Jason Unpacks Sequoia’s New Funds"| E2199

27:52And he was interviewing Barry Diller, whose book is amazing. And he was asking him if you could put all your chips in, like he did for QVC and make a bet here, what would you bet on? And he said, things that are alive that cannot be disrupted. So going to something in the real world, like an MGM resort, a casino, to go see a show, that's not being disrupted by AI, that's what I would invest in. I think this falls into the live entertainment category where live entertainment is going to be huge for these, and people will start doing them on the regular. I could see these being franchised everywhere, and I think that's Kimball.

28:28My friend Kimball Musk has a drone company that's a pioneer in the space, and I think that's what they're pursuing is these things being like touring versions of the Harry Potter show. So you can tour them around the United States, go to every single fair, and then people buy tickets to go see a drone show. And now imagine a drone show with a live orchestra, like they do at the Greek Theater in Los Angeles. Sure. They'll have, you know, they'll play Gladiator or Goodfellas or The Godfather and put the symphony orchestra to play the soundtrack live, which is really cool. Oh, here's the Vatican.

29:06Yeah, yeah, yeah. Check this out. So this is from, I think, Kimball's company. and this is the confluence of Christianity and modern technology. A famous combination. This is not sped up, in case you were curious. Beautiful. Yeah, so there's the Vatican with this drone show, and then I guess they're playing, I would assume they're playing the music live. Gorgeous, yeah. I mean, there's the Pope. Pope is dope. Tesla is training on synthetic data in its FSD regime. This is a video that Tesla shared on X recently, Jason. You wanted to take a look at it. So here it is, and we'll talk about it on the other side.

29:47Hey, startup founders. Are you tired of losing deals to endless procurement cycles and complex contract negotiations? There's a better way to sell your SaaS product. AWS Marketplace isn't just another sales channel. It's your shortcut to enterprise customers. When you list on AWS Marketplace, your software charges appear directly on your customer's existing AWS bill. No separate invoicing, no procurement headaches, no waiting months for approvals. But here's the best part. AWS actually incentivizes buyers to purchase from your listing with credits, helping you land your first 20 customers. And after that, AWS provides you with a budget to fund even more deals.

30:26Plus, you get one-click deployment, free trial capabilities, and access to committed spend budgets that enterprises are already planning to use. So stop watching your competitors close deals faster. Join thousands of successful startups already selling on AWS Marketplace. Ready to accelerate your sales cycle? Visit AWS Marketplace today and turn your biggest procurement obstacle into your biggest advantage.

30:55All right, so I guess we're seeing nine squares here it's a i guess this is all the cameras on the car maybe it's eight actually yeah the new tesla hardware stack is eight cars and this looks incredibly real if you showed this to 100 people 100 people would say that's real so they now have uh the ability to build situations for the car to navigate and then test it over and over and over again which is extraordinary because what are the chances that like, I don't know, a deer comes running out into the street in this, or a dog, you know, it's like to capture that on video very hard. But I would say, sadly, I know this from living in the hill country in Texas, like there's a dead, I always know when there's a dead deer, because you see all the hawks and everything circling.

31:49And sadly, there's a deer dead on the every day somewhere I'm driving. Like, you'll see at least one every day or two. So they could simulate all this. I think it's gonna get them closer and closer. They said they believe FSD has been solved. Of course, there's the natural criticism that Elon's been saying, hey, we're very close for a long time. But, you know, he was very close at landing rockets, you know, and then they did and had reasonable rockets. So he's often late but never wrong is what we say when people ask me for a quote always yeah you know frequently late never wrong is uh i think the track record so far but um i think this is this is freaking awesome and just to be clear to double click it in case you're watching the video version none of that tesla video was real it was all synthetic or as tesla put it video below is fully generated and not a real video but i love this idea jason we talk a lot about how you know once we get self-driving is going to save lives here's a way to accelerate our timeline to that point fantastic uh i did a little more research though i was curious if anyone else is doing this there's one or two other examples not that many to be clear uh but wave w-a-y-v-e we talked about them on the show a couple of times they just raised a few billion dollars they made something called the ghost gym back in 2023 seems to be a similar idea they said quote simulation supports large-scale synthetic data generation allowing us to leverage thousands with diverse driving scenarios blah blah blah And then Wabi, a company that I had totally forgotten about, but it's working with Uber on self-driving semis, is also working on a thing called Wabi World, which is a closed loop internal simulator that's powered by Generative AI.

33:28So companies are really putting this to work. And I'll just say, people asked what the use case was for AI 24 months ago, and it seems that we've stopped asking that question. But here's one more example of what we're doing with it. Yeah, Wave is probably underreported on. They have tests going on in UK, which is where I think they're based. And San Francisco, they've been testing. Germany, they've been testing all over the place. And I think what's going to happen very soon is we're going to see the commodification of self-driving software, and it's going to be built into the NVIDIA hardware stack and built into every car.

34:04Now, that's going to take a little more time, but there was an underreported story. Everybody knows there's a relationship with Uber and NVIDIA. just last week this underreported story you can pull it up um about how nvidia and uber have revealed a little bit more of what they've been working on together they're taking all the uber ride data they're giving it to nvidia as part of this partnership to be part of their thor stack they have a hardware stack that is doing self-driving so the uber team here uh uber teams up with NVIDIA to accelerate autonomous mobility. And you can find the paragraph where they talk about what they're doing, but this is going to be, I think, and you can read it for the audience.

