Elon Musk Is Leaving Washington, Meta Partners With Anduril, Kim Jong Un's "Side Launch" Goes Sideways | Peter Rahal, John Andrew Entwistle, Ashlee Vance, Chase Lochmiller, Matan Grinberg, Jonny Dyer, David Hsu, Sushanth Raman, Rob Toews

28:27Our answer is you work harder, not you go to jail. What has become clearer in the aftermath is how an unconventional choice of launch method, Kim's rush timetable and a top heavy warship overladen with weapon systems was a recipe for disaster. They just keep being like, yeah, just like throw one more missile battery on that. Yeah, yeah, yeah. Yeah, put a Gatling gun on it too. Yeah, yeah. Another cannon? No problem. You just keep doing it and then it just rolls over. It was a recipe for disaster according to Satellite Imagery Analysis. I haven't seen a failure like this one. I've never seen a failure.

29:02I said a retired Marine colonel that the Wall Street Journal called up and he's just like, this is insanely bad. Yeah, rough. Who are you on the phone with, Jordy? No, just calling up Mark Consian. Yeah, how bad is this? Haven't seen one this bad. Never seen one this bad. It was a 470-foot-long warship, Kim's second destroyer. They've been built in the northeastern city and did not go well. Anyway, our guests are going to start rolling in. There's a bunch more stories here. We've got Peter Rahal in the waiting room. Should we bring him in? Yeah. Let's do it. Peter! How are you doing? Welcome to the stream.

29:46Nice sweatshirt. Looking good. Thank you. It's a sample, too. Congratulations on your second announcement of the week. This is the first real one, I guess. Sorry that you got leaked, but very excited for you. Excited to have you on the show to break it all down. What is the most recent news? I only know the incredible revenue number, but what else is driving the news cycle right now for you guys? Well, so we acquired a supplier, an ingredient supplier called Effigy. And then, yeah, our valuation is$7.25. Wow. Forecast this year is$1.40. I think those numbers, and I think it's like we've accomplished more in nine months than...

30:38We did an RX over four years. Wow. So I think we're one of the fastest. What other brands have achieved that level of revenue within a year of launch? There's two. Feastables and Prime. Prime, yeah. Because just massive distribution cannons and really leveraging essentially$100 million worth of free marketing on day one. right yeah yeah so what's the secret for you because it feels product-led yeah is that is that is that the right assessment obviously you have a bunch of lessons uh you know war stories you can pull on from rx bar but like it feels like the product just sells itself in some way yeah i would say it's definitely product-led um we actually you know in cbg it's pretty hard to differentiate We appreciate measurably.

31:36And 75 % of our calories are coming from protein. The market is at like 50. So it's meaningful. It's an example. The market was on the surface. We don't apply rigor. It looks pretty competitive. But if you just spend a little time, talk to some customers, you realize NPS is pretty low. No one's really happy. um and a lot of people don't just aren't even in the category because of taste and texture and nutrition so the big opportunity for us in the bar business is like to really expand the category and make delicious nutritious products that that people are like oh oh protein bars don't taste like shit like the expectations are there's just they're just gonna be bad yeah yeah when you say people aren't in the category are you talking about the difference between protein shakes which can typically be like extremely high protein from calorie ratio uh more specifically people are like people don't consume protein bars like you'll you'll you'll hear people say like oh i'm just not a protein bar person sure sure sure sure yeah those are people like you ask them like why not it's fundamentally taste yeah i i i want to kind of push back on this idea that it's product led i agree that there's product differentiation but This feels like it's only possible from you because if you call up a major distributor or supplier or retail chain, they say, yes, I know you.

33:01You have a win under your belt. I'm willing to go big and we don't need to do some long extended test. This isn't to knock you. I mean, this is why you worked really hard and you're reaping the benefit of that. But is that critical or do you think if you were nobody and you were cold calling? My position is more so like Peter would have the relationships to get the bars on the shelf, but he's not making, you know, holding every customer that walks into the store at gunpoint. That's true. Buy my product. I mean, it is very different than Mr. Beast and Prime where, you know, you have like major influencers driving this.

33:33Like it is more product led in that sense. Yeah. And I would say my credibility or track record is like a lubricant to facilitate things. Yeah. But the repeats and it's all product driven. Yeah. yeah that makes sense talk about the decision to acquire one of your major suppliers uh was that something that you anticipated doing a long time ago or you know what was the point that that made sense yeah so when we started you know in our business you don't want any single source supplier it's just too much risk so it's like it can become a nasty dependency so we identified the technology, the ingredient, and it was like, this is like magic, like technology should feel.

34:18Got close with them, became 90 % of their volume. So we were basically, you know, 90 % of the, yeah, 90 % of their revenue. And it just made sense to vertically integrate and de-risk it. And so we can just control the supply. Yeah. And, you know, I can't imagine a company without without us being together um and it enables us it just like totally widens the aperture of like where we can go um and truly have a platform for across different different products different different brands different different consumer needs how how do you dig into like the defensibility around that intellectual property because i feel like the gold standard for defensible ip in you know anything fda regulated is probably um pharmaceuticals like drugs but as we've seen with like the Ozempic-Woguvi transition, like even GLP-1s, they were able to kind of eat at the edges.

35:16Now Eli Lilly is taking share from Novo. And it feels like if it can't work in GLP-1s, like how can this possibly hold in a protein supply chain, right? Like you're gonna get someone who spins something up that's like one molecule different or something like that. Is that a risk or did you dig into that at that level of depth? Yeah, I mean, there's really one process to make the molecule. Sure. And, you know, it's like, sure, someone can go try it, but we'll litigate and calculate that. And I guess the bull case is that like, yeah, you can go and fight on the IP stuff, but at the end of the road, you don't get a pharmaceutical product.

