_321: streets at the end. Um, but the edge cases seem to be the, where we're at in the industry, correct?

SPEAKER_361: Yeah. And, and the edge cases or, or the long tail of types of events you see is very long. There's a lot of random stuff that can happen. Longer than you anticipated when you started the company, I bet. Exactly. Exactly. So the approach for us is like, what's the minimum amount of, um, capability we can

SPEAKER_357: build in the system to address the largest number of those edge cases. So there are a number of things that we have a remote operator to be able to see telemetry from the vehicle and then guide it or navigate it in an example of being too far down a one way street, um, or any other of scenarios. We also have, um, a system to try to detect whether there's been a collision or even a person walking really near and brushing against the vehicle. We want to know if there's any contact between a pedestrian and a vehicle. And so we see those events. And so, you know, if there's a pedestrian mesh that messing with the vehicle, most likely it's going to trigger a call to a remote operator. That's going to review the footage and confirm whether there was, you know, something that needs to be addressed, um, or there's an emergency, you know, whether it's, uh, you know, the fault of the pedestrian or not, we don't care. We want to know that something has happened and make sure we disable the car and pull it over. And, uh, so the systems are there, the remote operators can basically determine whether they need to disable the car, where they need to pull it over, where they need to relocate it, a lot of tools at their fingertips to handle, you know, whatever we might encounter out, out in the wild, so to speak. And that's, that's the ultimate

SPEAKER_388: safeguard is that somebody can actually take the car remotely, park it, basically drive it like a,

SPEAKER_349: uh, uh, like an RC car or something remotely. They can go drive around something, et cetera. And then if the car does get attacked by a pedestrian, which is not inconceivable in our fine city.

SPEAKER_357: We've seen some strange things. Yeah. Yeah. It can get weird. It's not like driving an RC car. Cause we have these, you know, we're talking to a car over a

SPEAKER_361: cellular network, which is not necessarily reliable or, or, you know, high throughput. So the interface is more like you can reposition a vehicle and sort of drag its location, you know, like on a map from here to here, and it'll figure out how to maneuver over there. And the goal is to not have any safety critical actions like swerving or hitting the brakes. We don't want a remote operator to do that.

SPEAKER_357: We want them to give high level directives and the car, which can respond instantaneously is responsible for all those, um, safety critical maneuvers.

SPEAKER_210: What is it like to sell your company to a big company like GM selling cars? Um, how is that integration worked or they left the company on its own and just, you know, put it over

SPEAKER_349: here and said, Google did with YouTube. Hey, you know what you're doing, go get it. Or is it really

SPEAKER_07: tightly integrated into the product lines, uh, in the existing car company?

SPEAKER_357: Yeah, it's a good question. Well, to start with, I don't regret it. I think it was the right decision for the company, uh, in time. And I think it would have been hard, you know, really GM buying crews is, is what, in my view, created the, the flood of investments that flew into the self-driving space, uh, immediately after it would have been harder for us to raise, uh, the capital we needed to get to where we are today, absent a major OEM coming in and making that big bet and putting, you know, uh, a lot of money behind these things. And so that was, I think it was important for us to get to where we are. Uh, and I'm glad we did that, but it's hard. It's hard when you get acquired by a big company because in, and I think, you know, if you look at the statistics on these things, most times the company buys a small one, it doesn't go well. And I think, you know, a couple of things that worked out well here is the leadership team at GM, you know, all the way up to the CEO, Mary Barra and the president, Dan Ammon, um, really saw around the corner. And I think the, the, uh, where the future was, so this wasn't like a mid-level VP or a business development person pushing the deal. It was, it was from the top. So that's helpful. And then I think the other thing,

SPEAKER_361: frankly, is like, I'm really stubborn about this. I want to make this thing succeed. And so no matter what weird, you know, big company, uh, complications come up, you know, we're going to find a way through it and, uh, and be persistent and stick it out. But like everyone operates with the best intentions. It's just these large companies. There's so many different departments and groups and everyone

SPEAKER_357: feels like they have a role to play or a job to do that, you know, it can be very quickly, a small company can be completely engulfed by the parent company. And, uh, it can take the form of just lots of questions and communication, or it can take the form of death by a thousand cuts, where there's just like new policies and permissions. And you have to talk to these seven people and get them all to say yes before you can do anything. And so it really takes a deliberate effort to keep that, you know, natural tendency at bay.

SPEAKER_349: Yeah. And the, the hardware package, uh, when you and I last spoke very expensive, maybe it was over a hundred thousand dollars worth of lidar and things bolted to the top of cars. Where are we at today? You know, if a car costs, you know, let's, I think you're using the Chevy bolts, if I'm correct. So if that's a $50,000 car or something, what additional ballpark

SPEAKER_372: has to be put onto a car to make it self-driving?