34:48So millions of trips occur every day on Uber. It's a lot of data. They're going to pull that into the NVIDIA Cosmos platform and the NVIDIA loves brand names, the NVIDIA DGX Cloud to help AV partners build stronger AI models. The idea here, Jason, is that this is a platform that has state-of-the-art generative world foundation models. And so you can run around in a car inside of them and what do you need to build them you need lots of data what does uber have lots of that actually you know you talk a lot about you know tesla having cars on the road so many of them giving them a data advantage yeah uber has a lot of cars on the road too this could yeah make you build a more even playing field you know just obviously people will claim i'm like talking about my friend or talking about like an early investment of mine and they're like trying to figure out my motivation here i've said the same thing consistently for a decade i think there's three or four players who are going to get there all around the same period of time.

35:41So for everybody like, oh my God, Tesla's never going to get there. Oh, they're, you know, they missed all these different benchmarks over 10 years. They're going to get there. It's obvious. They are going to build something that hopefully, this is the dream, this is the objective, will be able to navigate things it hasn't seen before better than the people who are building perfect models like Waymo. Waymo is also doing this, you know, language model kind of version of this where it just reacts to what it's seeing. And then you have all the different people who make cars. So there's like four or five pieces to the stack.

36:21There's like the data and the cloud and your algorithm. And like the data is one piece, the algorithm and your system for doing this, your model for doing it. Then there's the hardware stack on your car. Then there's building a car itself. And then there's running a network. There's like many pieces here. And what we've seen is the pieces are, there are people who are specialists and there are full stack people. And then there are people who have two or three pieces of stack. And then there's the giant Uber network, which is betting on all of them. And NVIDIA, which is kind of betting on all of them and commodifying it.

36:51I think self-driving becomes commodified and then it's going to be a deployment issue and a regulatory issue and oh my god don't have an accident issue tesla's rightfully being very cautious as they should be and they have permits to run ride sharing in 10 cities and as i said here on this program many times before just do a year with the safety driver prove that you can do 10 000 rides 100 000 rides without any instance share all the data publicly waymo should share all the data publicly the regulators should share it all publicly because really what we have to do is convince the public that when a random death occurs, it's not a reason to shut all this down.

37:34That will only happen with transparency and cautious rollouts. Can I tell you why I'm not that worried about that, Jason? Because everything you're saying, totally valid. We had a friend of ours come visit the other week. She is a physician. So far, far, far away from our world. Doesn't care about xai versus open ai doesn't know who any of that stuff is right she was raving about waymo in la just raving about it she's a complete convert and is never going back and she is a normie so to me i think that as customers touch and see these cars in the real world get to take a ride on them it changes minds so i think you're dead on but i'm less concerned than i was six months ago i'm also we ride is the other one um and they're part of the uber network but i think you know a lot of these companies are going to be part of every network.

38:25Like for WeRide, why would they only be with Uber? They'll be with Grab in Malaysia and Singapore. They'll be with DD in China. They'll be with everybody. But WeRide and Uber have launched their robo-taxi in Riyadh. I'm going to be in Riyadh next week, Monday, Tuesday, and Wednesday in the UAE after that. And both of those places have these services operating now. So I'm hoping that my team - Can we do a twist from you inside of one of those cars? Literally is about to say that. Producer Marcus, let's get on it. If anybody's listening, I need WeRide contact info and then whoever else is in UAE, and I'd like to jump in those cars.

39:02And Waymo published just their incredible safety data. I don't know when they published it, but it was a couple of weeks ago. Miles through 2025, they've done 16 million in Los Angeles, 30 million in San Francisco, 46 million in their first market, Phoenix, and 3 million. And as you can see, 91 % fewer serious injuries, 79 % fewer airbags deployed, 80 % injury causing crashes, and crashes involving vulnerable road users. 92 % fewer pedestrian and crashes with injury. 78 % fewer instances with cyclists and 89 % fewer with motorcycles. So you have to be saying, well, wait a second. Does that mean they've hit a human or they've hit a motorcyclist?

39:59You know, all of that data. What it means is typically motorcycles, tragically, there was a motorcyclist that hit a Waymo and died last month. Nobody talked about it. I think we were the only show to ever talk about it. Why? The Waymo was stationary and it got hit. So the press is even like, that's not a story. Now, when Teslas do this, some motorcyclists will run into a Tesla. I guarantee you that they will be treated much different than Waymo. Why? Because Elon got involved in politics. And so there is going to be like a much higher - There's another reason for that though. Yeah. There's a more cynical reason.

40:34You're applying a lot of intent to that. One of the reasons is that Elon Musk is one of the two most clickable terms in the world. and news business does have a business element to it. We had a joke at TechCrunch in like 2014, which was put iPhone in the headline, just somehow just weasel it in there. Get the iPhone in there. And I joke, but I do think that there's a positive feedback loop between covering Elon Musk, people getting mad about it, driving more attention to it, which makes editors want to cover it more. So it's kind of one of those things. You're right too, but don't forget the money part of it.

41:07I think that, yeah, it's going to be one in the same. There'll be people who do it because right now, if you look at the number of journalists who identify as Republican, it's like 3 % or 4 % now. It was 30 % Republican, 40 % identifying as like, or it was maybe 35 % Democrat, you know, 27 % Republican, and then 40 % or so independent. Okay, great. That's how it was in the 70s and 80s when I was growing up. Now it's down to only 3 % of journalists report that they are conservatives. So then you have, that explains a lot of the downstream hand-wringing about this, which I've told all the Republicans I know, start publications that do journalism, not just commentary.