36:03You get a CPG brand. and so it's like our line's not there you're gonna have to fight real real hard and at the end you then you also have to build a brand also do the distribution and also whereas you know eli lily like if they spend a bunch of money to figure out how to you know alter the glp1 to get a drug that works like well then it's just a matter of doctors prescribing it right and it's and it gets paid and like in cpg you kind of have like two years runaway before someone copies you yep now we have like structurally nine years. Yep. And I think by that time, it's kind of too late. Yep, yep, yep.

36:36I mean, you already see that with the 140 revenue. Like as that gets bigger, it's just going to be, you know, ubiquitous and you own all the shelf space. So even if there's knockoff copycat, it's like, yeah, you're already big. So you're less than a year in post-launch. What does a good outcome look like? You've already had a multi$100 million exit. I imagine you're aiming a lot higher here, but but what is that in your mind yeah i try not to think about the outcome too much but i do

37:13that's the most honest answer ever any entrepreneur is like oh yeah i've never thought about the outcome it's just purely mission you know yeah it's amazing yeah um like i would be pissed if we don't get to a billion in revenue yeah yeah you know i that that would be disappointing what was rx bar roughly like what's your personal high watermark 240 okay yeah okay and i think maybe maybe a little bit more yeah yeah yeah yeah so i but i i do think i i do think we can build a really diversified portfolio of brands that really address a broad population and create a ton of consumer surplus so you know I think like yeah we just have so much work to do and yeah I do fantasize about being a public company let's go i love it and i think this is what my dreams are we'll have you back after your public company and you're just gonna be like it's such a headache it's terrible i can't say anything i'm in a quiet period i can't come on tvpn it's the worst yeah i might be the irreverent yeah these calls but but um um yeah I just think like there hasn't been a truly uh from scratch brand platform in the space it's all been done through M &A yeah I think you're Trevani's doing it actually now yeah sure um and I think that and you're a believer in in these brands living under the existing you know C Corp and or or like it doesn't sound like you want to do like a studio no I only studio analogy would like we we're not gonna do m a we're gonna we're gonna build them like yeah i i me and my team we've demonstrated we can like go from scratch like um yeah so so we would just build them it would be a decentralized model so like different orgs you know like so david's the master david's mars and then david protein is the operating business epigee will be its own subsidiary and it will create other brands um and then they just have to be decentralized organized like orgs because you you know like david in his life cycle is very different than a newborn baby yeah and so like you need to design the order for agility and and so so that's what we're designing for um and how much do you view yourself as a technology company i heard you bring that up earlier as in you're wanting to pursue opportunities where you can have a unique sort of durable edge through some type of effectively chemistry i don't know if that's the right chemistry um i you know we're in the chemistry business now i want to keep learning and investing you know i don't know we don't know but like once you get in the pool you'll figure things out so um you know like technology is not really welcome in food um so we we we restrain from the t word um uh but yeah we want to keep making really valuable products like that's just our focus and in obviously technology is really the best way to get there you have a technology you have investors that are traditionally sure technology investors I mean I don't think I've ever seen green Oaks do a CPG round yeah the last deal they did was like windsurf right they're like the last big outcome for that wildly different company have you have you taken a crack it kind of reverse engineering what it takes to really go after like a Nestle or a Unilever, like one of the major, major conglomerates that has been built through M &A.

40:57But if you think really, really far into like what it takes to create a hundred and a hundred billion dollar outcome in consumer packaged goods, no one's even come close. Like we all cite like Red Bull is a great outcome. But these are all like single things and very niche categories that are big categories like energy drinks. So the companies get big, Red Bull's big, Monster's big, but no one's really been able to figure out the right corporate structure to to just compound and compound and compound and create this like hundred million dollar hundred billion dollar behemoth yeah so the key way to address the really large tam and consumer food is through different brands that have different dna that address different sort of problems or consumer state like so like david i use the analogy of like d brands are just human beings they have fathers they have they have dna they have behaviors beliefs they you know have friends you got to let that brand or that asset be it be itself like you can't jam it into places it doesn't belong so like the david brand with our future portfolio and optimizing for calories coming from protein um so if we look at david you know it's like the most protein least amount of calories that that time is probably 1.5 billion in revenue in the u.s so so you really need multiple you need like a house of brands to go after very different parts of the population yeah that's one in an international yeah what's the we're mega scale and they're global so that's that's the key thing what's the pushback been like from kind of like the trad community that wants to do everything all natural, everything like, you know, oh, just, you know, farm to table, the Michael Pollan crew, the RFK crew, like there's all these different, uh, different segments of kind of, uh, like anti-modernity and food.

42:48What's the reaction been like? Yeah. So, um, you know, Peter Thiel talks about like anything with science at the end is not science sure so nutrition science is really today not been science yeah um and so so it's really been it's a really emotional conversation and and not a intellectual conversation so that's the first pillace and it is complicated and and our society is really confused around it and so in under the confusion and the way to simplify something complex people find like just simple correlations like if my ancestors didn't eat it i shouldn't and that that's like a pretty good framework or like it's not bad advice but it's clearly not sophisticated and it's way more complicated than that yeah so um and it's like the interesting thing with like the ancestral movement my friend told me this and i thought it was like really good it's like the ancestral food movements like it's like post-traumatic stress response like it's like a post-trauma response where you're like like you just stop and you like don't progress you actually go backwards because you're scared to make it worse and perhaps that's because food fucked up and food you know yeah got run by cfos who just cut the bottom line and just cut cut cut um but i do believe like there's a way to advance food in a way that and it isn't that scary and um yeah so i think i i goals to have like a more intellectual conversation on food like you know not something like oh i can't pronounce it therefore it's automatically bad yeah yeah makes sense like nutrition is about like it's pretty simple it's about the 80 percent like don't overeat calories don't be fat turns out that's terrible don't spike your blood sugar and get enough protein yeah and hit the gym i wish we had more i wish we had more time there's a there's a bunch more questions i have but uh congratulations on on the milestone.