SPEAKER_361: Yeah, it's a good question. I think taking a step back though, there's really been

SPEAKER_357: two approaches that have, uh, I guess, risen to the top in the self-driving world in the last few years. One is to take a minimal set of compute and sensing and try to put it in a retail car. And then through a software updates, you know, make it, um, full self-driving and that's unproven, but you know, from logic, from logical reasoning standpoint, you can see how it would work. Our approach has been let's equip these vehicles with whatever they need to make them driverless and start to learn from that and expand gradually and simultaneously drive down the cost, um, through increasing the volume and other tech advantages. But at least we know we can actually operate the business. Uh, and then once it shifts into less of a science project and more of a engineering and operational cost down exercise, we have a better line of sight to, you know, when we'll reach certain unit economics and when we can operate in certain cities and so on. Um, the other approach where you put some minimal hardware in the car and hope that you'll sort of invent magical algorithms to, to get to the performance you need is, uh, in my view, less proven, but it works if your goal

SPEAKER_349: today is to sell lots of cars. Yeah. And so of course we're referring to another electric car manufacturer that, uh, has autopilot and it seems to be working pretty well. Um, so what, what's, uh, is it twice? So you're basically saying you're putting more sensors than you probably need and

SPEAKER_345: over time that'll get cheaper and you may need less. Yeah. And it, and it works out because like, you

SPEAKER_361: know, these, these cars, the way we operate them is robo taxis and we scale up a little bit, we're

SPEAKER_357: going to operate and be operating them like, you know, 20 hours a day. Right. And so that means from a, from upfront cost perspective, you can afford to pay a little more instead of good unit economics compared to a car that sits idle 90% of the time, like when you would buy, uh, so maybe it's 50

SPEAKER_400: grand per car or something like that. I mean, the ballpark range, that's the whisper number on the

SPEAKER_357: street is that these things are below a hundred. It's, it's a lot, but it's not that relevant to us today because we're not building that many of the expensive ones. You know, we, we've announced that

SPEAKER_352: we have, we're building our own custom chips and custom sensors and other things that get really,

SPEAKER_405: really inexpensive. There's a really funky looking vehicle. The, um, cruise origin that you're working on

SPEAKER_210: that looks like, um, gosh, I don't know if it was, um, the, the original dread movie with Sylvester

SPEAKER_07: Stallone in it or minority report, but, uh, a bit of a boxy transportation, not a steering wheel type

SPEAKER_361: vehicle. Tell me a little bit about the cruise origin. Yeah. I mean, one of our goals is to, uh, drive

SPEAKER_357: down the cost of, uh, providing like a ride hail service. And the other is just to make it an amazing experience. And if you think about the car, cars that we ride around in today, especially if you hop into the back of a ride share vehicle, everyone's facing forward, your knees are at the back of someone else's seat. The driver can't look behind and see you. They don't know what you're doing. Um, you know, which can be good and bad. And then, you know, you don't have a lot of space. And so, especially if it's compact car, your, you know, leg room is minimal with the origin. We flip that around the seats face each other and you have an enormous amount of space inside. It feels like your own train cabin or something. It's, it's enormous. And why I think that's good is first of all, it gives you like a lot of personal space and you don't feel like you're in the back of a, you know, compact car at all times. Um, but it also makes these vehicles much more conducive for shared rides because everyone literally has a lot of physical distance between them and they're facing each other. So you're not worried about what the person is behind that you can't see.

SPEAKER_410: Yeah. Never thought about that. Yeah.

SPEAKER_357: And the, and the goal is if we can get more people to opt into pooled rides, which make a lot of sense, uh, from a, uh, utilization standpoint, we can reduce congestion a little bit because there'll be fewer, you know, need for fewer cars. And so that's, that's a big motivating

SPEAKER_370: factor behind that form factor. That was a real bummer that, uh, Uber pool and lift line got sidelined because of the pandemic. It felt like they were making progress. Certainly they were

SPEAKER_349: investing a lot. I think those segments for both of those ride sharing companies were maybe break even or perhaps even at a modest loss, but they did feel like they were figuring out the algorithms and in which cities, you know, it might work. Um, you, you believe in that as well. You believe that the algorithms can figure out how to make these either like bus routes or, you know, just a little bit more, um, conducive to multiple passengers. Yeah. Yeah. And then, you know,

SPEAKER_361: we can experiment with this just like, uh, the retro companies have been, you know,