41:55The reason Wikipedia, and I just got asked a comment for this New York Times story coming out about the Wikipedia's bias. It's very easy to understand Wikipedia bias. they only will let you put a fact on wikipedia if you can cite something what are citable sources they're trusted news sources or trusted like research studies you cannot link to a breitbart editorial you can't link to a podcast with you and i giving our opinions that's not a trusted source in wikipedia's definition at the same time they did that the number of conservatives in the number of conservatives doing actual journalism versus commentary has plummeted.

42:38You have a lot more doing comedy, a lot less doing reporting. Therefore, they don't get cited. The New York Times, Breitbart are not allowed to be cited. Now, sometimes Breitbart and the New York Post will actually do journalism, but often they're doing commentary. And this is why I think the free press is an interesting example, because they will, you know, I don't know exactly their politics. I don't know. How would you describe the free press? Moderate right leaning or moderate left leaning? No, it's MAGA with a thin veneer of intellectualism smeared across the top of it. Okay. So showing your bias, I think there's probably a million people who would say it's not MAGA.

43:15Because didn't they disconnect from MAGA massively with a lot of these things? I think they... Not really. Really? I also, I was around, I read a lot of the free press around the election time. So I don't think I'm even being that unfair about it. I think, I think they're, they're masquerading as the center right when they're actually a bit further right than that. Okay. And that's fine. That's fine. It doesn't bother me, but they're doing real journalism. Like they actually do about half and half. There's a lot of commentary. Sure. But they're like one of the few, few of the right, if they were considered right leaning, which I think they are.

43:45Um, I would say moderate to right, but maybe not. Um, yeah, she was, Barry Weiss is a New York Times columnist, correct. Anyway, I think to the point as well, there's two things going on here. The press is overwhelmingly left-leaning, and you have things like the Tech Meme Leaderboard here. To your point, who ranks high? The number one is Mark Gurman, who is an Apple pod, you know, who covers Apple. Apple for clicks, Elon for clicks, you get the idea. You'll move up the rankings if you cover these things. And when you were at when you were at TechCrunch, without revealing too much, people's compensation based on, in some way, or their career advancement based on traffic, views, leaderboard?

44:33No. No, we had a, not to bring it up again, but we had a very, very lightweight, tiny bonus system that might add up to like 1 % of your total comp if you wrote the most stories or got the most traffic. But we tried really hard, actually, to not. from the top down, think in those ways. But if you're a reporter and you publish a story and no one cares, and then the next time you pick up a little shorty about Tesla does X and then everyone reads it, that's a signal that you respond to, you as a market participant. I think while I would love to riff with you for about an hour and a half on the media thing, let's keep going.

45:10Next up, Sequoia has raised new funds. And Jason, I have a question for you because I need to better understand fund dynamics. So the news item here is Sequoia Capital, probably the single most famous venture capital firm of all time, has raised a couple of new funds, a$750 million fund aimed at Series A focused firms, and another$200 million fund for seed, pre-seed investments, the earliest stuff they do. The thing is, if you go back to 2021, Sequoia famously became a RIA, and they also announced a permanent capital setup. And my understanding of that, Jason, was that they were kind of done raising discrete funds, and instead they were just going to use their recycled capital in the so-called Sequoia Capital Fund.

45:53And now we're doing this. So just help me understand what's going on here, because I assume you know better than me. I think it's probably people... It's a little bit of both, and people are confused. There is the permanent capital model. Things go public. And then you, as, I don't know, some endowment, and you say to Sequoia, we're a giant endowment. There's a billion dollars in distributions over the last two years. You manage it as part of this Sequoia capital fund. You tell us when to get in and out of these stocks, because you're still on the board of Square or NewBank or, you know, Uber or whatever they were early into, Airbnb.

46:32And you will probably understand the fate of the company better than our internal people because you met the founders in the Series A and you've been with them for 10 years. So why not stick with them for the next 20? And so they have that permanent capital base. And I think you can redeem every year. You can have a chance to, you know, pull out some money from it. Then they're still doing their venture funds because some LPs just want to have the venture fund access. They don't want that. So that's what's going on here. And what's notable here is their discipline. Their Series A fund is$750 million.

47:04I don't know what the past ones were for Series A funds, but I'm guessing they were typically$350 ,000,$500 ,000. So they've gone up modestly. They've probably doubled in size in 10 years. But that's because the Series A check size has doubled in size. And the Series A feels more like a Series B in the industry. And the$200 million seed fund, our last seed fund,$45 million, is probably double what they previously did in seed funds. And the fact that they've separated those in the seed fund, they don't join the board in the Series A, they do join the board. And that gives them two swings at bat.

47:37The seed fund can just make these$500K to$3 million checks. They can put 40 of those investments maybe into the$200 million fund for about$5 million on average, a$2 or$3 million original check, and then, you know, maybe follow on for some of them. And then the Series A make their own independent decision and the growth fund makes their own independent decisions. The only challenge here is when you're a Sequoia CEO, you know, you have to then figure out, okay, which fund am I gonna be in? Is the growth fund gonna do this? And it can be a little conflicted. And then, you know, your later stage investors, let's say you go to Founders Fund or Andreessen for your Series C, the first question, who's participating from the Series A and the Series B?