44:48We're just getting started. Cool. And appreciate you coming on. Yeah, we'll talk to you later. You know. Congrats. Bye. Next up, we have Ashley Vance. You know that Green Oaks is underwriting that too. Small chance of$100 billion. For sure. It's like that's why they're doing the deal. Neil Meta. Undefeated. Next up, we have Ashley Vance coming in the studio from Core Memory. Bring him in. How you doing, Ashley? I'm good, boys. Oh, look, he's in the new studio. New studio. Reveal. Let's go. Exclusive. Let's go. Breaking news. Core memory. Interview Palmer Luckey. You're hearing about it here first, folks.

45:26You're cooking. You're cooking. It was good. Good morning. Yeah. Palmer came. I think he, well, he came in yesterday, I think straight from, there's this photo we posted of him and Zuck together. He's wearing the same clothes. I should have actually asked him, but yeah, I think he came straight from their peacemaking accord. Yeah, what was your read on it? How much of it was, how much were you focused on like the technical side of the deal between Anduril and Meta to work on VR for the military versus just the crazy full cycle narrative of the emotional journey? We kind of did both. I mean, he sat here for a couple hours and from my memory, you know, I think the first 45 minutes was going into the deal, the backstory, all that.

46:14The press release on this is super thin on the technology and the Wall Street Journal story didn't have that much either. And Palmer, he went into some detail on our podcast about the tech. And then, yeah, he got into the backstory. I mean, this is crazy. I've been following Palmer pretty close for the last couple of years. I never would have imagined that this would have been possible, given the intense, the depth of this like nine year war that he and Meta have had. So, yeah, I did not fully see this coming. Is it accurate to describe it as a war between Zuck and Palmer or a war between Palmer and Meta?

46:54I think it was like more Palmer and Meta. You know, if you go back, Blake Harris wrote this book, The History of the Future, and he spends like 40, I don't know, 40 or 50 pages laying out the saga. You know, Palmer put up this paid for this billboard and about Hillary Clinton and became a known Republican in public. It was fired from Facebook. And, you know, in that book, you see a lot of operatives within Facebook kind of pushing Palmer out. And we talk about this on the podcast. I mean, if Palmer's being truthful in what he said, he didn't lay the blame directly at Zuck's feet. He kind of felt like Zuck was doing what he had to do to protect his company and what the employees wanted.

47:40So I think it was more meta-directed. Yeah. It is crazy to think about the post-Oculus acquisition time period at meta and just not having Palmer report directly to Zuck. Like that seems like that's the real cardinal sin here more than the crazy stuff that happened politically. It's just like, how do you bring in someone who's like clearly the face of VR generational founder, like all these amazing things. And like, it would just have been so much easier to manage that relationship if there was no one else in between, like no management layers in between. Yeah. And like almost definitely, I think Palmer's probably still at Meta making, working on VR.

48:20I don't think he would have done the handle, you know? Yeah. Didn't Jan Coombe from WhatsApp stick around for a long time? And obviously the Instagram guys had kind of a rough go and they're upset about it now. But I think they reported directly and built that for a long time. It was just odd that VR was so important in meta that reality labs, huge investment. And yet Palmer Luckey was not, he should have been on the board. He should have been added really into the inner circle. It's been a while since I read Blake's 20s. Who cares? Zuck was 19 when he started the company or whatever. Sure. Break the rules.

48:55And it's it's been a while since I read Blake's book. But yeah, I mean, he kind of lays out all these layers of management that were put between Palmer and Mark. So but here they are taking photos as friends again. Have you gotten the demo yet? Any of the Andrel hardware yet? I haven't gotten on this headset. No. I mean, this has been and I've been down at Andrel doing some reporting. and Palmer's always like, because this is like an extension of this IVAS deal that Andrel took over from Microsoft. So I've been down there and Palmer's like, oh, my secret lab is over to the side. That's where these guys are working.

49:33I mean, people should know. One of the coolest parts about this, I think, is that, I mean, yes, they're making this warfighter helmet of the future, like the sci-fi thing we see in every military movie that doesn't exist. But beyond the like VR, AR data feed coming in, I mean, Palmer talks about this in the podcast. He's trying to like reinvent the helmet itself, right, to make it lighter weight, to make it the ballistics upgrade, all of that stuff. I mean, so he wants to build. It's not just like the sort of tech end of this. He just wants to make a better helmet for soldiers, too. And I know he's been spending a lot of time on all the materials and things like that.

50:15Yeah. Have you been tracking any of the other developments at Meta? There were some setbacks with the Llama team. It seemed like they kind of went a little bit too far into the pre-training paradigm and maybe not enough in RL. And so Llama 4 has kind of fallen behind in some of the benchmarks. at the same time. There's some really amazing VR demos that are coming down the pipe, like the Orion augmented reality glasses. And then even just like the next Quest headset, I feel like is going to be incredible because all that work that Apple did to get those incredible screens into the Vision Pro, well, like Zuck's probably going to be able to buy that because it's two years later now.

50:55It seemed like that was something that was unique to Apple, but only for a small period of time. And Apple kind of whiffed on actually getting that into the hands of millions. If it's in the next quest, it could be really cool. So any takes on what Meta's been doing in VR or AI lately? I mean, I do not follow them as closely as I do some other companies. The thing that's always stuck out to me was when they bought Control Labs, which was doing this weird take on a brain-computer interface by reading the motor neurons in your wrist. And, you know, we've seen Meta do some demos around this kind of like new interface using your body and your brain to navigate computers.