SPEAKER_357: cause obviously if, if you don't have that dialed in perfectly, these cars can fall back to just serving, you know, rides for individuals, but for us, you know, the, the transportation solution of the future should be better in every way, right? It should be, should be safer over time. It should be better for the environment. It should be, you make better use of our roadways. And really we're trying to hit all of those. Um, cause I don't think we want, you know, this, this glorious future we've been promised with self-driving cars and flying cars, only things we don't want to get halfway there. We want to, we want to nail this and really get people excited about it. Uh, let's talk about

SPEAKER_349: safety, San Frans and regulation. San Francisco has allowed you to do this, obviously. Uh, this, that's had to be a huge hurdle to tell the audience. What, what got you to, um, got you through all this regulation and red tape? Uh, are cities now really excited about this as opposed to maybe a

SPEAKER_415: little nervous about it when you got started? How is, how's the climate change for regulation?

SPEAKER_357: Well, uh, I guess to start with a lot of, a lot of cities have problems they, they want to solve and, you know, self-driving cars, I think can help with many of them. Um, you know, San Francisco has a, I think it's a vision zero where they basically want to get pedestrian, uh, fatalities down to zero. Um, and, uh, most cities want to be greener, have more environment, environmentally friendly forms of transportation. And so self-driving electric, self-driving cars can kind of help with both of those. But, um, you know, there are other things that we think down the road, self-driving cars can really help with one is the accessibility of transportation. So, you know, if you're disabled, you can use one of these things and our goal is to make them very, very accessible. And then also things we've done with the origin and bringing, you know, the sensor and compute costs down and all those other things, we think, you know, eventually this will be much lower costs than, uh, owning your own car or using Uber and Lyft, which increases job mobility and lets a lot more people do the things they want to do, uh, in a city. And so I think those things are, you know, all, all, all problems that cities want to solve on the regulatory side, you know, we really have two big, uh, groups depending on where you want to operate in the U S one is NHTSA who operates at the federal level, and they are in charge of basically making sure vehicles are safe on the road. And then at the state level, um, in most states, the equivalent of the DMV or department of motor vehicles, um, does permitting and registrations for vehicles. And so in California, they have a program for self-driving cars. So we've been working closely with the state, um, and making sure at least people in the city of San Francisco, like first responders and police departments and other people know how to interact with AVs. And then on top of that, we, you know, we've tried to go above and beyond. We have this program called cruise for good. And through that, we've delivered over 2 million meals to people in San Francisco, uh, using our self-driving cars while they're testing. And so we're using this, you know, excess capacity we have, and we've pledged to use 1% of our miles, um, to, uh, to, to give back to the community. And I think those things help build up some, um, willingness to, to let a service like cruise come in and operate, um, and ultimately

SPEAKER_350: hopefully do really good things for the city. If, uh, most ride sharing rides are about 10 bucks,

SPEAKER_210: 12 bucks today in a major city like San Francisco, uh, or New York, where do you think crews will be able to get it to? What do you think consumers will experience when, uh, they start paying for this?

SPEAKER_212: Cause the trial you're doing now is gratis, right? It's free to, to participate if you are lucky

SPEAKER_352: enough to be selected. Yeah. For the time being, we're waiting on one less permit from the California public utilities commission who, uh, regulates basically taxi type services. Um, but yeah,

SPEAKER_361: I don't know, to be honest, I don't know. Um, but what I do know is that it's ultimately going to be

SPEAKER_357: cheaper than hiring a human being to drive someone around and basically paying a full-time wage. Uh, and I think the experience can be better too. I mean, one of the things, and this may affect people's willingness to pay or excitement about paying whether the same amount or less or more, we don't know, but you know, the way these robot cars drive, they're robotic, but in a good way.

SPEAKER_361: And by that, I mean, every time they approach a stop sign or accelerate, they do it with the exact

SPEAKER_357: same deceleration force. They take the corners with the same amount of acceleration. It's predictable. And we've actually tuned it to minimize the amount of like, you know, jerkiness forward and backward that, that you get all the time in a rideshare car. So a lot of people will go on a ride and they're doing emails and stare at their phone and they get to the end of the ride and like, wow, I don't have any motion sickness. I feel fine. Whereas, you know, then if you take an Uber or Lyft somewhere else in the city, you're going to be like, Oh, this is, this is a bit, it's a bit rough. Yeah. I mean, I, I actually, one of the reasons,

SPEAKER_349: like I actually invested in Uber back in the day was because the rides were smoother compared to taxis at the time. Interesting. Where taxi, you know, drivers, I think they usually rent their cars for the first seven, eight, nine hours. They're basically paying back the lease cost.

