This Week in Startups - OpenAI's new models, the MrBeast leak, Waymo ♥’s Uber and more! | E2009

Starting point is 00:00:00 So here you have a nonprofit that took a software asset, spun it out into for-profit, gave, think, 100% of the equity to the employees or some large percentage. And then the nonprofit owns what? A share of the revenue, a license, unknown. And here, because it's software, maybe it's a little bit clujy or maybe it doesn't seem as dramatic as giving people a stack of gold bars. But, you know, seems like a bunch of employees went from the nonprofit to the for-profit. And we're given the goal beyond.

Starting point is 00:00:31 I do think it's really interesting to see how many people are going after coding. That's because of two reasons. Where does a lot of value in the world reside? And then what is AI good at right now? Bingo. This week in startups is brought to you by Washington Post. Stay informed with trusted journalism from the Washington Post. Right now, Twist listeners can subscribe for just 50 cents per week for your first year.

Starting point is 00:01:00 at Washington Post.com slash twist. Assembly AI. Get maximum value from voice data with Assembly AI. Build powerful products and features for your end users on the industry's leading speech-to-text models. Get 100 free hours to start building at AssemblyA.I.com slash twist. And Brex. The financial stack founders can bank on.

Starting point is 00:01:24 Brex knows cash is king for startups, so they built a banking experience that takes every dollar further. Get the business account trusted by one in three U.S. startups at brex.com slash twist 24. All right, everybody, welcome back to this week in startups. I'm Jason Kellyanis. He's Alex Wilhelm. And we're back. We both had projects. We were working on for the last couple of weeks. I just finished up the All In Summit. You've probably seen I've birthed all this amazing content. And we'll talk about, I guess, some of those highlights. But you birthed some of those highlights. something else, Alex, something arguably, much more important. Well, I didn't birth anything, to be clear. I walked around and brought people snacks and so forth. But yeah, we had our second kid and everyone is healthy and home. And we are right back, Jason, in the thick of newborn life. And I had forgotten how great and hard newborns are. It's great because you put them down

Starting point is 00:02:26 and they don't move, you know, unlike a toddler. But then, they don't have a concept of day and night, which is not good. So, yeah. They have a concept of calories running out and needing to attach to mom and get that next calorie fix. And also, if they pass out from a milk, a coma,

Starting point is 00:02:46 then it wears off and they're screaming. It's like milk addiction. Now that you put it like that, it's kind of like dealing with a really teeny little drug addict who wants one thing. The good news, though, is the new, Our new daughter is incredibly, I mean, she's, she's perfect in every way. So it's been edifying, challenging. And I bring all that up, friends, because I'm back and we're going to be easing back

Starting point is 00:03:10 into more live news. So make sure that you've subscribed on YouTube, hit the bell, all that good stuff, because we are going to be back in form on the All-in Summit, though, Jason. I was really. And by the way, how's mom doing? How's mom doing? Mom's good. Yeah.

Starting point is 00:03:24 Everyone's healthy. And all things are nominal. They are inside the house on the couch right now, all cuddled up. while I'm out in the shed. I like how it's like nominal, all systems nominal. Like he's launching a rocket. It hasn't blown up yet. How many SpaceX YouTube lives have I seen?

Starting point is 00:03:38 Turns out the answer is most of them. Correct. Well, just I have two pieces of advice for you. Night nurse. They go together. It's really one piece of advice. It's two words. I know how much you make.

Starting point is 00:03:49 You're doing okay. Family's good. You splurge out of night nurse once in a while. My lord. That is expensive but worth it. All right. I want to talk about all insomensement. that. Before we do our thing, I just want to

Starting point is 00:04:00 quickly double click on this. I'm catching up. I got to see a little bit of the Friedberg Sergei chap. I know Benioff swung by and JD, I think also swung by. JD the Hans, Travis. Yeah, how to go. So this is the third year.

Starting point is 00:04:17 In the first year, nobody wanted to do it. I dragged him kicking and screaming. I said, I think there's an audience for this and we had 800 people come. And it was a lot of scholarships, you know, low price, $500 tickets for up-and-comers and some people paid full fare. So I think at the time was maybe $5,000. Then last year, I had Freiburg take over all the content.

Starting point is 00:04:37 Now, we all reach out to our specific speakers and then I did the parties. And this year, I focused just on my performance as the moderator and, you know, the master of ceremonies, as it were. And I felt actually really good about that. It's the first time I wasn't on stage thinking about the lunch, the dinner, flow and all that kind of stuff. And so I was able to give a good performance, ask tough questions. It's a little bit strange, and we'll talk about it today, because I'm a journalist by trade. I like to ask hard questions, and I have this concept of a follow-up question, which I'll deploy. It's a very strategic thing, Alex, that you can use.

Starting point is 00:05:16 I'll teach you this master Jedi trick. If the subject doesn't answer, you then, you take another swing at it, right? You take the lightsaber doesn't go back on the belt. You keep swinging. And sometimes you ask the second or the third time. You get the actual answer to the question. So it's a different type of event. It's not adversarial.

Starting point is 00:05:35 It's very conversational. But then I also try to push people to, because I know the audience wants certain questions asked. They want a little bit of drama and, you know, getting to the hard questions. So J. D. Vance was great. He's super impressive. You know, you might disagree with his politics. I disagree with him on a lot of things, whether it's January 6th or taking away a woman's

Starting point is 00:05:56 right to choose and the national abortion ban that he was a proponent of. But, you know, I do think he's an impressive individual who will do a lot of great things for the country if he gets the shot and he's young and he's a business person. So, you know, that checks a lot of boxes for me and what I'm looking for more of in government. I obviously have fundamental issues with his boss and we got to that. And then, you know, for me, the Travis interview was like coming home. You know, he after what happened, you know, when it's outster from Uber, he just went underground and just, you know, He never stopped working. He just went all in on Cloud Kitchens and he hasn't talked.

Starting point is 00:06:30 So for the last, I don't know, six or seven years, you know, when I talked to him, I say, you know, if you ever want to, you know, we're friends, obviously, you ever want to have a conversation, you know, the platform's available to you, whether it's here on this weekend startups or at the summit or all, you know, whatever it is. Yeah. You know, inventee hosts, whatever. And this year, you know, we were wakeboarding and hanging out. And I said, you know, we is this year?

Starting point is 00:06:54 And he said, I think it might be. day before he says, I'll do it. And I was like, great. So we didn't announce him. We didn't announce Sergei. Sergei was coming to the event. We're obviously friendly. And he's been working too.

Starting point is 00:07:09 Yeah. People don't know, but he's going to work every day. And I had dinner with him a couple of weeks ago. And it was talking about work. And he was late toward dinner because his meeting went over. Yeah. And I was like, dude, come on, man. Can't be late.

Starting point is 00:07:23 But the great news is, you know, he's super highly engaged. We didn't announce him. So when he came out, the audience roared for both of them. And it was very touching, you know, because I think there's been a pretty adversarial, you know, press corps for the last couple of years. In a lot of cases for a good reason, a lot of cases, I think it's a little bit over the top. It might be link Beatty. I think they both felt like the audience wanted to hear from them. And so that to me on a personal level, interpersonal level. Like it's, as Sergey was taken back. He didn't think anybody would remember who he was. Oh. The audience roared. And then the audience roared at Travis. So I felt very nice about that because they're humans and we all want to be loved and have our opinions valued. So anyway, it was a great time. We had a fun time and it wasn't as exhausting as it normally as for me. Well, I mean, you stopped doing three jobs at once at a major event. And you and I having both worked to putting on major events know that there is plenty of work to go around.

Starting point is 00:08:16 It's hard to do more than one set of tasks. But on the follow-up question and then the adversarial journalism point, I wonder what percentage of media criticism is Stan culture and fandom being shocked that someone might push back against someone that they idolize or want to hold up. Because in journalism, if I say Jason, you know,

Starting point is 00:08:38 blah, blah, blah, and you don't answer me, and I ask again, that's standard give and take. But if you're a pop star stand or something similar, you might view that as I'm being rude or,

Starting point is 00:08:48 you know, harshing their buzz or something. That's standard given take. That's the roly poley of commentary. Absolutely. And you and I are, I think the reason we get along is because we call balls and strikes and we're kind of old school classic journalists. And then there's advocacy journalism on one side, which is kind of weird to call it advocacy journalism, which is call it opinion.