51:38I actually think that's like the under, you know, it's probably one of the hardest things they have to pull off, but it's kind of underreported in some ways because, I mean, if you look at what, trying to read the tea leaves on whatever open AI and johnny ive are doing um you know we're clearly we're lurching every some companies are lurching toward a new kind of computer we've been on the same basic computer for decades well to be clear the open ai thing they were just they were pretty clear that they said it's a third device besides your phone or your computer that will work with those with those yeah yeah no that That is fair.

52:18That is fair. I still feel like we're... Everybody wants a new toy. Give us a new toy. Somebody will figure this out. I'm confident. I'm old enough to remember when I first started reporting. It was exciting. People had different operating systems. They had different takes on computers all the time. And I just feel like we've lost that. So this is one of the most exciting. What are you talking about? The new iPhone, it has a button on the side. There's an extra button now. We can only refine the bezel so much. You're looking a gift horse in the mouth, actually. Come on. This thing is awesome. I have a bone to pick with you guys.

53:00Please. You got this glowing profile information about your new media empire. And I was quoted in the story, and they didn't even bother to mention my new media empire. I mean. Boom. Sorry, sorry. I hit the wrong effect. I hit the wrong effect. I mean, it was half like complete Glazegate, like just so over the top, like lavishing praise on us for genius and reinventing media as a whole from the ground up. And then the other half was like a complete hit piece. It was brutal. And I guess you balance those out and it's like, you know, kind of a nuanced take. But yeah, they did some people dirty. They put us in quotes.

53:44They put us in quotes. They put us in quotes. This will be the last time TBPN ever appears in quotes. And the last time that Ashley Vance has ever quoted in the information without mentioning Core Memory. CoreMemory.com. Go subscribe right now. That hurt. That hurt. I like that everyone wants you guys to be the ringer or Grantland, and neither of you seem to know. I don't know what those things are. But it's even worse than that because people are like, oh, it has like SportsCenter. I haven't watched that. That's Squawk Box. Also, I haven't watched that. Can you comp it to something I've watched?

54:15We're reinventing the wheel. We're reinventing the wheel. At TVPN, we're focused on reinventing the media's wheel. Yes, yes. I love it. But yeah, give me the broader update on core memory. I know you have a bunch of different products. You're filming movies. You're writing books. You're doing short documentaries. Give me the overview. And I want to know specifically about the latest update with Neuralink as well. Yes. Okay, yeah. I mean, we've launched a bunch of stuff. I'm writing on Substack. We launched a podcast. Palmer is on it today. And then, you know, I think the thing that we're maybe most proud of or one of the things is, yeah, you know, I used to make a TV show for Bloomberg and we did very well there.

54:55And we kind of killed that, built our whole team from scratch. And so on our YouTube channel and on our sub stack, we've got, I don't know, they go anywhere from like eight to 20 something minute episodes that are, you know, dives into different inventors and scientists and startups. And then so anyway, please go. You're kind of telling the story of these like various Brotopias that are existing all over Silicon Valley. You know, unlike some people that only Brotopia, we do do many female scientists. But, you know, we're going all around the world, really. So we just went to so far. The early episodes were Silicon Valley based just because I had to get this up and running really quick.

55:34But we just went and filmed in Switzerland for a couple of weeks. And so part of that was, to your other question, you know, we're working on a movie about brain-computer interfaces, and Neuralink is at the heart of that. And so we're following this journey for many months, years, maybe. And then we filmed a couple episodes with some Swiss companies. Yeah. How do you think about what's happening in Switzerland? It's kind of, you know, it's always like this mixed, oh, God, you're going to get me on like a Europe rant. I'll try to contain myself. Switzerland is my second favorite country in the world.

56:13It's like, you know, if you're going to go with the well-made museum version of Europe, it's great. They do have some, I think, a lot of, well, okay, they have a lot of good biotech stuff coming out from pharma and just like an incredible education system. And they have finance stuff. But we went robotics. You know, I went to a couple of university robotics labs. And it's interesting. They're always doing pretty cool stuff. And then it's always hard for them, I think, to make the leap into forming a startup and like really getting money behind that. And sort of the same ambition that you would see out of a similar Silicon Valley company.

56:55But yeah, we met a bunch of young kids who were doing cool. You will see it coming up in a future episode. We had a robotic swan, robot swans dancing at a lake, shooting water and lights. And we were up till two in the morning filming that. They actually have a fantastic robotics industry. The American thing would be to put a gun on the robot swan. I'd love that. Over in Switzerland, you've got to do a profile on this. They have a fantastic robotics industry all around telling the time. And so they have these amazing companies. is you should do a whole profile on Patek Philippe, Adhemar Piguet, Vacheron Constantin.

57:34It's like this machine. It's robots. It's basically robotics, but just to tell the time. And so I can imagine an Ashley Vance style deep dive on those companies being really, really good. It's hard tech. I would enjoy that, I think. That'd be great, yeah. But yeah, it was cool. It's cool. They got good energy. The Swiss government really backs EPFL and ETH, these two universities kind of on a level that I don't know you you know I've been all over Europe filming the tv show and I'm always I'm impressed when I go to Switzerland do you think American investors should be posted up in Geneva just slinging checks is the is are they ready for American industrial venture should they go from spending just the entire summer there to spending some time during most of winter and most of winter ski season this is a ski season summer but then they could also spend some fall and springtime there.