48:18Is Sequoia participating? Are Benchmark participating? How much is available in the C? Is that a negative signal? If they don't, if they have a growth fund, Now the growth fund is looking at the seed fund. So if the seed fund owned 6 % and they didn't do the series A, now the growth fund is looking at Uber and saying, huh. And just as that example, I did the Uber investment as a Sequoia Scout from the seed, where even though it was its own fund, it was the Scout fund. Then they did do, my understanding is that the growth fund had invested in Uber. I can't confirm that right here, but that was what I had heard.

48:53so but they in the middle they didn't do the series a b or c they didn't get there or they didn't compete for it i don't know what the terms were but they didn't benchmark and then eventually shervin did the b with menlo so you can uh look at those two and uh you can see that they have discipline i don't think they want to chase the average return no no i mean that's why i started off by saying this is sequoia this is not you know interesting yeah i was trying to avoid that example but the reason to avoid it i mean this is what everybody would say in a bar uh talking so laura day so lower fund size higher expected returns more concentrated positions um sean mcguire a partner over at sequoia did a separate interview with oh man i forget the name of the publication the israeli focused publication calicastec or something like that anyways he was talking about how individual sequoia investors invest and he said look we only do per partner three deals a year so we're talking about fun size discipline we're talking about partner discipline in terms of what they have to choose so uh in that vein jason i went through the sequoia announcement i pulled out some themes from individual partners and what they're looking for over there so if you're a founder that's working on apps built on top of cutting edge image and video models or voice That's very interesting to them.

50:11Several mentioned cybersecurity. Observability was in there. On a geographic basis, one partner was really into Europe. McGuire had mentioned Israeli founders. And then they said that they think there's a trillion dollar opportunity at the intersection point of commerce and AI. So something in the e-commerce realm. I don't know what that would look like, but I liked the boldness of that bet. Hmm. Yeah. Um, there's tons of opportunities in Israel as well. I don't know how you run a seed fund from the U S and invest there without boots on the ground. Everybody asked me to invest in their region.

50:47And I've always had the same position, which is I don't, if I don't speak your language, I don't understand the culture. I don't understand the legal stack. Okay. And I can't use the product myself and I can't come to your office and hang out if I need to, it's going to be really hard. You're better serviced having a local investor. So I'm not sure how they're going to get that done. Um, but there was a big, um, another story with Sean. Uh, I don't know if you saw that story in the financial times, but the CEO of Sequoia reportedly left in protest, uh, because she felt his, um, uh, comments on X were, and I think it was the financial times who broke this story.

51:30Yeah, I have it up on the screen. But maybe you could explain what his spiciest quotes were and maybe the background on that story. I think if I was to choose which of his comments I think are the spiciest, I think I would invite a whole bunch of invective on myself from various online people. I'll just say that if you are sympathetic to criticisms of the Israeli military operation in Gaza, um well i feel like i'll just say it you just threw me a hand today comes from here's no it's a very simple like uh oh you mean just the mom dommy just the mom dommy stuff i think that's the one that she was objecting to oh okay so mommy comes from a culture that lies about everything it's literally a virtue to lie if it advances his islamic agenda the west will learn this lesson the hard way islamist i'm sorry islamist agenda sorry that's an important distinction so So that is a pretty spicy comment.

52:30If you were talking about radical Islam, I think that would be an important way to say that, like if you were talking about Al-Qaeda. But I don't think that applies to all people who follow the religion of Islam. So I think this is where politics is toxic and having leaders spout out about it all the time can cause chaos. Sean's tweets are very, very spicy all the time, Jason. They're spicy at night. They're spicy in the morning. Yes. Well, and also Paul Graham's. Paul Graham's been, you know, accused of anti-Semitism countless times. You know, now Sean's being accused of being an Islamophobe, I guess would be the term that people are using online.

53:13You know, if you're independently wealthy and you don't care about the fallout, you know, you can spout off, but it's going to have downstream ramifications. My lesson here for startups and founders is you may not be in that position and it could cost you your funding. It could cost you your employees. It could cost you partners. Why would you even do that? Like, why would you make that part of your, um, so this is a very important case of do what I say, not what I do. If you're Jason Calacanis and you're 54 years old and you're on one of the top 10 podcasts in the world and then you have This Week in Startups and you want to say spicy stuff because you already made your money and you don't care if you lose some partnerships on this side of the margin or that side of the margin or you're David Sachs and you're a partisan and you're part of the White House and you want to defend their position or if you're Paul Graham and you've already made your money and YC is so big that you don't care about losing Israeli founders who think you're an anti-Semite.

54:15Is Paul Graham still that involved in He's the patriarch of the program. So there are plenty of people. And yes, he still owns it. He owns it. Period. Full stop. He benefits from it. He gets all the carry. So yes. Oh, well, that's clarifying. I did not know that wrinkle. I mean, he didn't gift it to anybody. There's still, I think, three founding partners. I'm sure they get the bulk of the economics or half the economics. And then, you know, Gary Tan and the current partnership get some percentage of it. and he gets to hang out in London and enjoy his life. Very smart move. I think smart. But what I would tell any young executive, what I tell my young analysts, like, not worth it.