Starting point is 00:09:08 And then on the other side, you have people going direct and not facing any tough questions. And, you know, I think, you know, where we sit, I think is the ideal spot. I'm not trying to lead the witness. I don't want the audience to come away with my worldview. Right. I want to ask the question. And I asked JD, would you certify the election? And he's like, no, there's a lot of things.

Starting point is 00:09:28 Would you have certified it? Yes or no? And then I asked him, I said, listen, I'm going to ask you a third time. And that's my little device to let people know that I'm not going to stop and that the audience also knows that we're here on the third one. And he said, yeah, listen, I would have not signed it, basically. That gives you a lot of information. I'm not telling you you should vote for either party. I just want you to have information.

Starting point is 00:09:48 That's the first clip I saw from the Allens Summit all over my Reddit feed was people discussing. that particular comment from him. And it is good to know what someone thinks about key things, especially if they want to be in charge of the nation. So yeah, I think that was, that was well done. I'm going to watch the Sergei one, of course. I didn't know if the Travis came. So I'll watch that one too. Yeah, you'll like it. Yeah, it was a very emotional one because, you know, there's a lot of, you know, for all of us, PTSD with his ousting, you know, and it's like, it's the first time he's really discussed it. It is the first time he's discussed it openly, publicly. A couple of tears were shed. It was pretty heartfelt. You'll enjoy it. Twist is doubling

Starting point is 00:10:25 down on news this year. Me and Alex, man, we are reading the news. We're bringing it to you. We're giving you our opinion. You know how we stay sharp? The exceptional journalism done by the Washington Post. You know all of their great national news coverage, you know, politics. Obviously, it's in Washington. But you know what they do a great job on? Tech and business. They crush it. It's for you newsletter is a daily personalized roundup of the stories that you care about. most. And they've got great reporters like Drew Harwell going deep on AI. The Washington Post is going to help you broaden your perspective. You're going to get really smart going deep on the technologies and topics that you care about. And you know what? I'm an audio lover. You can listen

Starting point is 00:11:03 to the articles there in addition to reading them, which is great for me when I'm driving or I'm working and I want to have something on the background or I'm on the treadmill. So here's your call to action with the election rapidly approaching. Now is the time to sign up for the Washington Post Go to WashingtonPost.com slash twist to subscribe for just 50 cents per week for the first year. $26 a year. What a deal. That's 80% better than their typical offer. So this truly is a steal.

Starting point is 00:11:31 They want to get young people and technology people understanding how great their coverage is. Washington Post.com slash twist. Go ahead and support them for supporting this week in startups and get all that great content in your email box on your phone. Subscribe 50 cents per week, Washington Post.com slash twist. But we got a full docket today. Why don't we just get to the first story? Because the audience wants to hear what's on our mind. All right.

Starting point is 00:11:53 So the biggest thing that I think that's happened while I was out in the woods is that Open AI dropped a new model at last. If you have been reading the fan accounts of OpenAI, there's been talk about strawberry and Q Prime and et cetera, et cetera, et cetera. We've been waiting, Jason, for this new better reasoning model to come out. And here it is. Oh, one. I have been reading lots of commentary.

Starting point is 00:12:17 analysis about this. Overall, I'm pretty impressed. We'll get into in a second how it works, but I'm curious just from what you've read, has Open AI moved the state of the art forward? Yeah, I mean, we are in an incremental game right now where it reminds me of the early days of, you know, the iPhone and smartphones where it's like Christmas every three to six months. You know, a new app comes out and it's Twitter. It's four square. You know, it's a flashlight, whatever it is, Spotify, the YouTube app comes out. It's just, this is Christmas for tech people. Yes.

Starting point is 00:12:53 And so I do think this reasoning is something that I've heard Sam talk a lot about. Hey, we need to get more reasoning going. And then there was an announcement by Google as well that they had released some data integrity solutions where it's an open source project to kind of check the work, check people's path, check people's statistics and stuff like that. and it was reducing hallucinations by 60%. This is all to say, you know, the car's on the road. It doesn't feel safe.

Starting point is 00:13:22 It feels like you're in an experimental plane or an experimental vehicle. Sure. Now, if you're going to put the whole company on it and people are going to start flying and they're going to go see Grandma and Grandpa and the new baby, you're going to want that plane to have certain features. And that's the error we're moving into, which is safer transportation, safer knowledge gathering, I feel like the whole name of the game right now is to move from these early adopters, the Vanguard, have been playing with this stuff for two years. Now we need to get to that big,

Starting point is 00:13:55 fat middle, which is, you know, college kids are generally part of the Vanguard, but somewhere in between, you know, rank and file employees, and then eventually we'll get the laggards. And so as we move into that, it's just got to be, you know, more reliable. And that seems to be what they're focusing on. What was your impression of it? Did you play with it at all? Did you, did you, Did you see any good examples online? Well, I was going through just prepping for today all the different metrics discussing how good it is compared to other things. Because to me, I'm always interested in the underlying tech.

Starting point is 00:14:23 In this case, we're talking about what is called reasoning tokens and the fact that this is using reinforcement learning. And we'll get into that just a second. But I was curious like, okay, what's the actual improvement? And the metrics are very impressive. So just to give everyone a taste of the progress we're seeing this three to six month Christmas or Hanukkah, we have an image here of what's called CodeForce ELO. essentially how good at programming is an open AI model, Jason.

Starting point is 00:14:47 And as you can see, GPT40 on the left over there, didn't do terribly, made into the 11th percentile, you know, shout out. But if you then look at 01 preview, 01 and another version of the O1 model, you can see dramatic and massive improvements. And right now, just everyone knows, we are looking mostly at the O1 preview. And then also there is O1 Mini, which is smaller and faster and more designed specifically for coding work. And Jason, I picked this example because it seems that the thing that is being solved or being attacked the most quickly with modern LLMs is code generation. And it seems to be getting better faster as that than by pretty much anything else that I can see. And so I wonder if the real use case, the magical moment really is just going to be this is going to help people write a lot of code a lot faster because it just keeps getting better. And eventually it's going to be good enough.

Starting point is 00:15:37 It does seem to understand reasoning much more. So I did a very simple test. So I was born in 1970. I said, what are the 16 most important technology since 1970? And it thought for 12 seconds, you see that here. And it says, you know, when it thought, identifying key developments, if you hit the drop-down carrot, it explains to you what it did. So this is interesting.

Starting point is 00:15:58 This is where you can get into reasoning, evaluating technology. And if it's tracking technical approach, tracking technology, the advancements in technology, tracing things. And it kind of explains it. And then it comes up with, hey, the personal computer, the internet worldwide. Web, mobile phones, smartphones, global positioning, GPS, MRIs, DNA sequencing, renewable energy, etc. Then I said, put these into a bracket and run the first round. It thought for 13 seconds, it designed the bracket, it mapped the technology, advanced to the next rounds, weighs choices.

Starting point is 00:16:26 You get the idea. And here it goes. It says personal computer versus blockchain and cryptocurrencies, the internet, but it didn't actually run it. And so then I had it run it and it did its matchup. Okay. Computer versus blockchain, person computer wins. Internet versus CRISPR. Ooh, spicy. The internet. I would agree with that so far. World Wide Web versus 3D printing. That's an easy one.

Starting point is 00:16:46 Worldwide Web one. Mobile phones versus Wi-Fi. Easy one. Mobile phones work. Smartphones versus social media, smartphone one. Yeah. And it's reasoning. Smartphones combine the functionality of mobile phones with the advanced computer campus enabling a multitude of applications, including social media. They have transformed how we interact. So it basically came to the reasoning, Alex, in this Facebook. that you can't have social media without smartphones. Now, I don't know how it's doing that. Exactly. And this is where I think we're kind of looking at something here and wondering, did it get that from a blog post that somebody else did at TechCrunch where they, you know, pitted these things.

Starting point is 00:17:22 And 10 years ago you pitted these things or somebody said in a Reddit thread or on Twitter or Facebook, hey, you know, smart, you can't really compare social media smartphones. But it did happen. And, you know, I think that's what you're going to start seeing people do, which is, thank you for doing my marketing plan. Thanks for helping me write my blog post or telling me what we should have on the podcast today. But a little more reasoning and getting to an answer is always welcome.