58:29I mean, I try and, you know, I don't want to overgeneralize and crap on an entire country. But, you know, I do think there's opportunities there. But then sometimes you walk through the European startups and the energy level is not the same as you'd find in the U.S. They're not sleeping on the floor of the factory over there. We have billion dollar companies where teams are sleeping in tents. Yeah. We're about to talk to a billion dollar company where the founder is sleeping out in Abilene, Texas, where you got Chase. Have you been to Abilene yet? I have. I went with Sam and, oh, yeah, I'm going to name drop.

59:09I went with Sam and Greg a couple weeks ago and got the tour. That was like a big, everyone in Europe was asking me. They're like, are we hosed on AI? and I was like, well, I just went to Abilene and each one of those buildings cost about$50 billion and they have a lot of them and you guys have precisely none of them.

59:33So, I mean, it was kind of like a harsh, I gave a talk in Poland as part of this trip and people were not aware of the scale of like the investment. Wasn't MGX setting up a data center in France? Yeah, yeah. I mean, the overall Stargate project is available to other countries, and those deals are being negotiated right now. Because obviously, NVIDIA wants to sell chips, and Crusoe will want to sell energy. And there's a lot of different companies in the Stargate supply chain that want to be a part of that, even if it's in another country. But from what I was gleaning from chatting with Sam on that trip, it sounds like France is the only country that's kind of ready to pony up at the we would like to participate level.

1:00:17is it going to be like luxury ai

1:00:23lvmh gpt like branded tokens yeah chat artisanal tokens that is what they do best anyway this has been fantastic thanks so much we'll talk let's make it a regular thing thank you boys thank you i would love that yeah i'll see you this weekend all right thanks guys cheers up next we have chase lockmiller from crusoe energy coming in the studio welcome to to stream. Chase, how are you doing? Welcome. Where are you in the world? Are you in Abilene, Texas? Better be in Abilene. It might be. I don't know. It might be in San Francisco. Oh, we don't have him. Oh, okay. He is not here yet. I thought we had him, but we can do some timeline.

1:01:00What do we got? Oh, this is interesting. There's a shakeup in terms of who's in charge of AI at Meta. And I want to go deeper here because a little bit of OpenAI Day is coming together on Thursday. We have OpenAI and Anthropic on board. We need someone from Meta. So this is a little call out. If you work in AI at Meta, let us know. We'd love to have you on the show and duke it out with the rest of the foundation model, companies, CEOs, and research scientists. Lab on lab violence. Lab on lab violence. Anyway, speaking of the man who powers it all, we have Chase Lockmiller from Crusoe Energy in the studio.

1:01:42How you doing? Welcome. Welcome to the stream. We have no sound. We are missing sound. Are you muted? Are we muted? What's going on? Let's check it out and get to the bottom of this. And we can hear you now. He makes data centers. He's not a Zoom expert. Give him a break. What is new with you? How are things going? Give us the latest on all the news that came out. I feel like the last two weeks have just been Crusoe, wall-to-wall coverage. But how are things going in your world? Things are good. Things are busy. It turns out, you know, AI needs a lot of power and all these chips need a lot of data center capacity.

1:02:23So what we announced last week was, you know, the completion of funding for the expansion of our facility in Abilene, Texas. That's going to consume a total of 1.2 gigawatts of total power capacity. And that funding we did in partnership with Blue Owl Capital. So the total funding is about$15 billion in total capacity to build out. That's amazing. We love to see it. How is it different working with an investor, a financial institution like Blue Owl, versus some of the investors that you've worked with on the venture side? Is it more Excel and less vibes and decks? Is it a wildly different underwriting scenario?

1:03:08Like, you know, we're more familiar with the venture style where it's usually just a handshake and a term sheet on the back of a napkin. And a prayer. Yeah. I feel like I left the vibes investors, you know, behind, you know, a long time ago. Okay. That was more like kind of the earlier stage stuff. But yeah, so, you know, certainly different. And I think with Blue Owl, they've been a great partner, you know, one of the leading real estate, you know, private equity practices in the world. So very sophisticated. I mean, they've been super supportive and helpful across getting the entire deal structured and financed.

1:03:46So, and then obviously very, very deep, deep pocketed capital providers to really help us, you know, make these projects happen at really significant scale. Now, what's different is like, this is not an investment in Crusoe, you know, right? This is a partnership with Blue Owl for this specific project. And just, you know, Crusoe has a business model that is, you know, not asset light, is very CapEx intensive, which requires, you know, being able to tap into those very large pools of capital to basically make these large scale AI factories happen. So really break down the anatomy of like how these deals work.

1:04:28Is it like there's a new LLC or a new C-Corp that's created? And then is there something that looks like a mortgage with like a 30-year payback period, interest only period? And then I imagine that this facility is going to make no money for a few years while you're building it out, but then it'll start making a bunch of money. And so is there some sort of repayment schedule that's responsive to that dynamic? Yeah. So the breakdown of the structuring is, so this also came out, there was, you know, the exact number. Yeah, I'm sure you can't share. There is construction financing that's being provided by JP Morgan.

1:05:08So on the expansion, it's a little over 7 billion. That's been provided by JP Morgan and then a number of other banks and capital providers in the syndicate, including Bank of America and a handful of others. But the way it works, I mean, it is a prop co, right? So it's a property company that basically, you know, it's an LLC that ends up owning the individual, you know, campus or the individual buildings. All of those buildings have an affiliated lease with them, with our customer, which is a long-term lease agreement. You know, being able to partner with a large scale investment grade customer is really what helps unlock a lot of the capital here.

1:05:55You know, when people are talking about these very, very large quantums of capital, credit is like the magic unlocked. Right. So when you have big long term offtake agreements with high credit quality customers, that's where you can really unlock these larger pools of capital. and we can put$15 billion to work in a positive capacity. And then, so that's happening over there. How do you make money then? Sure. We make money as both a project developer as well as we are a partner with Blue All and the ownership of the entity. Makes sense. Our customer, it's basically like we're the landlord. We have a customer that is paying us a monthly rent for 15 years and, you know, we make money.