54:57Yes, that's... Not worth it. Just stay out of politics. It gets messy. Keith Raboy went back to Khosla. And Khosla backed Amjad Mossad of Replit, I believe. If I have all my things correct. Yes. And Keith was just going after Amjad, who we had on the show. I've talked to him. He's lovely. Great guy. We talked about our favorite boats. And - Raboy, also a great guy, by the way. Raboy doesn't talk to me. But I'll just say like, the thing I want to bring up here is that it seems that the era of expected founder friendliness, even across political lines, has been broken. And I think that that is a loss.

55:34So I'm with you. Yeah. I mean, if you think like chiming in on one of the oldest and most bitter conflicts in the world is like you're going to make some incredible progress with your opinion that hasn't occurred and you've never been to the region and you have no influence in the region but you feel you need to sound off by all means i certainly have fallen for that many times you know what i do is i in situations like this i call my jewish friends i call my palestinian friends and i ask them how they're doing i tell them i love them and i let everyone else who knows more geopolitics than me sort out the middle east because it's not i'm i have nothing novel i love i love you all all right i mean the the arab israeli conflict in modern day uh like the modern era of this conflict is 75 years old and it's in 50 years on the planet 40 years of which i can remember it from the age of 10 on i remember reading about it yep i'm not sure that this is the one where a venture capitalist or the ceo or the startup founder or the person who and the accelerators gonna make much progress on it i don't think that like their statements on twitter are gonna move anything now if you want to go there and start an inter start yc in israel or palestine you know paul graham feels so strongly about he should start y combinator in palestine he should make an you know a version of it there to help the people have a sustainable life etc if you know um and if sean feels so strongly about this he should quit sequoia or you know and create a venture firm in israel and and put all his energy into that i just think these you know incessant all day long i wonder if incessant all day long harping on these issues is actually good for the sequoia brand the y combinator brand or if it doesn't make a difference because those brands are so strong well i do know for the small folks who are coming up stay out of it it's not going to work for you it's just can i read that one can i read the comment from long okay all right lon's making jokes but let's keep moving yeah lon disrupting the show all right uh we're going to do one more awesome story jason then we're going to talk to a great founder who's a pitch for us but first i want to talk very briefly about mercor m-e-r-c-o-r we've had them on the show uh i think was back in august episode 21 59 this is what do they do this is a company that is working on connecting experts to ai companies so that way they can do reinforcements learning for human feedback or rlhf it has blown up uh it is raced from very little to 100 million arr and then all the way up to about 500 million in their annual run rate and they're raising 350 million dollars at a 10 billion dollar valuation blah blah blah blah blah blah blah 20x arr but is it i was going back through reporting from the information and they said in april that mercury pays up 60 to 70 percent of its top line revenue that means that its net revenue is far less than half of the then current 100 million dollar figure that they announced so this yields an interesting question about gross revenue versus net revenue we talked a lot about people kind of doing fuzzy things with arr but i don't think mercury is i just think they're counting the bigger number.

58:54I also found a great question from Reddit about people talking about how easy it is to game ARR and recurrent revenue in that context. So Jason, could I please better understand from your perspective when something should be counted as recurring revenue, when annualized run rate makes more sense, and how founders can avoid accidentally tying their shoelaces together and getting a pink flag from firms like Launch? Because I keep seeing this story come up. Yeah, so this is a valid issue, and we have an investment in Micro One. You may have seen. We are the seed investors in this company, and it's now grown to, I think...

59:31I think they've reported a raise at$500 million, and these two companies, Merkor and Micro One, are always mentioned, I think, in the same sentence. So I have a horse in this race. Got it. Of course, I wish all founders well. And just so people know, I don't see the items on the docket. I mean, I do see them if I want to, but often I'll respond to things on the show if I'm busy. Like, I didn't know Mercor was going to be on the docket today. I had like three or four stories I wanted to talk to, and then I rely on my team to present them. So this is an editorially independent program this week in startups.

1:00:05If I have an investment in Uber, we'll talk about Lyft and we'll talk about Grab, and I interviewed the Lyft founders. You know, Travis never had a problem with Lyft founders speaking at the event. Elon's never said to me, oh, you can't, you know, or got mad at me because I had a Tesla competitor on the program, whatever. We have done 2 ,300 episodes of this podcast. I don't pick the topics anymore. Probably a third of them. And we don't give favoritism to the portfolio companies here. I might say I'm really excited about the space. So just looking at this, there is a new group of companies like Athena, which I was the first investor in, and I was the first investor in Jonathan's other company, Thumbtack, where investors in Micro One, I won't speak for them, they can speak for themselves, we'll have Ollie on the program at some point.

1:00:52But MeriCore and these companies, if they are in fact not reoccurring, they're working on projects and it's low margin for MeriCore, let's say 70 % goes. By the way, you know, another company that gives 70 or 80 % of the revenue? Uber. Apple. Oh, yeah. Good point. Yeah, yeah, yeah. 55 % goes to YouTube creators. So it's pretty standard to when you have a marketplace to give the bulk of the revenue, the majority of the revenue, Airbnb would count in this as well, to the people providing the service. And that's marketplace dynamics. If you wanted to call it reoccurring revenue, yeah, that might be dishonest or not accurate because unless the person says we want to do, you know, I don't know, 100, I don't know how they price, but 100 million questions answered per month to train our data or 100 million, a billion per year, we're going to pay you$3 per or whatever it is.