Starting point is 00:17:47 And so we'll see, you know, over time, how correct it is. Yeah. And so those are the reasoning tokens that are part of the reinforcement learning process. And I did go through some papers this morning trying to make sure that I understood this. The best actually breakdown was from a silly YouTube video that I found in which they said, okay, so this is essentially GPT4, but they can kind of, and a self-prompt itself to refine its thinking as it goes through reasoning, as we saw,

Starting point is 00:18:09 to get a much better response. The upside is more complex queries, I would say better results. Frankly, what you just showed was very impressive. The downside, though, Jason, is just that it's expensive to run, uses a lot of compute power. And because it takes more compute, it takes longer. So it's actually currently right now slower and more expensive. But to me, the curve of those two complaints and technology always bend one way, which is down.

Starting point is 00:18:33 things get faster, things get cheaper. So I don't really see a downside to this other than in the next six to 12 months until it gets cheaper and faster. It will be slightly hard to use because it'll be costly. But who cares? What a dramatic improvement? I mean, this feels like progress to me. And I love that.

Starting point is 00:18:51 Now imagine different language models, self-correcting each other and then agents on top of it. So, you know, when you put an agent on top of three or four of these and we ask an agent to do this for us. And there's a third co-host here. And we say, hey, you know, tell us, you know, the history of supersonic flight and give us, you know, the key moments in a timeline. Instead of having a guest on, like, oh, my Lord, there's going to be a third person in the mix every time. And that's going to be a little bit strange for people to kind of understand when there is a replicant in every zoom call. There is going to be, you know, like in Blade Runner, a replicant in every. aspect of our lives. It may not be physically there at the start. Optimists and Humane will be

Starting point is 00:19:37 later, but we'll start having them on call. So I think, you know, it would be really interesting on this Zoom call or in a Slack room or Glue is doing this, David Sachs, Slack competitor. Yeah. It has like some chat in there, but when they start interacting, that's going to be really interesting. And then there was some weird news around their corporate structure and then a financing. So maybe quick hit those for us. Yeah, just keep everyone up to date on all things Open AI because it's still technically kind of mostly counts as a startup in some way. So Fortune recently reported that Sam Altman, we all know Sam, has been discussing next year shaking up the Open AI model. And we've all gone through this discussion before.

Starting point is 00:20:16 Why is it a foundation that owns a nonprofit, that owns? It's a mess, Jason. It's always been a kludge, I think. And it's very impressive that the company actually managed to fundraise and grow as much as it did with this relatively ridiculous corporate setup. I mean, if you're going to back a startup, what do you want to see? A Delaware Sea Corp or maybe a Texas-based Seacorp, but you don't want to see a foundation that owns a non-profit that owns a for-profit entity, right? Yeah. Hey, when founders ask me, what corporate cards do they use? I automatically think of Brex. That's what I use. And when founders ask me where they should store their cash, well, of course, I'm going to think about Brex. Why do I think about Brex so much? Because knowing

Starting point is 00:20:54 everything about your money is not negotiable. Nearly 40% of startups that fell, well, they fell because they run out of cash and 100% of startups can't afford to make financial mistakes. We all know that. So you want to be on top of this stuff. And thankfully, Brex has built a killer banking experience that helps you take your dollars further. And they'll protect your cash while extending your startups runway. How do they do all that? Well, they offer the best parts of what you need, checking, treasury, FDIC insurance. And they do this all in a single powerhouse account if you need the best because you are on to something big. A Brex's banking solution is going to get you 20 times the standard FDIC protection through

Starting point is 00:21:40 program banks and industry leading yield on your cash and the ability to send money worldwide at lightning speed, which given how global everything is now is super important. So Brex is used by one in every three U.S. startups, as you know, because you go at the dinner with other founders and we talk about it. So go get the financial stack that founders trust at brex.com slash twist 24. BrexB-R-E-X.com slash T-W-I-S-T-2-4. You know, I think where we're at right now is going to be more lawsuits and eventually the IRS and the government doing a full, you know, exam of what happened here because there is this discussion. of, and I was talking on a private chat about this, you know, with some high profile people.

Starting point is 00:22:33 And I just said to myself, well, what if you did this and you had physical assets, right? So here you have a nonprofit that took a software asset, spun it out into for profit, gave, think, a hundred percent of the equity to the employees or some large percentage. And then the nonprofit owns what? a share of the revenue, a license, unknown. Okay, now let's do it again. You and I start a startup. It's going to be for real estate,

Starting point is 00:23:04 and we decide we're going to build a bunch of Airbnb, you know, small, tiny homes on a ranch, let's say hypothetically. And then we get all these people to donate and they get to get tax donations against their gains. Maybe you and I have some gains. We donate it. We get the tax gain and then we spin it out.

Starting point is 00:23:23 And then you and I become employees of the spinout and get equity in it. Yeah. After it's appreciated in value, but we didn't take the risk. Like, this seems like a really good playbook for venture. And in that case, if you donated the actual physical assets, how do you justify that? And here, because it's software, maybe it's a little bit clujy or maybe it doesn't seem as dramatic as giving people a stack of gold bars.

Starting point is 00:23:45 But, you know, seems like a bunch of employees went from the nonprofit to the for profit and we're given the gold bouillon. Yeah. And I wonder what the nonprofit got. That's, I mean, it could be the nonprofit has 90% ownership, 90% of the profits, you know, and this for-profit thing is just transient, but it seems like the opposite from the outside. Yeah, I mean, what's the incentive to give the nonprofit most of the upside? I mean, there's a lot of investors who are putting money into this company. And Reuters said very recently

Starting point is 00:24:15 that as Open AI looks to raise, I think they said $6.5 billion more at a up to $150 billion valuation. everybody's a little bit tired of warrants and foundations and strange orgs. Have you ever seen a Chinese company go public in the U.S.? And then in the F1 filing, they had that diagram of ownership and there's like 74 entities? That's what Open AI feels like. And it's too clever for its own good. So look, I don't want to get into the is Open AI, you know, betraying its early vision and manifesto and so forth. Because that's in the past and the people who are there fight that out.

Starting point is 00:24:51 But I wonder, Jason, and this is not cynical with me, if it's just going to be much better for Open AI to be a regular company and not this Frankenstein hybrid. I love the good vibes behind it, but has it really helped? Yeah, I mean, it does seem like a bait and switch, and here we are. So, you know, I'm kind of tired of talking about it. I'm interested to see it resolved. Yes. As a capitalist, I'd like to understand the playbook. If somebody could write the playbook, I'd actually like to know the attorneys working on it.

Starting point is 00:25:23 So let me start there. If anybody knows which attorneys are doing the work for Open AI, can somebody just leak that to me and my DMs are open? My email for life is Jason at Kalakanis.com. Just tell us who's doing, who are the attorneys doing this? Because I want to know what they're thinking, right? And also what their specialty is, because if they are like, you know, if their IP lawyers tells you a lot, if they are M&A lawyers tells you a lot.

Starting point is 00:25:46 So I'm curious if they have a special specialty. that would tell us what their heads down on. But I wanted to put this to you, because I've been mulling on this $150 billion number for a long time. So the information reported that Open AIS considering subscriptions of up to $2,000 a month. To me, what that sounds like is if you want access to the new model for your company, you're going to pay $25K a year. That seems reasonable enough. That's not crazy.

Starting point is 00:26:11 But how much of your own net worth, Jason, presuming it was all liquid for the sake of this conversation, would you be willing to invest in Open AI tomorrow at a $150 billion post? Zero. Wow. Okay. Tell me why. There's so much potential downside here that there are other ways to place the bet that don't have the downside.

Starting point is 00:26:30 Now, while they are the category leader, they're being priced like the category leader. So, let's say it becomes a trillion dollar company. You made six times your money. Congratulations. Seven times your money, something like that. Yeah. I think you could play other angles here, whether it's Google, Amazon, Apple, Facebook, and maybe get the same result without the downside and the risk in it.