1:06:44That's in excess of, you know, sort of our debt service obligation. Yeah. Then talk to me about the energy side. That's kind of the bread and butter, like the history of Crusoe as I understand it. How important is it to find unique kind of combinations of resources to provide energy to these large scale data center projects? What's happening in Abilene? What's unique about Abilene? And then I want to dive into a little bit more about the life cycle of that energy production plan. Yeah. So, uh, so much of the bottleneck of scaling AI has boiled down to just lack of energy or lack of access to energy.

1:07:24And, uh, you know, Crusoe from its founding seven years ago has always taken this energy first approach to building computing infrastructure. And instead of thinking about, you know, how do I build the next data center in Northern Virginia? We've always kind of thought about like, where can we access low cost, clean as much as we can, and abundant energy to power computing infrastructure. And sort of the revolution you see unfolding with AI and sort of this complete transformation of the digital infrastructure landscape is pretty mind boggling when you really think about it. Because Northern Virginia is sort of like the center of the world for data centers.

1:08:06Everybody's like, okay, the Northern Virginia corridor, that's sort of what the internet is happening. That's probably where this Zoom conference is being hosted. Just so much of the internet happens in Northern Virginia. AWS, US East. We know and love it. Yeah, exactly. It's the backbone of our industry. Yeah. Totally. Exactly. So all of the data center capacity we've ever built in Northern Virginia is about four and a half gigawatts. Wow. What we're doing in Abilene, Texas is 1.2 gigawatts. Wow. And, you know, we're looking at trying to do more. Yeah. We're one company. This is for one customer.

1:08:42We're looking at other sites that are five gigawatts. Right. So you're talking about building a whole Northern Virginia that's been built over the last three decades. Yeah. One facility for one customer. Right. You know, there's just fundamentally not enough power there in Northern Virginia to make that happen. And so what's happening with AI is you're seeing everybody start to take an energy-first approach to developing this infrastructure. And that's really what led us to Abilene. Abilene is a market where there's an abundance of energy, a lot of wind particularly, and solar had been built on the back of production tax credit incentives.

1:09:19And their problem was actually they didn't have enough demand for energy. They would frequently get curtailed, which means they would have to sell power at a negative price, so they're shutting down their wind farm, or pricing would go negative and they would sell at a negative price. So their issue was actually just not enough demand for power. So it was a good natural fit between AI factories and low-cost, clean, abundant energy. It's a pretty awesome setup. we're going to account for about, I think the number is about 30 % of the total tax revenue for Abilene, which is like, we're like, well, that's amazing.

1:09:59Can you give me like kind of an energy 101 on, on energy production in Northern Virginia between, uh, I'm sure natural gas is in there. There's solar, there's wind. I don't know if there's any nuclear, are we still using coal at all? Like I really have no idea. Okay. What about like crude oil or fuel or like just gasoline? Does that power AWS at all? Like, I'm just curious about that mix. Yeah. No, there is actually, you know, especially during moments of peak demand. If you have peak demand at night. Diesel generators, probably. Extremely cold night. There's obviously no solar. Sure. And, you know, people are, you know, oftentimes having to use oil, you know, oil to produce power, which is like not a good, you know, not a good.

1:10:46And yeah, it's just it's like the dirtiest possible option here. Like we can be a lot cleaner. It's expensive. And it's expensive. And then contrast that with Abilene. What's the energy mix look like there? Is it similar or is there something different? You mentioned wind. Is there more wind in Abilene? Yeah, I mean, Abilene is one of the windiest places in the United States. Sort of this corridor that just gets a tremendous amount of wind. And that's why a lot of renewable energy developers built there. Sure. I think it's actually important to understand this production tax credit. So the way this works is that the independent power producers that build these renewable energy facilities, they get paid a production tax credit for producing and selling a kilowatt hour of clean power.

1:11:34Now, they get paid that regardless of who they sell it to and at what price. And so that has led to these consequences where you'll often see power prices go negative because their actual realized price is, after the factory in the production tax credit subsidy they're getting, is positive. But it's like they're having to pay someone to take that kilowatt hour, and then they go collect that subsidy through the production tax credits. So now the issue becomes those production tax credits only exist for 10 years. And so at the end of the 10 years, you still have this working wind farm and you're like, okay, power prices go negative.

1:12:13Now I have to curtail and I have to shut off, which I could be producing power, but I'm not because there's literally no marginal demand for the power. And, you know, I think this is where, you know, having this alignment of, you know, markets where you can produce power in a very cost effective capacity and you can actually build an AI data center there to soak up that energy is a very good alignment that, you know, Crusoe has tried to facilitate. Can you talk a little bit about the history of the company? I know that at one point there was some crypto mining with gas flaring stuff going on and then the transition from that into the current AI boom.

1:12:52Was that a major emotional roller coaster or did it kind of overlap in a perfect way where it was just like all growth? um yeah totally so uh you know we started the company one of the first applications of energy was uh uh was was as you as you talked about was basically capturing waste methane from oil production that would otherwise be flared okay and then utilizing that to power initially bitcoin mining data centers but you know we also powered uh our early versions of our ai data centers sure like smaller training runs right yep um and this was like pre-chat gpt this was like you know we were working with like you know the the mit department of physics uh doing uh you know early simulations of the big bang you know which news and that's uh you know ccl department at mit like you know early early work in our ai cloud development but ultimately you know when i started the company i really wanted to build an ai cloud platform that was like uh so a lot of people are like, wow, this was a great pivot.