1:01:59$100 million, we're going to pay you$3 per year, you're going to make$300 million from us. And that's a two-year contract paid monthly. So that is reoccurring, and it's a two-year contract. But it doesn't feel exactly like Slack reoccurring or Microsoft Office reoccurring or Salesforce, where your lifetime value would be considered four, five, six years. So the opportunity here is you're getting into something where people, this is a new product or service. That's what makes these new companies exciting. It's not just like the standard software stack. So I think all of these are going to do great.

1:02:32I think these are great companies and they're providing a massive service that's never ending. I don't think the need to train data is going away. Ah, that was my next question was, how long is this business model going to last? Because Mercore has a much bigger vision for what they want to build as a company and their AI work was the wedge originally. And I think it just got bigger than they expected. But they had a much larger plan that I thought they would just use this new capital to go off and pursue. because if you can sell 3.5 % of your company, Jason, for nine freaking figures, sure. Well, and wasn't the other company here, the one that Facebook bought?

1:03:05There was another company. Scale AI, which they bought. They would have been competitors, yeah? Yes, absolutely. In fact, Merkur and Micro One and other companies we've talked to saw their business just absolutely accelerate in the wake of that deal because no one trusted Scale AI anymore because of their meta ties. Sounds reasonable. I'm just reporting what everyone said. Jason gives that a thumbs up. but I think that they're benefiting quite strongly from this. I just wonder if VCs are properly de-risking the downside of this is contract labor and it's going to be price competitive. And, you know, that's what I think.

1:03:38The same argument was made about Airbnbs and Uber drivers and, you know, Athena assistants like, can it scale? Can anybody replicate it? What I'll say is my understanding of this data engineering work is that it's incredibly sophisticated already, and it's only year, like, whatever, two, three, four of this really starting to click in. It's not going to end, folks. It's only gonna get bigger, because as lawyers and accountants, as but two examples, or doctors, radiologists, nutritionists, engineers, as they start using these services and they go, wait, it's hallucinating. Wait, this is wrong information.

1:04:20Wait, cite your sources. Yeah. You're going to want experts to be in there saying, yeah, somebody wrote this page on Reddit. Somebody did this clipped article and they rehashed it with somebody in the Philippines for, you know,$6 an article. And it's slop. It's, you know, outsourced slop versus AI slop. somebody's gonna need to make sense of this and say, okay, this is slop, this is 60 % accurate, this is 80 % accurate, this is 95 % accurate, except the 5 % that's wrong in this article is because they didn't update the article in seven years and it was written in, you know, 2013, and boy, we found out that you really don't want to be taking this statin anymore because there's a new one or it's, you know, the whole philosophy has changed because of this reason, that's going to keep happening.

1:05:11Information is going to get harder and harder to discern. And if you look at what we just talked about in, I think, the second or third story today of creating synthetic data, well, I'm sure these companies are involved in that pursuit as well. I don't know that, but I would assume... I would assume so. I mean, if you were like, hey, let's make some synthetic data, they'd be like, okay, let's get some experts in here to tell you what synthetic data to make. Like, we don't need a bunch more Austin streets since there's 17 people on the streets mapping it. And, like, we know where every, like, pothole is at this point.

1:05:47We know every, you know, homeless person wandering into the street at every intersection. Yeah, literally, like, these things have... But we do need to do, you know, the drive between Austin and San Antonio and the cattle coming out of this ranch and the construction being done on this highway. Seems like those things are going to just keep popping up. And going from Biloxi to Lake Pontchartrain. We need other routes too. Yeah. All right. Really quickly, I want to do a polymarket, Jason, because we talked about a very fun thing last week. If you recall, there was a game going on in which a company set all the AIs against one another to do trading.

1:06:25And I'm not going to name the name of the company because I didn't want to give them points, but we did enjoy this chart of watching. watching the crypto guys trade uh sorry watching the ai models trade cryptocurrency and as you can see currently gemini and open ai are down quite a lot grok and claude are flat up 10 12 percent and then the chinese ais from alibaba and deep seek are crushing the game now why but that's because they're using inside information um maybe maybe but i mean you would think that they maybe are crushing, I mean, I'm making a joke here. Yeah. I think there might be something to the fact that they maybe don't respect copyright or, you know, paywalls and that kind of stuff.

1:07:05Maybe they have an edge. Maybe. And maybe the LLMs from America have been trained to not use inside information or they could have some red teams that have constrained them in some way. Maybe. So that makes this competition to get the most returns, it would give an unlicensed, you know, more freewheeling LLM a massive advantage it could be all i can say is by looking at this i want to set up my own deep seek trading bot but here's the fun thing over on polymarket watching ai's trade crypto wasn't good enough there's now a market for which of the models is going to become the best one so now jason we have humans setting wagers on ai setting wagers on humans setting wagers on crypto and this is what technology is for.

1:07:53And if you're curious, over on Polymarket, currently DeepSeek and Alibaba, as you might expect, are fighting. I think we can say this is the Alpha Arena AI training competition. This isn't like content marketing, I don't think. It's not like it's, this isn't like Carta or whoever saying like, let's get ourselves ads on this week in startups to sell our product. I'll put this as separate. Alpha Arena AI training competition winner. Everybody seems to think it'll be, I mean, 90 % chance here that it's going to be one of the Chinese language ones. But what's fun is, since I took the screenshot and I put it into the docket, this changed quite a lot because this changed quite a lot.