Starting point is 00:26:52 So, yeah, it seems if they're doing two or three billion dollars in run rate, maybe they do a billion and a half this year, that's a hundred times revenue. If they're doing three billion next year, it's 50 times revenue. So maybe you split the baby, 75, maybe a bad analogy for today. But, you know, you split the difference. You split the dad in half and turned them into a husband. Yes. I think maybe 75 times top line revenue while losing, and we don't know what this company's

Starting point is 00:27:23 losing. Is it losing a billion a year or $5 billion a year? I would say $5. You know, and how do you amortize these, you know, what is the lifespan of the H-100s, five years? And then they have to get chucked and replaced and they're like the new ones are that much better. So I do think it's like an absurdly speculative kind of bet that people would make for vanity reasons. It would be like, I think there's a lot of vanity involved in here. If people don't realize

Starting point is 00:27:50 this, but being able to say I was the investor in Uber gives me credibility, Robin Hood, you know, I invested in, you know, Open AI will give you some credibility. Now you talk to the next people. Oh, yeah, we're open AI shareholders. Of course, we have an allocation. Oh, we're SpaceX shareholders. There's like, you can also just buy it on the secondary market. It doesn't mean you came in when it was a $100 million company or a billion dollar company. It means you came in to $150 billion. So these things trade like public companies now. And I think people are using these in some ways because they get management fees.

Starting point is 00:28:23 And there's not many places to put a billion or $2 billion to work. So if you're, you know, Thrive capital and you've got all this money laying around or in Dresen, I don't know who's involved in these rounds. I know if I've had done it before, Josh Kushner's company, like, okay, well, he's getting management fees on the two billion he put in there. He's getting carry. So you put a billion dollars into something. If it only doubles only, you get 20%.

Starting point is 00:28:47 It's $200 million. So for the capital allocator, there's no downside. Like literally no downside. And you get the fees on a billion dollars of 10%. So you're making $100 million in fees. You net that out. You get that now guaranteed. If it becomes worth two or three times, you get that.

Starting point is 00:29:03 three, four hundred million dollar carry, and, you know, your LPs would have been better off putting the money into Facebook and being liquid for the next 10 years. Yeah, I think the vanity point's super huge. I think people really often want to be seen in the best companies and the best rounds. It's brand building for the venture capital firm. And the other thing that I would say, Jason, is if I'm an LP and I give Thrive Capital money, and they put it into Open AI, and I don't get the return that I expected out of it, It's a bit like the old adage about no one gets fired for buying IBM. You probably won't get in that much trouble with your LPs for buying into OpenAI

Starting point is 00:29:35 because of course you were trying to buy into Open AI. Why wouldn't you? So I wonder if it also de-risks the reputational hit of a less than performing fund if you put a lot of the capital into what people expect is going to be the next Microsoft. Though I do want to say that your point about you'd want to make a different bet. It makes a lot of sense to me. There's a new company called Super Maven. Tech Wins just covered them $12 million round, Bessemer lead.

Starting point is 00:29:57 Open AI's co-founder, John put some money into it as well. And it is yet another help people code startup. And if you go into our notes docket and you click the little toggle that says compete with, look at the names that I found that are in this space. Amazing. It's a lot of names. And I just, I'm trying to sort out, is that the better bet to pick one of the 30 courses chasing what is clearly going to be a big prize or do you back the model companies themselves? I don't know where I put my money, frankly. Okay. Are you worried that your startup is the only company out there that's not building with AI features. Well, if your company records

Starting point is 00:30:34 or collects any voice data, I've got a secret weapon for you. Assembly AI has the AI models you need to turn that data into new features that your users are going to love. So don't fall behind. Put this incredible technology to work for you. Whether it's podcasting, audio books, meetings, especially the virtual ones, and all the videos that people are constantly uploading. We're producing and consuming more voice data than ever before. If your startup is building a product that uses voice data, you'd need to check out Assembly AI. Assembly AI build speech to text AI models that can turn your voice data into new product

Starting point is 00:31:14 capabilities. All you need is voice data and a few lines of code. For example, VDIO makes video editing tools that generate captions. That's what you see on all the This Week in startup's clips. That wouldn't be possible without Assembly AI crunching the audio data. Assembly AI's speech-to-text models are simple, accurate and fast, and they've got the industry's lowest word error rate, 30% fewer hallucinations. If you want to put your audio data to work and build powerful AI experiences for your customers, head to assemblyaI.com slash twist and get 100 hours for free and join over 200,000 developers who are building amazing apps with voice data. Go to assemblyaI.com slash twist today. If you had a billion dollars to invest, it would be more work to put 50 million into each of these 20.

Starting point is 00:32:05 I think you would net out with a better result. Yeah. You know, play the index. Now it's a lot of work. And this is where, you know, GP dynamic, the general partners at firms come into play. Place one bet. Go back to Malibu. Meet with the founder every month for an hour.

Starting point is 00:32:22 You're done. billion dollars put to work. You know, what we do, what Y Combinator does, what David does at Tech Stars, you know, like, we do a hundred new investments a year. Obviously, by statistics, 80 are going to struggle and die in the first three, you know, and then you do it every year. You know, your whole life becomes people sending you desperate emails. Ah, we got to save the company. Nothing's working. And, you know, we're basically in hospice care with 80% of the portfolio at any given point in time. And that's a really tough life, whereas other people have the other tough life of a general partner watching public markets go up and down every day and having your LPs

Starting point is 00:33:01 call you, what's going on with Apple? Why is NVIDIA not doubling? Why are you in this? And then you're dealing with, you know, having to get up at 6 a.m. on the West Coast to deal with the markets or 5 a.m. every day. So you pick your poison. And I do think it's really interesting to see how many people are going after coding. That's because of two reasons. Where does a lot of, lot of value in the world reside. And then what is AI good at right now? Bingo. I don't know if there is a better category than developers, lawyers, accountants, right? And we have a great company taxed VPT that went through our programs, then went to YC. And, you know, they're getting a lot of attention, you know, a lot of illegal AI startups because those are constrained data sets that can be

Starting point is 00:33:46 mastered by AI without hallucinations quicker than other things. And then, those professions get paid a $50 to $1,000 an hour. Therefore, you know, if you make them 10% faster, it goes really well. I mean, 10% faster. I mean, I flip it around a little bit and I'm thinking about what if you need 50% fewer developers, you know? I mean, what is the biggest line item at companies that build software? It's humans. Well, and cloud credits, but like whatever.

Starting point is 00:34:14 I am really curious about what this technology is going to do to the arc of the number of developers. There's two ways it could go. One is the number of developers goes down because we don't need as many of them. Or the number of developers goes by 10x because suddenly everyone is one because they can just run the creation system themselves. It's the latter. And I think the reason is if you were to think about typing or spreadsheets, you know, the spreadsheets, when they were done physically, there was a person on like a drafting death. Yes. And they had those sheets with cells in it and they were, you know, writing it and the accountants were doing it.

Starting point is 00:34:50 Oh, we made a mistake, rip it up, start over, and you would go to that room. When I started my career at Land Systems in the 90s, we would go and there would be a typing pool at the law firm. They still had some electric typers and then they had word processing units. And then eventually, attorneys started using keyboards. But those the same attorneys were like, I'm not a secretary. I'm not going to touch a keyboard. That is like, just like we might be like, I'm not going to write code. Get a developer for that.

Starting point is 00:35:16 Now it's just like, I'll write code. It's not that hard. And so... Yeah. So, I mean, but in your analogy, though, then writing code is going to eventually become something that is both considered to be menial humdrum and low skill. Like, you know, typing is a skill for sure, especially back in the days of typewriters because you can't do what I do, which is just spray typos all over the screen.

Starting point is 00:35:38 I'm just, I'm very excited about seeing which of these startups from the great group of them that are working on this does come out as the winner. Let's make sure we make that a category on the Twist 500, like specifically... AI for devs, because that's like worth tracking. I was thinking that this is kind of under the static team size and AI meets reality. Because one thing we did talk about a couple months ago was, hey, you know, not every enterprise generative AI software service is doing well or meeting up to kind of expectations. But here I think we're seeing static team size run into AI in a more positive sense in the

Starting point is 00:36:11 trend perspective. Like I do think this is a place people are having impact. And developers, you know, love this stuff. And that's why GitHub copilot is doing so well. GitLab has one, Amazon has one, and a bunch of startups. So I will make us a new theme to match this. There's a tweet from Michael Seibel that I had been sitting on for a couple of days. And he said, Michael Seibel, of course, formerly head of YC, now just one of the partners

Starting point is 00:36:33 over there, someone that I've talked to a bunch, I like him a lot. He said, too many startups and investors are exploring the B2B opportunities with LLMs and not enough are exploring the consumer opportunities. This is the moment consumer founders have been waiting for. And everyone did SaaS, Jason, forever because it worked. Sell out the companies, make a lot of money, huzzah. Is the generative AI moment, as it is, a catalyst for a new wave of consumer startups like we haven't seen in a while? I'm just curious if you're seeing deal flow along those lines.