1:13:55And I always tell them like, no, this was like the plan from day one, believe it or not. But I always felt like energy was the thing that tied together all computing infrastructure and any computing application when really scaled out, energy does become the bottleneck. And I had seen that and experienced that in these proof of work blockchains like Bitcoin and Ethereum. And I felt like if AO is going to scale, energy would be a massive component in the overall operating cost of operating intelligent systems at scale. And so we did a lot of early investment in terms of like making the platform, you know, work with, with Crucial Cloud and, and trying to, you know, figure out what, what we wanted to build and for who.

1:14:42And then, you know, we kind of, you know, with the, with the launch of ChatGPT, it really sort of catalyzed just massive investment and attention to, you know, purpose-built GPU infrastructure. And we had done that from the ground up all the way from energy data centers, as well as managed infrastructure as a service at the software layer. Did you lose any sleep over the DeepSeek news or were you Jevin's paradox pilled from day one and you knew that it was just going to keep going? Yeah, I think just like the way that got spun up in the media was like so misdirected. It was very pro-China media.

1:15:22Yeah, it was crazy. Like very early on. I completely agree. I think anybody, everybody was like, wait, like there's no chance this was like a couple of hobbyists that had like a couple of GPUs in their garage and they trained this model off of like, that's just like not what happened. They did this training run with scraps. With scraps in a cave. Powered with a bicycle. Yeah, yeah, totally. Yeah, it was a couple of guys with pens and paper just. But I guess the bigger question is, it does feel like we're somewhat shifting from a pre-training to an RL environment. The scaling laws are holding in the macro, but it seems like there's a series of S-curves in terms of the different training and improvement paradigms that lead to just better products.

1:16:08And so, yes, we can do another 10x increase in pre-training, but maybe we hit a data wall or there's some problems there. What are you seeing in terms of trade-offs for demand on the data center side as we go through these paradigm shifts in terms of what's important to create a really, really performant AI product? Is it just we're shifting from training to inference and that doesn't even affect you? Does it affect you? Do you need a different build out for a large training run versus just mass inference of complex models consistently forever all the time because demand is so high, but it's smaller models all over the place?

1:16:48Like, does any of that affect the way we build data centers? I think it does affect some of the ways that we build data centers. But to the question of slowing demand, I mean, the conversations that I'm involved in, there's like, if anything, we're seeing demand accelerate. and for bigger, larger scale clusters. And just the overall demand is increasing quite a bit for inference as well. And I think you see kind of this transition where folks will use a very large AI factory for a training run that gets some state-of-the-art model. that infrastructure is still useful for a long period of time to serve both inference workloads as well as any of these post-training, test time compute scaling, chain of thought reasoning models that are really basically taking inference queries and then thinking about them and sort of playing out a whole bunch of different scenarios and then coming up with better, smarter, more intelligent answers.

1:18:02But, you know, and I think that's like the crux of, you know, the infrastructure. It's like what we're building are these AI factories, right? So they are factories that manufacture intelligence, factories that manufacture intelligent outcomes that, you know, are prompted by, you know, inputs from users. And I don't see any near-term shortage of demand for more intelligence. Are you bullish or bearish on sort of upstart or SMB players that want to build AI factories or data centers that see the broader opportunity and then are maybe putting together sometimes sophisticated teams, sometimes less sophisticated teams, but able to pool together capital and want to bring data centers online and just assume there's going to be demand waiting there or just assume that they can actually build something that's state-of-the-art?

1:19:02Pretty bearish. We've seen a massive influx of we call them two guys in a pickup truck where I got my cousin Lenny who has this plot of land out by his ranch and there's a power line that goes through it and he knows someone that works at the power, you know, just like put up a barn and throw some racks in there. We're a business. Hit up a hype. Hey, I'm a Google shareholder. I'm going to just call up Sundar. We got a contract. No problem. Yeah, I got some 3090s in here. I got a couple 1080 Ti's right over there. A couple of propane tanks. Yeah, propane barbecuing GPUs. I think the thing is that these projects, they're so big that the capital investment to make them happen is so massive.

1:19:59And you're seeing people speculatively build smaller scale stuff. But the really big stuff that's a couple hundred megawatts or gigawatt plus, you really need credit to make it happen. I said it before, but credit is the unlock to all of this infrastructure getting built. And we have companies with the greatest balance sheets in the history of business that are going all in on this technological paradigm shift underway. And with that, you can unlock a lot of infrastructure capital to make all of this happen. But, you know, if you're going to speculatively, you know, spend, you know, 20 million bucks kind of trying to build an AI factory, you know, it's like shooting a BB gun at a grizzly bear, you know, you know, you're not, you're not, you're not.

1:20:55Well, yeah, at the$20 million scale, you can get a group of smart people that can make a deck. And then investors are going to see that and be like, I want to make money on this AI thing and be like, well, yeah, we'll throw 20, 30, 40. Chase has the best animal-based metaphors. Because I remember we were at that nuclear conference and you said that the demand for energy is so high that companies would burn whale oil if they could. And for some reason, you keep coming back to these. But the BB gun at the grizzly bear is great. So I'm sure you work with hundreds of different vendors for different components, parts, etc.

1:21:32Where do you think there is major supply chain risk or shortages? Where would you like to see more investments? I was about to ask this. We heard something that a large portion of the transformer supply chain, not the algorithm to the transformer, where the physical infrastructure comes from China and maybe there's a risk to the supply chain with the trade war there. Would love to know what the key inputs are outside of, we all know power, we all know NVIDIA GPUs, but what else could we be constrained on, whether it's cement or transformers or copper? I don't even know. Lay it out for us. Yeah, I mean, you know, the bottlenecks move around.