1:08:32When you say this, the people listening don't know what you mean. The two leading lines became much closer together, so people are now making a much more even trade between DeepSync and Alibaba than they were because the results from those two models have changed dramatically in the interim, as you can see right here, Jason. So I love seeing synthetics on top of AI because it just humans will speculate on anything and I love seeing polymarket kind of cross over into our little world so shout out to our friends over at PM Microsoft has a service called Microsoft 365 it was previously called Office 365 previously called Office it's Word Excel PowerPoint Cloud Storage you pay by month yes and now it includes AI because Microsoft of course is working a lot on AI now down in Australia they reached out to their customers on the consumer side of Microsoft 365.

1:09:18And they said, hey, you got to pay us more money to get AI or you got to cancel. So a lot of folks were like, well, I can't not have Word and Excel. So they did convert to the higher paying tiers. But if you clicked cancel, then they offered you the ability to keep your current plan without AI. Scummy, scummy from a multi-trillion dollar company that doesn't need to grind out the extra 48 cents, Microsoft. This is something Satya Nadella needs to get his hands around. We call these dark patterns. I give Lena Kahn a hard time because she destroyed the industry for four years. One of the things she did well, always I'm calling balls and strikes, you know, as the official chairman of the interwebs, which I'm now serving my third term.

1:10:02Third term, wow. Third eight-year sequential term. I'm now 24 years as the chairman of the interwebs. I'm pretty sure that's his dictator at that point in time. I mean, if the people vote me in, the people vote me in. It's up to them. I'm not saying I won't run for a fourth term, even though the limit is three. Let's start making the hats. I'll start making the hats now. Jake out for his fourth term. Here's the thing. Dark patterns are not good. They, especially to your point, Alex, when you're a big company, you don't need to do dark patterns. Now, there's somebody trying to hit a number for retention and for their AI goals.

1:10:41So there's somebody who tried to hit their goals and it's understandable. They want to do great. So they did a dark pattern, apparently, allegedly here. The problem with doing these is you always get caught and then it always gets amplified and you become an abusive, large company and the resentment builds up over time. If you're winning, you need to be magnanimous in life. as I started to win in life, investing in companies, doing podcasts, whatever it is. I always kept this in the back of my head because my book author, John Brock, my book agent, John Brockman pulled me aside one time. He was like, you made it kid.

1:11:21You can stop fighting with everybody. You can be magnanimous in victory. And I like, boom, it hit me. And I was like, oh yeah, you're right. I've been battling for 20 years to try to get something going in my career, but I hit. So now I have to flip to magnanimous. What does it mean to flip to magnanimous? It means I don't need to be fighting with people over things that don't matter. I need to be helping founders and remember what got me here. What got me where I am is, you know, entertaining on a podcast and I help founders. Got to stick to your knitting and just really lean into that. That's why I love having founders on here and I love Founder University and getting involved in year For Microsoft, how did they get there?

1:12:01By building productivity tools that helped people. Yes. Not dark patterns. No. Dark patterns make you look lame. Like you're using your technological business savvy to abuse your own customers. That builds resentment over time. And then when you get three, four, or five of these strikes against you, you know what happens? Your customers are gonna start talking about it, the press is gonna get on it, and then you're gonna get different federal agencies who are going to go, oh, I'd like to have that pelt on my wall. I'd like to have that scalp on my wall. That's an example, Jason, of just what you're talking about from September.

1:12:42And this is all allegedly, by the way, I give them the benefit of the doubt. Fair enough. But we talked about the FTC case against Amazon for their prime renewals upgrades. I forget the exact details of this, but last month, just about a month ago, The FTC secured a quote, historic$2.5 billion settlement against Amazon. So you will get caught. This happened a month ago. I didn't even see this. The FTC... That's because it's the Trump era, Jason. Everything's blowing up all the time. Got it. So between the White House, East Wing being knocked down, the Epstein files, Tower of I didn't actually know that the Federal Trade Commission secured a historic order against Amazon that enrolled millions of consumers in Prime subscriptions without their consent, knowing it made it difficult for consumers to cancel.

1:13:34Amazon would be required to pay$1 billion civil penalty and$1.5 billion in refunds back to consumers harmed by this. And here's the thing. If you want to raise the price of Amazon Prime, I'm here for it. You know why? You lunatics. I was ordering some camera equipment. I'm teaching my daughters how to take pictures. I found my old cameras. Turns out one of my nannies is in photography school. I was like, hey, when you're here, if I gave you this old camera equipment, will you take the girls out on the ranch and take pictures of animals? She was like, I can do that? I was like, yeah. And I had to buy a bunch of old compact flashcards and batteries because this literally, I found a duffel bag of the equipment I had from the Weblogs Inc.

1:14:14and Gadget days. That's awesome. It was just in a box. So I took it out, I cleaned it up, and I was like, oh, the batteries don't work. They brought the batteries at 4 a.m. today in the compact flashcards. I don't need them at 4 a.m. I don't need them till next weekend. If you want to charge me an extra$10 a month, $5 a month, we love you, Amazon. We love what you're doing. You're getting us stuff too fast. I have now gotten to the point where I'm like, I don't need it that fast, and opening up four boxes on three different days is more painful than getting these things early. Wait till Tuesday.

1:14:47Put it in one box. I only have to open one. That's how good Amazon is. You don't need dark patterns. Exactly. You'd also don't need Amazon to start putting unskippable blocks of minute long insurance ads into my prime video, because let me tell you, everyone in the world i'm not buying insurance not all state not state farm not the annoying yellow brand off off anyways here's the other thing yeah how about communicating instead of in a dark pattern why don't we become the white wizard gandalf gandalf the white right i think he was the gray and he came back as the white if a return to the king yeah if this is a good analogy let's go with If there's a better one, pop culture expertise.