Starting point is 00:37:05 You know, you are seeing it in the social media apps like Facebook and Twitter having grok. And I'm seeing people make images inside of Twitter and then post them. You're seeing something similar on Facebook. So I do think the big companies are going to keep adding to it. But yeah, I would like to see more folks look at, say, Yelp and Google, you know, local and say those services, finding local, you know, vendors, finding local restaurants, coffee shops. Now do it with AI. You know, take anything out there and do it with AI. Now you're in a race versus the incumbent to add AI.

Starting point is 00:37:41 but an AI first version of Yelp would be quite interesting. Yeah. You know, where it's, you, you don't show up and start doing bullion searches or clicking off buttons. You just say, I'm looking for a sushi place,

Starting point is 00:37:55 uh, in Austin that's open on Monday nights, uh, that has an Omel Costa menu and show me some great photos and boom. It doesn't, you know, I wonder if the consumer thing here could just listen to you describe this. It feels very personal because if,

Starting point is 00:38:09 you know, I don't care about sushi in Austin. because one, sushi is disgusting. People shouldn't eat fish. It's gross. And two, I don't live in Texas. But what I do want, though, is my own thing that I talk to that has all of my context built into it.

Starting point is 00:38:20 So I wonder if the consumer killer app here is just kind of a personal agent, if you will. And I would love to know who's working on that. But enough about AI. Let's talk about Jason's favorite company in the world, which is Uber and my favorite company in the world, which is Waymo. Because Jason, they are teaming up in a new wave Friday. They expanded their partnership.

Starting point is 00:38:39 I'm going to quote here, they're going to quote, bring the Waymo One experience to Austin and Atlanta only in the Uber app beginning in early 2025. This is big freaking news. The exclusivity really hit my, uh, the hairs in the back of my neck. Because it feels very like a deep integration versus a lightweight partnership. Okay. So important for me to do a couple of disclaimers here.

Starting point is 00:39:02 I don't have any inside information. You know, I know Dara, you know, uh, but I, I don't talk to him on the regular. And so, uh, it was always. Travis's thought that when this happened, Uber would be participating in the technology in some way. They had their own autonomous unit and that if other people produced autonomous cars, they would be the fleet manager or work with fleet managers. Now, why do you need a fleet manager? Well, cars require maintenance. People kuk in the back of cars on Saturday nights.

Starting point is 00:39:35 They need to be cleaned. cars need air pressure in the tires. They need to be washed. You know, there's maintenance that, you know, if they're a gasoline car, they have to have gasoline in them. You get the idea and breaks, you know, all the standard of maintenance. So it makes sense that if there's a range of providers and you've got an established network of 250 million rides per week and somebody makes a great car, they may just want to focus on making that great car. Mercedes and BMW don't run cab companies.

Starting point is 00:40:14 They also don't run rent a car companies. It's enough work to produce a Mercedes or a BMW or a sprinter van. They don't run fleets of sprinter vans. And I'm sure they've tried. I'm sure they've thought about being in Hertz's business. And so what you realize, one business is manufacturing, one is software development, and one is running fleets, the other is running a network. So if you were to take those four circles and you start overlapping them, Uber has a network,

Starting point is 00:40:40 but they don't make the technology. They have investments in a couple of companies. You have BYD makes the technology, doesn't have a network. You have Waymo testing a network in San Francisco and L.A. They don't make the cars, but they make the software. Tesla makes the cars and the software, but they don't have a network. So, you know, and they don't have fleet management. So you start getting into all of this. It feels like it's going to be hard for one person to do everything. It was hard for Uber to try to be a technology company and make this. It's going to be hard for Waymo to deploy this in, you know, a thousand cities around the globe. It's going to be hard for Tesla as well.

Starting point is 00:41:17 And it's going to be hard for Tesla owners. Like if you're, I'm a Tesla owner, do I want my car in the network and somebody puking in the back of it or whatever? You know, I need my car during prime time. Do I want to put my car into the network during prime time? No. I need it available. So the times I will put it in will be when I'm on vacation or. I don't know one.

Starting point is 00:41:36 So I do think there is a lot of complication here, a lot of heavy lifting, and I'll just leave you with often deep partnerships that are exclusives result in mergers and acquisitions. Yes, I can see where that would trend. The interesting thing is Uber historically has worked with people who have their own cars, and Waymo has had to hand build its own technology because you know, if you're at the frontier, you can't go to Walmart and buy it off the shelf. In time, though, once Waymo sorts out more cities, more data, more traffic, I can see them licensing out there, their hardware to someone else to build, maybe Mercedes builds the cars, and then, you know, the aggregation of demand stays on Uber's side. But in that division of labor, I think Uber is the fleet provider, which they're not mostly currently right now. And so

Starting point is 00:42:28 that would have to be a new competency or cost center, maybe, for Uber, but they're saving. So I can And kind of, we're not going to end up with the same companies at the end of this, is my point. They're going to look different because they're going to have to start deciding where to play. But what I care about is this feels like acceleration. Like, this is more cities, more partnerships, more driverless cars out there. Therefore, more data, better models, more safety, more speed. I just want to fast forward this by five years, Jason.

Starting point is 00:42:54 I'm too impatient for the progress here, but it is good. Yeah, the co-CEO Ked to Kedra spoke at Orland Summit last week, and I asked her about these things and their number one goal is to deploy the technology widely. So I said to her, okay, if you want to deploy it largely, why wouldn't you do what Android does, which is open source? She said, open source isn't on the table. I said, okay, if you want to deploy it widely, why wouldn't you sell it to car manufacturer? She said, that is on the table. I said, okay. I said, what about partnering with ridehaling network? She said, that is on the table. And then this announcement came out on Friday later in the week when the stock went up, I don't know, five or 10 percent

Starting point is 00:43:29 based on it. Yeah, yeah. So I think what we'll see, and to your point, like, Uber is not in fleet management right now. Mm-hmm. There's two possibilities here. One is they work with existing Uber drivers and fleets because there are people who own multiple cars and essentially run fleets now and they find drivers and have them use their cars for the drivers.

Starting point is 00:43:51 You know, if they were, let's say, immigrants, they don't have to worry about insurance, buying the car. They just can drive the car, essentially. And so I think they'll work with the existing fleet managers, and those fleet managers out there will eliminate the drivers because they do work with them already. And they'll just deploy the cars and clean them. If you look at the economics, Uber takes 25 percent, generally speaking, the driver gets 75 percent. So I'd love to see the economics of this deal. Let's say they give, you know, Waymo might be very happy to get 50 percent, 60 percent, 75 percent of this and not have to deal with all the other stuff.

Starting point is 00:44:28 they could just focus on software, making the cars, and then let's say there's a who's going to pay for these cars. Waymo just said they're going to raise $5 billion. Yeah. Uber's a $150 billion company. $5 billion. I think that buys like 50,000 cars. It's a lot of vehicles.

Starting point is 00:44:44 Yeah. Yeah. So 50,000 cars, you know, if they're doing, you start doing back of the envelope math, you know, okay, if they do 10 a week, if they were to do 10 rides, 50,000, 500,000, if they were 100,000, if they were 100 rides, that's $5 million. Okay. maybe they could do 300 rides a week each. That would be a lot.

Starting point is 00:45:02 Maybe 400 rides a week. So you start to realize that maybe with that $5 billion investment, they can do, I don't know, 5%, 3% of what Uber's currently doing, which is another way of saying, you're now going to need $100 billion in cars. And then somebody's got to manage a fleet of not 50,000 cars, but, you know, 50 million cars? maybe 5 million cars, as this thing grows from less than 1% of people using ride chair into 20%.