SPEAKER_390: And it turns out the last four hours of their 12 hour shift is where they make their profit, right? So they're, what do they call it? Like the gate or something?

SPEAKER_349: In New York, it was 130 bucks. So, you know, they, all the medallions, I mean, this was like kind of the tragedy of the whole thing. All the medallions got bought up by rich people,

SPEAKER_283: lawyers, accountants, like the professional class heard you could buy the medallions in New York and then rent them out and you'd make, you know, whatever amount percentage on it and you get a, and yeah, they would rent them out for, it was 130 bucks. It depends on if it was a Friday, Saturday night or a Sunday or a Monday or whatever, but yeah, you'd basically think 130 bucks. So they, you know, they get to the third or fourth hour, you know, they're halfway there and then they get to seventh, eighth hour and then they're finally break even. But by that time,

SPEAKER_185: you've been in a car driving for eight hours and you, you just got to race for those last four hours to try to, to try to, you know, get into the black. It was kind of gnarly and unfair. And then this doesn't surprise me that they're going to be much smoother, uh, rides for everybody

SPEAKER_361: and much safer. And well, I mean, think about if you, if you could pick your top one or 2% of Uber drivers that you have ever had, I'm assuming you've gone on hundreds of Uber rides, pick the top one or

SPEAKER_357: 2% and say, I want that quality of driving every time. You can't do that with people because the distribution of driving behavior is too wide, but you can do that, uh, with robots.

SPEAKER_350: It'll be perfect every time. And they're going to start to learn which streets are also

SPEAKER_370: the most unsafe or have the most unpredictability. I'm certain now in San Francisco, there are problematic intersections or streets. Uh, am I correct? And then has the feedback loop

SPEAKER_349: gotten to the point where you can say to a government agency area, have you done this yet? You know, this really shouldn't be a stop sign. It should be a, a light, or this is a light. It could be a stop sign or this block is just absurdly challenging for whatever reasons. There's a club on it and people spell out into the middle of the street. Yeah. I think, I mean that,

SPEAKER_361: that feedback loop would be a good thing to have. It's probably going to be slow in terms of generating

SPEAKER_357: change, but you know, it, that reminds me one thing that I, you know, this got in the news a lot recently, recently, but, uh, you know, these automated driving systems treating stop signs differently than humans would. I think ultimately one day, once you can prove that, you know, driverless cars don't need to fully stop at stop signs because they're actually doing all the looking that a human driver is supposed to be doing and maybe more, you know, I wouldn't be surprised if in a few years, robots can kind of blow through stop signs and stop lights legally with permission, so long as they've been, you know, proven to be safe in this situation. So I'm really excited for that because, you know, anything that improves utilization or throughput on our roadways is fundamentally good for cities. And so if we can get people places faster or put more people on a road, that's a good thing. And I think there's a lot of things there that just humans are never going to be able to do just because we're fallible and we're human that, uh, the robots can do. And that's opportunity.

SPEAKER_370: Yeah, I mean, we have to turn our heads and we can look in one direction at a time.

SPEAKER_349: Robots can be looking in all directions at all times. So instead of coming to a full stop, they can come to a row or as we would say in Brooklyn, come to a full rolling stop was that

SPEAKER_210: was the joke. But yeah, I mean, I think the Tesla's were, uh, I guess it had to be tweaked a little bit. There was a little controversy. I think they were going through the, maybe they were not fully stopping, I think. And so that is a, I guess, a question is how, how much do you have to stop, uh, at those stop signs? And yeah, that'd be amazing if you could just roll. If the street,

SPEAKER_437: if the speed limit was 35, the, the rule for self-driving cars is, yeah, you can go 15 miles an hour through a stop sign and it's got plenty of time to stop.

SPEAKER_439: You know, it's probably more context dependent, you know, these, these vehicles,

SPEAKER_357: the way we build them, they know where they can see clearly, you know, that there's nothing there and there's no cars coming and where they can't because there's a truck blocking an oncoming lane, or there's a building corner or a tree. And so if you can demonstrate both internally and obviously to regulators and others that when the situation is right, there's no risk by proceeding. I think that's a cool opportunity, but you know, that's, that may be a bit a ways out. I think until then, we can all, we have a lot we can do, uh, driving, uh, within the existing laws.

SPEAKER_415: Yeah. And, and tell me about weather, because that was the other big thing that, you know, we, we live in a foggy town. There are towns that have snow.

SPEAKER_349: There are towns that maybe don't have lines drawn as crisply, uh, you know, as a city might, you know, a suburb might not be as well maintained, or you could have very narrow streets. So

SPEAKER_210: let's tackle weather first. How, how are we doing with weather, specifically snow, rain, fog? Those seem to be the big three, right?