1:15:33Sorry, you can't bring up Lord of the Rings and not get my nerd back up. Okay, but is it close to, like, he's as good as it gets, as the white wizard. Perfect. Be the white wizard in this, which is, say to people when they're watching Amazon, your Amazon Prime subscription is$12.95. If you want to get rid of these ads, which we only have two per episode, you have to pay$16.95. If you'd like to do that, that option's always available to you. Would you like me to remind you of this? Or would you like me to never remind you of this? Take the opposite approach. Be super generous. You know who does this?

1:16:07I knew it with Stratechery, which is like one of the best newsletters ever. It's like the OG of like newsletters really considered. He would send my$10 a month bill to me every month. I had never seen somebody in a newsletter do that. They all just hid their updates of their monthly billing. He sent it to me every month. And it said right on the thing, click here to cancel. And I was like, I'll never cancel. Yeah. Strategy because a it's great. B you're so upfront about it. So upfront about it. Now I want to take all of this and apply it to startups because briefly, I know we've gone a little bit long today, but the, the Microsoft thing we're talking about from, from the perspective of here's a company that doesn't need to do this and they're burning, uh, goodwill.

1:16:49But if you're a startup out there trying to raise money, competing with other companies for capital customers, attention, it's pretty tough, Jason. And there might be a temptation to make it a little harder to cancel or maybe a little dark patterny. So how do you, as a founder, with less capital, less cash, more pressure, always stay on the right side of things and not fall prey to essentially growth hacking gimmicks at the expense of customers? You have to be obsessed with your reputation at all times in your life. now when you're young and you're just on adrenaline and nicotine patches and caffeine and you're burning the midnight oil you might not always be thinking sensibly you might have pressure from well done with your vape and uh coffee uh you might not be seeing straight just always go back to if my mom knew i was doing this just do the mom test if your mom if you work for or Microsoft, and you did this, or Amazon, and your mom saw it, what would your mom say?

1:17:46My mom, when I would drop an F-bomb, she'd say, you know, I thought the book was spectacular, but the F-bombs I'm not comfortable with. And I started saying frackin' or frickin'. And my mom gets upset if I curse. God bless her, you know? And she's right. This is, you know, if I'm cursing too much, it's my Brooklyn side, and it means I don't have clarity of thought that I can't come up with a better word to use to express my frustration or feelings or my point of view. And so, just run the mom test. Hey, Mom, I tricked people into upgrading their Microsoft 365 package. Can I show you what I did and how many people, percentage of people I tricked at Christmas?

1:18:28If your mom's not going for it, and she's like, I'm very disappointed in you, then don't do it. Don't do it. And if you're the leader of a company, what happens is you have to be careful about about incentives. Incentives matter. Show me an incentive. I show you the outcome. So in the case of a startup, if you make sales, hey, if you book a sale, you get 10 % of the sale. And if you book the sale for the premium product, you get 20 % of it when they sign up and give us their credit card and we charge them. Okay, fine. I get it. What you want to do is say, you know what, instead of 10 and 20%, you're going to get seven and 15%.

1:19:04Then you're going to get the other 7 % and another 5 % if they use the product and they're in and they use the product more than 10 times a month you'll get another bonus and if they renew you're going to get another bonus so that you're not incentivizing them just to churn people and sell it to the wrong people and trick people into getting it you want alignment alignment so a lot of times people will make the renewal of a sale much less than the initial cost of the sale when in fact the renewal speaks more to the entire team doing their job so yes think about hey if our renewals are 80 the product team the sales team everybody gets you know uh two days off at christmas more valuable than any amount of money if our renewals are 90 percent or 85 everybody gets three days off in the holiday if we hit 90 everybody gets four days You come up with something like that.

1:20:05Now, you get that esprit de corps. You ever hear about the esprit de corps? Yes, I do. I don't know what it means. No, it means the spirit of the group, right? Yeah, it means essentially how does a team feel. And esprit de corps usually in a positive sense, as in we have good esprit de corps, which means we have lots of spirit. All right, everybody. It's been another extended, amazing edition of This Week in Startups. If you care about startups and technology and you want to be the tip of the spear, if you want to be part of the vanguard, the early adopters. Tune in Monday, Wednesday, Friday, here.

1:20:38And we'll be announcing a second show in the new year shortly. We're going to have a second show, a spin-out show from This Week in Startups. This Week in Startups family is going to grow. I'm excited. I don't even know what's going to happen. So I will be tuning in to hear about our show's expansion. Tune in. You know what I'm talking about, but just don't spoil the surprise. And please sign up for our YouTube channel, ThisWeekInStartups.com slash YouTube. And if you could do me one favor, please write a review or post a comment on Spotify, YouTube, or write a review on Apple Podcasts. If you do, we're going to include your comment, if it's an intelligent, funny, insightful, spicy comment in the YouTube comments, the Spotify comments, and you can tell us what company you're from.

1:21:25We'll give a shout out to your company, and we're going to play those at the end of every single episode. So if you make it to the 75 minute mark, 60 to 75 minutes, or if you want to just advance to the end and see if you got picked for that, for today's show, go ahead and do that. We'll see you all next time on this week in Star Wars. Bye-bye.