Starting point is 00:45:33 So if it goes 20%, it goes 20x. I mean, we're talking about a large number of vehicles. And even if Tesla's entire fleet was in the network, they might have 7 million cars. I think they've produced to date. Yeah. Let's just say they produce 2 million cars a year and they have all 7 million in the fleet. If they're producing 2 or 3 million cars a year, they would need decades to fulfill the demand. So that's, I guess, the big challenge here is I don't think one company gets there alone.

Starting point is 00:46:01 I'm not just talking my book. I've sold a lot of Uber, you know, before it was public. I still own a lot. I could sell my shares anytime and just buy Google shares or Tesla shares and play the other side. I actually think the pie is going to go from less than 1% to 20% of rides, either induced or replaced public transportation or new rides is going to go this way. And I don't think young people are getting driver's licenses. So it would be a decade journey. I don't think so on the driver's license point.

Starting point is 00:46:28 I agree. Also, just as a parent who lives in a city, I think the first place we should get rid of human drivers is the small side streets, just because people drive like absolute idiots and they're distracted and they don't watch street signs, traffic lights, turn signals. It's brutal, just trying to keep my kids or mean hit by cars. So I would love that. All of this makes me very excited. The thing is, I was leaning towards as you were talking, saying, oh, okay, well, think

Starting point is 00:46:53 about it. If everyone needs to have a new fleet, it's going to be larger, two companies, GM, Tesla, because GM has cruise, Tesla has its own self-driving. They're already manufacturing. But even then put together can't make enough cars in the near future to come up with the amount of capacity we need for our relatively optimistic view of where this is going. So there's like a $100 billion or $200 billion in spend missing that someone's going to have to come up with.

Starting point is 00:47:17 And I think go back to an earlier point, Waymo, Uber, whoever it's going to be, someone's going to have to write that check. And you know who can afford that? The federal government and big tech. That's it. Yeah. And Uber is part of big tech. You know, Uber could at any point in time save the public markets.

Starting point is 00:47:34 We're raising $10 billion or do a bond for $10 billion to buy these cars from B.D. And even if BYD has a 100% tariff on it, a $20,000 car going for $40,000 into a ride-chairing network, still pretty great deal. So I... Absolutely. I think that's the thing people... People like to think of the world as winners and losers. If we were to look at e-commerce, Amazon and Shopify, both crushing it, and then other legacy people, Target, Walmart, still crushing it, Instacart, still crushing it, DoorDash, Uber eats, crushing it. Toast is crushing it. So I think what we're going to see is, as dynamic as Uber has been and Lyft and DoorDash for the first, for the last, you know, 10 plus years in terms of our imagination of how the world could change.

Starting point is 00:48:22 I think that's like the epilogue to the story. The autonomy endgame is upon us. This is the beginning of the autonomy end game. We will be sitting here in 10 years and the number of people who have cars is going to have dropped significantly. Old people are going to stop driving completely. This idea that 70-year-olds have to drive, that's done. People over 60 are going to stop driving because they have money and resources.

Starting point is 00:48:48 So they're going to move fully autonomous. And then this next generation are kids. You know, when your kids start, you know, in 15 years driving, my kids in three or four years and 10 years, respectively, you know, they start driving in the next decade. They're not going to drive. No. I mean, I'm going to teach them to, but they're not going to need to learn. And they're certainly not going to spend all their hard earned cash buying a car. Why would you?

Starting point is 00:49:11 I mean, it's a depreciating asset that you use 3% of the time. Yeah, totally agree with you that. I'm just giggling because I learned how to drive my dad's F-250 on logging roads in Oregon. and big manual V8. And I used to kind of think about, you know, I'm not going to teach my kids how to use a manual transmission and wrong question.

Starting point is 00:49:29 The question is, will I teach them to drive at all? Because our new daughter's two weeks old, which means that she is 16 years away from the legal age of driving in the U.S. And 16 years is five eras in technology? I mean, that's a long way from now. It's a long way from now.

Starting point is 00:49:43 Long way. And, you know, there's been, I think Uber is setting itself up to be a platform. They have B-YD buying 100,000 cars and putting them out there. They've got Wave AI. They just put money into last week. Cruise, their restarting cruise.

Starting point is 00:50:03 Cruise is going to, NGM is going to be looking to Uber to run that side of the business. Ride Alto and now this Waymo deal. And they've got a couple of other partnerships. And they've also invested in a couple of companies. It's like Aurora, I think, is the spinout that was their autonomy. And why did they close the, I mean, there's a question for, our live audience here. And we go live on YouTube when we do this news programs.

Starting point is 00:50:26 You can type in this week and start up and subscribe and hit the bell. You'll get an alert when we do. Why did Uber stop its autopilot program? They were doing too much. And when Dara took over, he said, we're going to focus on a small number of things and do them really well. Part of that was culture. Part of that was regulation.

Starting point is 00:50:44 And then part of that was dealing with legality around independent contractors, getting back on the road in London, you know, selling their shares in different regions where they weren't going to compete. So there was a lot on his plate. And, you know, I think they did the right thing by becoming a money printing machine now. The only person who can really go it on their own is Elon and Tesla in my mind. I don't have any inside information there. But, you know, I do think he shared an app and he has a unique ability. And they're going to be announcing on October 8th, I believe. The rumor is they going to show like a two-seat dedicated car that will go into production in the next couple years that they could flood the market with. That's my

Starting point is 00:51:22 prediction. They could flood the market with a two-seater that cost $25,000, but only make it for Robotax. You can't buy that car. Oh, and that protects their gross margins on their actual core on their core business or their existing business. Okay. You still have to buy the Y. You still have to buy the Model 3. Yeah, yeah, yeah, yeah. And S, you can't buy this two-seater car, even though there might be demand for it. That one, they produced with, you know, bare bones, two-seater. with stuff on the trunk. If you got four people with you, just take two of them.

Starting point is 00:51:52 Yeah, why not? They're electric and they're cheap. Like that's, ah, it's going to be fantastic. I will say I have like a smart car. You know, you know what my neighbors have?

Starting point is 00:52:02 A smart car. You know what their kids have? Imagine smart car. So often they park their two smart cars back to back. It's the funniest thing. It looks like two jokes. It's just kind of come together. I like the Mini Cooper better.

Starting point is 00:52:11 That was my first new car ever bought in my life. It was a Mini Cooper. The second year I came out. I love the Mini Cooper. Wow. So that would be, what year was that? Gosh, it had to be 2004 or five. Yeah, yeah.

Starting point is 00:52:23 Have you ever been in an original mini, the old British? I have not, though. I've seen them on the streets in London and in Paris where, like, parking is at a premium. Yeah, those things are real small. You think a mini is small, stack it next to that. It's very different. Now, thinking about things that might be getting smaller, the interest rates might be going down this week. Jason, we are finally on.

Starting point is 00:52:41 And I, we're going to get this quick, everybody. I'm not going macro on you, but it is rate cut week. So, Jason, we got to get some predictions on the board here. 25 basis points or 50 basis points? What's your prediction for this week's Fed cut? I mean, think the prediction markets are saying there's a decent chance, 50 to 70% chance of a 50th cut. I think that might make them look a little bit desperate. Oh, okay. Maybe they waited too long. I am a fan of 25, 25, 25, just consistently doing 25 while you get more information.

Starting point is 00:53:15 In other words, they were late to raise rates. And that kind of created a bit of chaos. Obviously, we saw with Silicon Valley Bank and dated bonds and treasuries and everything. So I don't think they want to flip the market here and look panicky. Just think like every month, 25, get more information, look at the employment picture now. I do think politicians, seemingly Gamala, would love a 50 or 75 because what happens when you cut rates? Stock stock market go up. Stocks go up.

Starting point is 00:53:45 I don't care any more about this stuff because I just like the idea of predictability and in markets and thoughtfulness, whether it's politicians or our monetary policy. I like predictability. I think they get that they need to be predictable and independent of administration. So people are saying like this was all done in some grand conspiracy. I don't think so. No. I don't think they wanted that whipsy.

Starting point is 00:54:15 saw that happen with Silicon Valley Bank. Also, they would have, I mean, let's be honest, if Kamala and Biden were secretly running the Fed behind the scenes, interest rates would have come down already. Do that in the second quarter. And then you go into the debates with a roaring stock market. You go into the third quarter. And then all of a sudden, it's November. And hey, look, it's been six, all people know is six months of going up.