SPEAKER_361: Yep. So, uh, taking a step back here, like, you know, we've been working on this for like,

SPEAKER_357: a cruise is about eight years old, but in the grand scheme of things going from nothing to driverless cars and operation in a major U S city. That's a relatively short period of time. And the way we got there is by picking the one thing that's most important for the company and doing that. And basically only that so that we can get there quickly and, you know, make progress as a company. And so we've deliberately chosen to launch an, you know, an urban environment that has less crazy weather weather than maybe like Boston and New York city in order to do that sooner and get some real world experience with customers, which we think will help us scale up faster than if we waited until we could solve everything before we deployed anywhere. But we've been working on a lot of like our next generation sensors and compute systems and everything else is designed to handle all kinds of weather. And the way these hardware cycles work is, you know, the cars that we have on the road today, a lot of the hardware and sensing decisions were made years ago. And as they go into production and get matured and validated, then they show up, you know, and they're on the road today. So we're working now on the technologies needed to handle, you know, also all sorts of weather capabilities down the road. But in the meantime, you know, there are a lot of places in the US that have climates that, you know, don't have a lot of snow, like this southern portion of the US. And so you're going to likely see us expand into those kinds of markets in parallel to working on the hardware and software solutions for more challenging weather. But I will say in general, the systems work pretty well, even in even the presence of rain or, or fog, we just haven't fully validated them or not comfortable operating them in a driverless mode in extreme circumstances yet. But you know, as we expand out

SPEAKER_444: of the sunny, sort of clear sky parts of the US, we'll take that, you know, much more seriously and

SPEAKER_210: spend more time on it. So once again, California gets the new cool technology first, because we don't have snow all over the road, most roads here. So snow is gonna snow and ice are the hardest conditions for humans. They're going to be the hardest conditions for the for the robots and the AI as well, right?

SPEAKER_361: I think that's right. I mean, it's one thing to handle, you know, a road surface where you still have full traction or most traction, but there's just stuff in the air, whether it's dust

SPEAKER_357: or fog or or light rain. It's another thing when the coefficient of friction, you know, on the road changes, because it's icy, or in the auto world called split mu mu meaning the coefficient of friction, where you can have one tire on an ice patch and another on a dry area, and the control problems get more complicated. And it's not something that's never been done before. It's just something we haven't spent a lot of time on yet, because we're kind of focused on what's ahead of us right now.

SPEAKER_415: How necessary is LIDAR? I know Elon was talking about we don't need LIDAR to do this. How much of your system is dependent on LIDAR versus computer vision today?

SPEAKER_361: LIDAR is very convenient, because it helps you get a really good picture of what's immediately

SPEAKER_357: around a vehicle, especially in the short range. And if you're trying to distinguish between, you know, person laying on the ground, that's wearing a dark jacket, that from a camera looks just like a patch of road, and a person lying on the ground. You know, it's nice to have a sensor that can give you an accurate, trustworthy depth measurement, especially in around the vehicle. You know, when you get to longer ranges, I think cameras and radars are the types of sensors to rely on more. You can get great velocity information from radars, and the precise, you know, position or distance or orientation matters less when objects are far away. So you'd mostly just need to know what they're there. But as they transition from being far away to being closer, you really need a lot of cues, you want to know the exact heading of a vehicle, you want to know its precise velocity and other things. And so, well, you know, many years in the future, I think you can maybe do all those things with cameras, I would certainly want radars as well. LIDAR just kind of shortens the time to get to market with these things, because you don't need a camera based system to be bulletproof and perfect. If you have more than one sensor, you know, to compare readings from what dependency is there

SPEAKER_349: on GPS, if any, because a big part of the early days of self driving was GPS centric, oh, we know where you are on the road. Now it seems the cameras and the fidelity has gotten so amazing, the sensors have gotten so amazing. And GPS is kind of table stakes. You know, you're on Lombard Street, you know, you're on whatever market, it really is about what's going on around the vehicle. So what role does

SPEAKER_361: GPS play these days? Not a lot. You know, in GPS in cities is really bad because you have this,

SPEAKER_357: I call it urban canyon phenomenon where the GPS signals from the satellites are bouncing off the tall buildings that you're driving between and shift your position a lot. And so you can't really rely on it in any way. But the way our systems work is they have a basically a memory, which, you know, it's sort of, you know, maps are one form of memory. And once they get a lock on that map and know where they are, and you may use GPS to assist in getting that initial lock, once you're locked on, you don't really need it anymore. And once you're locked on, you stay locked