Starting point is 00:54:39 Instead, they've been dealing with the highest interest rates we've seen in, you know, a generation. I'm Team 50 just because I think it's time. I think we're a little bit late. The labor market's gotten weak enough that I think right now a shot in the arm is more important than the optics of it. But I do very much agree that no matter what this first one is, either a treat or just an expected snack, consistency moving on us forward. I think we should get down to like 3% next year. Three, three and a half sounds pretty good to me. So a smooth path there.

Starting point is 00:55:09 And it would be good for exits. It'd be good for IPOs. if you go for stock market prices, startup valuations. Like, this is going to be a holistic good. And maybe it's going to make next year a lot more fun than this year for folks in your profession, Jason, the investment community. I love my job and I do it independent of this hand-wringing. And if it's a little bit hard and there's headwinds, I consider that part of being a pilot and sitting in the seat that I'm lucky enough to have. You know, if you're a quarterback and you get sacked once in a while and you eat a little dirt, well, you also get to be the quarterback and throw the touchdown pass.

Starting point is 00:55:42 And so I take the go with the bad. This has been brutal the last three years with no DPI coming out of funds, no M&A, Lenacom. We've talked about it here over and over again. And this crazy political cycle, just like it to end the political cycle and the whipsawing, stop steering the car all over the place and slamming on the brakes, slamming on the gas. We don't need to ever go back to this zero interest rate phenomenon as far as I'm concerned. It's healthy to have people be able to make a little bit of a little bit of a little bit of

Starting point is 00:56:12 of money from their savings and understand that savings is virtuous. And, you know, it obviously, I think puts a little pressure on my profession and other people's professions to perform above whatever that number is. Yeah. The risk-free rate. And so I like that. And when it's zero, it's like, well, where else do I put money? I have no choice. It's got to go into real estate, it's got to go into equity, and venture. There's no way to put it if you don't have that. So let's Keep it at three and a half, like you're saying. Let people make a little Vig on their money and compete for those dollars, those investments. My favorite email that I get every month is when my money market accounts give me free money. And I'm like, thank you for the free money. Because when I started building a real emergency fund, interest rates were zero. So you put in any amount of money and they flipped you a quarter at the end of the year. We're like, be grateful. And that was pretty tough. You know what I did?

Starting point is 00:57:06 I was looking at the price. I stayed at the Beverly Hills Hotel. My friend Freeberg sets us up there for the summit. Nice. A bit pricey. And I'm at the pool. It's quite a nice place to say quite charming. And it was $50 for a peanut colada, $42 for chicken fingers.

Starting point is 00:57:23 A Frappuccino was $25 for like a frozen coffee beverage with no booze in it. Why? Because. And then I was at Shudder. It was my favorite hotel to stay in Santa Monica. I moved out when it was on my name. not the conferences. I moved to shutters, which is still a pretty penny. Don't get me wrong, but it's like a third of the price. And it's on the beach, my favorite hotel to stay at in on the west side. And you know what I did? I ordered Uber Eats for every meal. I didn't order room service once because the room service bill is literally two or three times the equivalent and you get better food.

Starting point is 00:57:55 Yeah. Uber Eats. And this was the first time a hotel was not passive aggressive and sending it up. All these other hotels meet you come down. as punishment for ordering not from the room service menu. And I'm like, can you send it up? I can't send it up. You have to come down. I'm like, but you can send me up a shaving kit. A toothbrush. Yeah.

Starting point is 00:58:14 A tooth brush can come right up. So like when you stay at like a mid-tier hotel, they won't send it up. And I understand not sending the Uber Eats or DoorDasher into the hotel randomly. But I mean, you got people. And then I give them five bucks. Still cheaper. I get to try all the great restaurants. It was wonderful.

Starting point is 00:58:30 So shout out to my Uber Eats and DoorDash. I just, I feel. Whenever I'm getting charged an insane amount of money for something, I'm like, you have me in some sort of armed twist. Like, if I'm at a concert and I need to get a bottle of water, I understand that I'm going to pay $7 for it. And I'm going to be furious, but I'm thirsty. But when there's options, man, I'm not, I'm not paying 50 bucks for a frozen coffee. That's going to be middle at best. Well, and Travis at the All In sum, we did release the video.

Starting point is 00:58:56 He showed one of the things they're working on with some members of the former autonomy team at Uber, went to Cloud Kitchens, randomly. They just applied. I guess they got in. And they have this new machine. You fill the machines up, tomatoes, lettuce, chicken, barbecue chicken, steak. And then it, you know, does a little swirl. The bowl comes by and it puts everything in the bowl. Then it puts a little sauce on it. And then the bowl goes into a bag. And then the bag comes out at the end and it has the label on it. And the driver just picks it up and there's no intervention. So like, you know these bowls you get in Australia bowl culture? Yeah.

Starting point is 00:59:38 So here it is. We're showing it from the All In Summit. And you can see, like, they made their own hardware to do all of this, you know, purpose built. Those are the containers. And so this is, this means fresh, healthy food, 24 hours a day. And you're going to take out, I don't know, if there was a back order, you could be taking out a half hour of time, right? because this thing can just churn and you see how many days it has there.

Starting point is 01:00:06 It looks like some crazy 3D printer and it's not printing the food, but they 3D printed all these parts to make it and there goes the cardboard through the machine on a conveyor belt and right into the bag, boom. So this is coming to Uber Eats soon and they're going to be able to make you a breakfast ball,

Starting point is 01:00:26 a lunch ball, a dinner bowl, whatever it is, a salad and you're done. And the folks over at what's the salad place? Sweet greens. We had the CEO on the pod. Yeah, they also have this technology they're working on. So, Brave New World, autonomy.

Starting point is 01:00:43 The autonomy end game is I think what I'm going to dub this theme. Everyone a dev and the autonomy endgame. I got to take notes, everybody, so I don't forget. It's in the show. We got it in the show. We make a lot of content here at Twist, Jason. It's good to have notes. That way I don't have to scrub through my own video.

Starting point is 01:01:00 One of my favorite parts about you is that you take notes That's why you're a great journalist Obsessive about it Because my brain currently Especially because I have a newborn It's made of Swiss cheese Baby brain

Starting point is 01:01:11 You gave me an assignment On my first day back Which was here is a Mr. Beast leaked production document It's 37 pages Summurized that and have it ready to go For the show so I did I got to read

Starting point is 01:01:23 Lots and lots of Jimmy Snack off Yeah I honestly Honestly Jason There's a lot of you in here It was kind of funny because he's very clear about what kind of person he wants, how he wants them to communicate, and what people should be obsessed with. And also, there's a lot of clarity of goals. So if you read this and you want to learn how to make YouTube videos, you will learn how to make YouTube videos. But more it struck me as how to run a high performing mid-sized company that has successive recurring deadlines. Like this is a company. that has to ship by definition because they are producing something that it's consumed and then loses value. And there's a lot that I really like. So I'm going to just call a couple things out here.

Starting point is 01:02:10 What is the goal of someone joining the Mr. Beast production organization? Really simple. One sentence, your goal here is to make the best YouTube videos. And he goes on about this. He's like, look, we are not Hollywood. We're not trying to be Hollywood. We are trying to be the best at YouTube. That's it, full stop. So clarity was huge in this document. I really like that. And then, And for example, another one, what is the goal of our content? One sentence's answer, he goes on, but he says to excite me. Hmm. I like it.

Starting point is 01:02:37 Yeah. Haste. Well, it's, and also, if you are building for everyone, you're building for no one. But if you're building for Jimmy and his sensibility is, I want something exciting. I want something that snackable, like a 20 minute or 10 minute thing. Okay. And it's a YouTube thing. And, you know, one of the great things about YouTube is,

Starting point is 01:03:00 because it came from Google, and they took it over so early, it's a metric-driven platform. So you can see how you did in the first 24 hours of a video. You can see when people drop off. And so everything for him, and I've talked to him about this personally. I've played cards with him a couple times,

Starting point is 01:03:16 and he's a fan of Wallin Pod, and some of my friends are involved with his business. So he believes, like, you've got to get that first minute right. You've got to have a bunch of edits in the first minute. once you pull people in, then there's like this setup for the first next couple of minutes. And it's very much like the three-act screenplay in Hollywood. And, you know, these screenplays were kind of built into a science for our generation.