SPEAKER_210: on. So the GPS isn't playing a huge role, the cameras in the lighter are playing the big role. And that means and lidar may be less of a role in the future, you think? Is it possible that these systems will be able to get past lidar? Or do you think it's lidar is always going to be part of the

SPEAKER_361: the winning package? Well, my personal view is that today, you know, people are relying on long

SPEAKER_357: range, really high power, high performance lidar to see things far away, especially on highway driving and other things. I think that's got to go away. Because having a, you know, high power laser that you're trying to keep stable when you're bouncing up and down on a highway is just hard to do. But I think it's really useful for short and medium range, immediately around a vehicle and maybe out to 50 or 60 meters. And if that's your goal, that becomes really inexpensive with today's technology, to the point where you know, it's practically the same cost as a camera. And so again, if this makes it convenient, and simpler, and potentially safer, I don't see why you would why you would ever go away from that.

SPEAKER_349: Is cruises business going to be running an Uber lift competitor or supplying cars to ride sharing

SPEAKER_357: companies are undetermined at this point? Well, we're flexible. I mean, I think, you know, I think this has a potential to be very disruptive for ride share companies. And you know, your investor in Ubers, you know, all about their their model and everything. And so, you know, I think these companies can coexist potentially work together. Right now, we're really excited about having, you know, fully integrated service, because I want people when they experience their first driverless ride, I want to control every touch point on that experience and make it really magical and really fun. And maybe things stay that way. Or maybe they don't down the road, we don't know yet. But at least for now, when people have their first experience, you know, in a driverless car, I want

SPEAKER_210: to make sure it's good one. When will consumers be able to buy these? Or is that not going to be

SPEAKER_361: cruises business? Well, today, given the higher cost of the technology that we put in these,

SPEAKER_357: as I said, they make more sense for a robo taxi application. But down the road, you know, we're, we're aware that not everyone wants to use a robo taxi service all the time, even if it's super cheap and super available. There may be times when people want to own their own car or get a different vehicle form factor or whatever it is. And for those reasons, you know, we want to pull that benefit forward in time as well, in terms of, you know, how self driving can make it better. So, you know, with GM, we're likely going to put this technology in vehicles that you can buy or lease or some form of ownership. And, you know, we're working on basically the hardware and sensing and other solutions to make that happen now. And you can see it, you know, in the second half of this decade, for sure.

SPEAKER_353: Kind of a wonky question, but you're collecting massive amounts of data about your city. Waymo,

SPEAKER_349: Uber, Aurora, just so many companies are also collecting all this information about cities. And then of course, you have open maps, and you have the GPS companies, Garmin and Google Maps, Apple Maps. What is there going to be some collective? Or has anybody talked about this, where you could as a self, one of a half dozen automation companies here, be able to share data on challenging situations and maybe have some sort of open mapping type product or project?

SPEAKER_361: Yeah, I wouldn't be surprised if down the road, things standardize. Because, you know, I guess

SPEAKER_357: it depends on how things shake out. If there are multiple AV players that are still around years from now, because right now, everything is kind of consolidating around a handful of companies. If there are multiple players down the road, then it may it may become something where, you know, companies aren't competing or really differentiated on that map data. They differentiate in other ways. And in that case, I wouldn't be surprised if that happens, you know, because it makes sense on the surface, if you have more vehicles contributing to this pool of knowledge, the pool of knowledge is going to be more comprehensive and more up to date.

SPEAKER_445: That's Elon's big advantage right now is that he's got so many Tesla's on the road, he's got the best data set, I would think of anybody.

SPEAKER_357: Yeah, I think, you know, certainly in terms of exposure to diversity of data, having cars everywhere is helpful. You know, it also depends what's the fidelity of the data, like obviously a fully equipped car with lots of lidars and radars and camera, the value of that mile of data or whatever could be more. And then the question is, really, like, can you turn all that data into insights or improvements into the product? And I think that's what a lot of people are trying to build right now is this magic machine, where you just feed it data, just like send in images and other things, and somehow it gets better. Yeah, it's not really the way it works today. There's a lot of people hand labeling data or scripts that are trying to label data and humans cross check to make sure it's right. And then once you use all that data to retrain these neural networks, you still have to go and check and make sure that these new neural networks are better than the old ones, and they didn't get worse in some way. And so we're not quite in this, you know, glorious future where you just throw more data at a system and it magically improves. But I think, you know, around the road, or down the road,

SPEAKER_453: if we do get to that point, and you don't have access to a lot of data, you could be left behind.

SPEAKER_349: Yeah, it's one thing to have the algorithm, you know, throw up a bunch of posts of your friends

SPEAKER_283: and be like, Hey, maybe this birthday announcements more important than this, you know, other, you know, kids picture or a bar mitzvah is more important than this or whatever, like, there's no downside to

SPEAKER_349: getting it wrong, really, or but getting it wrong and self driving, you literally if there's a stop sign that's challenging, and the car needs interventions, you have a human look at that video, and multiple humans look at it and train the data, right? Yeah, it's very, on this edge case journey, I think a lot of the companies are on, there's a lot of manual consideration and thinking through what should happen at this intersection, correct?

SPEAKER_361: Yeah, in this, right. And you know, you can inadvertently teach a neural network,

SPEAKER_357: slightly worse behavior, just by throwing more data at it. Yeah. And so the challenge is to curate data and show it, show these neural networks, examples of mistakes they've made, in a very targeted fashion. And then we have like, call them miners, little scripts that search through our data for certain types of mistakes, and then feed that back into the training data. But that's a it's pretty complicated procedure. And it's not really a turnkey solution quite yet to just go from lots of

SPEAKER_444: data to continuously improving product. But that is the goal for us. And I'm sure many others

SPEAKER_350: guys have made it so far. It's amazing. As we wrap here, what's the what's the hardest thing now?

SPEAKER_210: Is it getting the hardware tighter? Is it this algorithm? You know, the is it the the machine learning aspect of it? Is it the edge cases? What's the most challenging thing you deal with

SPEAKER_361: every day? I mean, you know, the we have so many technical problems, we're solving time simultaneously that there's not any one that I'm particularly worried about, or keeps me up at night,

SPEAKER_357: really, it's about execution. So continuing, this is going to sound simple, but continuing to bring in and retain really smart people to build the right tools and infrastructure to accelerate development. And then to just make steady progress and not, you know, sort of slow down as we grow in terms of size, or start saying yes to too many things and lose our focus. I really think if we just keep doing what we've been doing, and stay focused and execute well, you're going to see driverless cars

SPEAKER_210: everywhere. And when do you think me, you know, consumers in major cities? Well, this will become

SPEAKER_445: the predominant way people get around in taxis, you have to pick a year.

SPEAKER_357: It's hard to say, you know, and the reason for that is, you know, we're really seeing the tip of the iceberg right now, there's been years and years of exponential improvement to these products to get them to the point where you see your first driverless deployments, and people can use them. And so that rate of progress has been rapid and exponential during that time. Now we're starting with a very small number, probably going to see exponential scaling. And so it's really hard to predict the y axis at any point in time, you know, when you have these exponentials. So I'm not going to not going to lock myself into anything other than to say, you know, we're in one major city, a lot of major cities look similar, and we're moving as fast as we possibly can.

SPEAKER_349: Yeah, it's fantastic. Last time you were on in 2014, we did our bet, which expires in 2024. And we made this 10 year bet when debts would be reduced below 5000. We all know debts in the United States about

SPEAKER_283: 30,000. Is that still the number? It's gone up, I think it's closer to 40 now.

SPEAKER_185: Geez, Louise, it's got to always it's and it's got to be people distracted driving, right? I mean,

SPEAKER_357: that's gotta be distracted driving or maybe from COVID people are driving less or behaving more erratically, whatever it is, it's going the wrong direction right now. It's just unbelievable. So we

SPEAKER_349: made this bet $1,000 Omokase sushi dinner was inflation adjusted. So who knows what this will cost?

SPEAKER_478: Yeah, inflation's through the roof. So this could be can be expensive.

SPEAKER_345: I expected a reduction to about 10,000. You went five. You know, it's, it's, I think we're both in

SPEAKER_349: the running here. I mean, time is moving quickly. Overall, when you look at this, is this moving

SPEAKER_350: faster than you thought it would or a little bit slower? Feels like we're going a little bit slower.

SPEAKER_361: Absolutely. It's slower than I anticipated. You know, I think everyone in the industry thought

SPEAKER_357: this was going to happen sooner than it did. And the reality is until you start putting these products out there and exposing them to tough urban environments, you just don't know how difficult it really is. But our belief is, you know, like you said, if we're going to get to this lower number of fatalities in the future, it's not going to happen on its own. It's going to take driverless cars and people pushing forward, challenging the status quo and really putting a lot of energy and firepower into

SPEAKER_392: this to make these products and to get them out there. Absent that, you know, the future is pretty,

SPEAKER_350: pretty bleak. If you had to put a percentage number on how close we are to solving this problem,

SPEAKER_357: put on it. Well, will, will we solve it? Absolutely. Absolutely. If, if cruising, if not cruising, someone else. Yeah, it feels like to me, three or four people are going to

SPEAKER_412: figure this out. So the question is, what percentage are we at now? I have my own thoughts.