Starting point is 01:03:42 They backed into what makes, you know, I don't know, Silence of the Lamb, Star Wars, you know, pick an iconic movie, Rain Man into that. And it's like act one, act two, act three, the resolution and the journey and, like, like the problem solving in the middle act and all that great stuff. So he's kind of produced, I guess, a playbook here that really works for him. Now, if you're not doing stunt reality television,

Starting point is 01:04:10 I'm not sure this is applicable to every vertical, but it's certainly there's lessons in there. Any other lessons you took from it, Alex? Well, clarity of how to get good from him. Quote, get rid of Netflix and Hulu and watch tons of YouTube.

Starting point is 01:04:24 Like, don't go drink from the, pond we're not, you know, fishing in. Only double down on this. I get back to the clarity and focus point. Notes in there about communicating how to communicate inside of an org could go up, then over, don't go over and then up. How many people who are having teams know what they're doing.

Starting point is 01:04:42 There was a big thing on creativity, saving money, which I think is his way of saying that constraints breed solutions essentially. Yeah. The example was about Doritos. Like if we say we're going to give away $20,000 to the winner of a video, no one cares. But if we say we're going to give them a 10-year supply of Doritos, it could actually be less

Starting point is 01:05:04 money and it's way more exciting. So creativity, saving money. And then I love this at the very end. One of his last kind of like notes was never do anything that can make us look bad from a PR perspective, which is not a bad thing to have in your document. And they've had challenges like doing these kind of, all the YouTubers have had challenges around, you know, when you're doing stunts or you've got contestants. Fear Factor had this issue.

Starting point is 01:05:29 Everybody's had this issue. All the reality TV shows, and he's got that reality TV event. You're going to have a lot of issues. And so I think that's very wise. Yeah. Do not do anything that would embarrass the organization. I want to throw one more thing in here, though,

Starting point is 01:05:45 now that I'm just thinking about it, this is not a polished doc. There are typos. There are, he writes, ha-ha at one point in time. Like, there's a lot of just very, it's like he sat down and literally wrote you a letter. And that's, I think, actually why it's very effective.

Starting point is 01:06:01 Because if it had been McKinseyed or BCG'd, you know, it would have been impossible to understand. But here is a very, quote, quote, founder mode document from someone who knows what he wants from his org and how he has gotten historically. And I just, I think that people can, people mistake polish for quality in writing. And I think here we have the. inversion of that. The polish is crap. The writing and the actual information transfer, Jason. And so that struck me as impressive. I'm not shocked that he's a smart guy, but he's smart.

Starting point is 01:06:37 Yeah, I think this quote is the one that's worth pointing out. You kind of previewed it here, but I'll just put it on the screen for a second. Your goal here is to make the best YouTube videos possible. That's the number one goal of this production company. This actually relates to the purpose-driven church, a famous book where it's like, each church has to have one thing they do, not everything. It's not to make the best produced videos, period, not to make the funniest video, period, not to make the best looking videos, period, not the highest quality videos, dot, dot, capital, it's to make the best YouTube capital videos possible. Everything we want will come if we strive for that, who's you sentence. It sounds

Starting point is 01:07:19 obvious, no comma, but after six months in the weeds, a lot of people tend to forget what we're actually trying to achieve here. So this is a very like almost like somebody transcribed him in a meeting or having a drink after work. And I think this is critical because when you have an organization that gets bigger, you want people to point to something. And you know, I tell that to folks here at our investment company. We want to do the seed or pre-seed round of companies and we want to help them get to their Series A. That's the goal. We want to find the best companies with technical co-founder, multiple technical co-founders and product velocity and get those folks to release a product, get traction, and get follow-on funding, or what's called to pull through in our industry. And we track pull-through. 33% of our companies in the first year pull through to another round or so. You know, just ballpark figures here. Sure. And so if they're not pulling through to another person investing at a higher valuation from us, what did we do wrong is what we have to ask ourselves. Did we pick the wrong team, the wrong idea? Did, uh, Did this team not achieve product market fit?

Starting point is 01:08:28 Did they know what they needed to achieve in order to get another investment? Did we not introduce them to the right people? Did we invest in a category that VCs don't want to invest in, like event planning, you know, apps like what we're going to do this weekend or, you know, split the bill apps, you know, things that are maybe not going to be large businesses, not enough TAM? So we know why people will not invest. The team's not technical. The team doesn't have product velocity. The team doesn't have product market fit. The team is not going after a big enough market.

Starting point is 01:08:59 So we should be asking those questions early, and we should be communicating that to the founders. Hey, we're going to have you come to our accelerator. We think the downstream investors are going to think your TAMs too small. What's our strategy there? And then we just hear their answer. Yeah, we're starting with this. It's our beach head. And then we're going to expand into this larger beach head and we're going to go internationally.

Starting point is 01:09:19 And that makes it a billion dollars in revenue a year. So, okay, good enough for us. Let's go. Yeah, clarity of vision. I mean, I would boil down your launch proposal there too. We help the next $10 billion tech companies reach their series A or whatever. But I think you could play gel it even more if you were doing a similar document, but just knowing what you're doing. And this is, I think, where we see rot at large companies, because if you think about like what Google does as a company, it does like 55 things. And inside of it, everyone's vying for

Starting point is 01:09:50 probably a bit more like self-focused like progress inside the org versus like the goal because I don't think you can have 50,000 or 100,000 people sharing the same goal because you can't coordinate work like that. I would love to learn more about like management and systems theory to understand why big companies end up always looking and feeling and acting the same in ways that I don't appreciate. There's probably something here. There's a whole segue into founder mode. Yeah. Wow. But given that you've got a kid right now, I don't recommend you getting into founder mode. No, dude, I have two founders at my house and I am the thing that has been mo-ded.

Starting point is 01:10:31 Yes. There's a lot of, okay, one little story. Have you seen the cars that children can sit in, their plastic and they have a handle behind them? You push them around? Yeah, they're like strollers. Yeah, we didn't have one. And now we do because our nanny's former family outgrew it. So we got it as a hand-me-down.

Starting point is 01:10:47 Love hand-me-downs, by the way, as a parent. People give you the best stuff. Anyway, this car is terrible because it has plastic wheels. And Ada demands that I push her around the neighborhood constantly. And it's the loudest thing. And it sucks. So I've been bullied into that by a two-year-old. So that's what I've been doing.

Starting point is 01:11:03 There's one of those that has two seats that you'll be doing soon enough where ones in the front, one's in the back. And we had twins. Oh, we did the side-by-side double stroller. Not the, not the, not the, yeah, I know you're talking about. We were awesome. We bought tons of stuff. And then we just put it all out in the driveway and our friends who were having kids. We said, this Sunday, everything must go.

Starting point is 01:11:27 Yes. Take it on Sunday or it's going to donations. People came. God bless them and their families, you know, didn't have to buy this stuff because you use it for two years and it's done. Yeah. People as, some of our neighbors had us over for brunch and literally had a room set up of, we are going to donate everything you don't want.

Starting point is 01:11:43 But we have two daughters. You're having another daughter. Come pick. And so we left with like, you know, loads of stuff. Because, I mean, newborn onesies, you don't wear it for two years.

Starting point is 01:11:53 You wear for two months, maybe. I love the idea of like, we put all of our 14 year olds clothes as she got older, put her age and her size, put them into plastic tubs, put it in the garage. Then the eight year olds now,

Starting point is 01:12:08 when they hit her age, we take out the box and they've raided. And it's like, oh, this is cool stuff, right? Yeah. So it's just absolutely fantastic.

Starting point is 01:12:16 Otherwise, you spend all your money on parenting stuff. But Jason, We are back to doing news live. So next week, we're going to go back up to two. I'm going to quickly get back up to speed here. Awesome. Love it.

Starting point is 01:12:26 It's great to be back, man. I miss doing this. Good to have you back. And, yeah, next week we'll do two. And in the meantime, lots of great interviews. Twist 500 is growing. And you're doing those short ones. Let's keep doing some of those short ones.

Starting point is 01:12:40 And we'll see you all next time. Bye, bye.

This Week in Startups - OpenAI's new models, the MrBeast leak, Waymo ♥’s Uber and more! | E2009

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.