SPEAKER_361: Oh, I'd say 100% given that we've got it in a small, small piece operating in San Francisco, but to us, like going from R&D phase for seven years to having a product that real people are using,

SPEAKER_349: that's it. It's out there. Yeah, I feel 80. I feel like you guys are 85% of the way there. I think the snow is like in the inclement weather in the Northeast and the five points. I think these education are points, you know, pedestrian wackiness and then winning over the public,

SPEAKER_419: which, uh, if you're talking about general availability, I'd agree with your, your assessment.

SPEAKER_07: That's fair. 85 feels fair for general accessibility. Yeah, sure. But I mean,

SPEAKER_349: congratulations. I know this has been like, a lot of hard work. Uh, and it didn't go as fast as people thought it would. And it's not been easy. There were a lot of people trying and now you're down to,

SPEAKER_210: you know, let's face it. What is it? A five horse race. I think here in the United States for a five horse race. And I think, you know, society's gonna look back on this and the work you did. And, uh, it's a huge mitzvah for humanity that all of these people who die in the United States unnecessarily in these car crashes, it's gonna go to zero. I think it's gonna go to zero. Uh,

SPEAKER_07: I really sincerely do. I think it'd be impossible to crash a car. It's gonna be like some random

SPEAKER_345: thing like, uh, like commercial aviation. Nobody gets on commercial aviation now. And, you know, on a major carry and expects the plane to go down thing. People get in cars. They're like, yeah,

SPEAKER_185: this is real, you know, high risk, uh, behavior. You can a hundred people dying a day in the United

SPEAKER_357: States. Yeah. It's gonna go down. We, we bump up against the laws of physics. Like the only crashes that remain are physically unavoidable. Um, and those that are preventable, we can prevent.

SPEAKER_372: That's what's gonna happen. It'd be like some boulder falls has nothing to do with driving,

SPEAKER_210: just a random act of God that a boulder fell. Uh, but seriously, congratulations. I think it's awesome what you're doing for humanity and, and to the team over there, you know, just it's really just impressive to watch that you guys didn't give up and, and you know, the other players in the space are still grinding it out. I mean, obviously there's a big prize here. It's great, you know, for economics and business, but there's a bigger issue here, which is young people die in car crashes. I mean, and it's just tragic and you're gonna save a lot of lives. So to the cruise team, uh, congratulations really, and keep it up. We can't wait to be riding around in these. It's gonna be a wonderful future. Even if it took an extra five years. That's great. Well, thanks Jason. We appreciate

SPEAKER_345: that. All right. I'll see you in a couple more years. Well, when you're taking these things out in Boston and getting through tiny streets in Spain. Uh, but for now, I can't wait to ride in one

SPEAKER_210: in San Francisco and, and, and have that nice smooth ride, uh, in Pack Heights and into the inner sunset and everywhere else you are. So continued success. Everybody can go to getcruise.com.

SPEAKER_349: They're always hiring. They need more engineers and people. And how do you sign up for the San

SPEAKER_357: Francisco? It's right on the, right on the website, getcruise.com. You can sign up on the waitlist.

SPEAKER_350: Go sign up and get on the waitlist and I can't wait to try it. And we'll see you all next time on

SPEAKER_487: this week. Bye bye. Hey everyone. Producer Nick here. I want to tell you about the SAS syndicate.

SPEAKER_489: If you're a founder of a SAS company with a product and market, our investment team wants to talk to you, head over to the syndicate.com slash SAS S A A S to apply to raise from the SAS syndicate. And you can join Jason syndicate of over 9,000 accredited investors at the syndicate.com

SPEAKER_490: producer, Justin here, no cool startup. Check out open scouting.com where anyone can refer a startup to our investment team here at launch. Even if you don't know the founder, if you're the first to flag a company for us and we decide to invest, you'll get five K in cash or 10% of our

SPEAKER_494: carry. Hey everybody. Producer Rachel here. Are you an early stage startup that has product and market some traction and are looking to raise at least $500,000 apply today to remote demo day for your chance to pitch to over 9,000 investors in Jason syndicate. Submit your application at remote demo day.com. Our next event is on April 27th. And if you want to learn how to invest in

SPEAKER_489: startups from the world's greatest angel investor, and no, we're not talking about Chris Saka, then head to angel.university to apply. The four hour workshop costs $300 and all proceeds are donated to charity. To date, we've donated over $175,000 to various charities. And you can see the full list that angel.university.com.