1:22:13You know, the bottlenecks for AI infrastructure builds kind of move around. You know, you sort of had this moment of infinite demand for H100s when they first launched. And, you know, I think Elon famously said that, you know, it was way easier to acquire illegal drugs than get an H100. And so, you know, it's rapidly shifted into energy and data center capacity. And what does it mean to actually build that stuff out? You know, high voltage transformers are definitely like a big bottleneck. A lot of that capacity does get built in China. You know, it's a diverse enough supply chain that, you know, I'm not that worried about, you know, trade war kind of impacting high voltage transformers, but, you know, they are kind of long lead time assets.

1:23:00You know, outside of, you know, there's a whole stack in the transformer side too. So like you have the medium voltage transformers, switchgear can be like a major long lead time item that Crusoe actually started manufacturing in-house. So we have factories in Tulsa, Oklahoma and right outside of Denver, Colorado. When you say switch gears, that like network switches, like ethernet routing or something else? No, sorry. It's electrical switch gear. This is basically like your electrical room that has all of the breakers that feed into the actual data halls that are powering the... Yeah, it's like a power strip you plug it in the wall you get six outlets out of the back kind of like that but like the big version of that is that right it's kind of like you know it's kind of like your uh you know it's kind of like your uh your breaker box in your house okay got it you know you trip a breaker yep you gotta go down and you gotta go flip the switch it's like that at you know a gigawatt scale data center something yeah that makes sense um uh but switchgear is definitely a bottleneck sure Chillers is another big thing.

1:24:07I think an interesting trend in data centers right now is with the introduction of the GD200, the new NVIDIA chip, basically you're seeing this massive transition to liquid-cooled computing at significant scale. So a lot of the government labs and high-performance computing communities have been experimenting with things like immersion cooling, single phase and two phase, as well as water cooling DLC for decades. But no one's ever done it at the scale that's unfolding right now. And the reason it's happening is because you just have so much energy density, so much heat being produced by these new NVIDIA chips as we move on to more advanced architectures that there's simply just not enough heat capacity to basically move that heat off the chip from a traditional aluminum heat sink or something that's so big that it just...

1:25:12Can you talk about the life cycle of water in some of these AI factories? We had somebody on the show, I don't remember their name, but I do remember that they said, we don't have enough water in Abilene to run these data centers. and that didn't quite feel correct. So I'm not going to call them out. Is that a bottleneck? No, it's not. It depends on how you design it. So we've tried to, you know, I think water can be a very sensitive topic depending on the communities that, you know, you're engaged in. And Crusoe's always tried to be a phenomenal partner to, you know, the local communities that we're working with.

1:25:51So, you know, the way we've designed our AI factories is what's called a closed loop architecture. So that means basically you have cold water that flows into the rack and it flows over the chips over this, you know, through this copper pipe. And you have this heat exchange between the silicon that goes through the copper and then to the water and then hot water sort of exhausted from the rack. that hot water that then goes out to a heat exchanger that's a chiller that's outside. And it's basically just, you can kind of think of it as like a massive maze of copper pipe. And then you like blow air over it.

1:26:35You try to blow cold air over it. And then the heat basically gets exhausted out of the water. And then you basically have cold water from that that then feeds right back into the system. So we have 1 million gallons of water per building. We only fill it one time, right? It's not like we're using a million gallons per hour. Yeah. This is what the media has implied is that like every time you make a cute Studio Ghibli image, you're like dumping a gallon of water. It's like, yeah, a gallon of water might flow over the chip while you're doing that, but then it flows over the next one and the next one.

1:27:06This is recycling, which is not. Yeah. And at that kind of scale, even a million. So there are architectures you have a fresh water supply and that is, you know, you are consuming a little water in that scenario. But in our case, you know, we've designed it with a closed loop architecture that, you know, is like you fill it one time, then you're done. Let's hear it for closed loops. I love closed loops. Last question I have, how do you evaluate your pipeline? I'm sure you're a very popular guy getting, you know, phone calls and emails from all over the world, but you're building physical infrastructure.

1:27:40It's not like you can just copy and paste, you know, what you're doing in Abilene or some other areas, you know, a million times. So I'm sure you have to be pretty. Yeah, I mean, we're trying, you know, we are, we have a couple other projects that are underway that, you know, hopefully we'll be able to talk about, you know, more soon, but, you know, similar scale or bigger is kind of like, you know, what we're seeing. So, you know, a lot of demand unfolding within the ecosystem. And, you know, I think we try to be thoughtful about our partners. I mean, we really ultimately want the space to be successful.

1:28:18You know, I view AI as a generational opportunity to transform, you know, human prosperity around the world. And we just want to help make that happen. And, you know, we don't think any one company is going to do it alone. we want to help support the entire industry in terms of uh making this technology successful scaled and and really rolled out to the masses makes total sense fantastic i have a ton more questions but we'll have to have you back on because this was a fantastic conversation we'll talk to you soon chase thanks so much for stopping by cheers great to have you take care before our next guest let me tell you about ramp ramp is uh ramp.com time is money say both easy use corporate cards almost tears earlier tears of joy yes using ramp travel it is the most amazing product i know this is such a shill that no it's not a show it was off air it's it's an incredible it's an incredible experience it is a beautiful thing that i can get imo gold medalist to make me my travel business travel app yes any i mean when we talk about software that is lacking you You just, the most obvious example is like the airline app, right?

1:29:25The airline app is notoriously buggy and you're not logged in. Ramp travel saves your, all of your information immediately. It shows you all the flights across everything. You just click one button. It just books it. And then it's just, boom, you're just booked. No chasing receipts. No chasing receipts because it happens inside of ramp, which is amazing. But even aside from the expensing, even if I had to do something else on the expensing side, But just the experience of actually booking on is so much easier. It's like, I mean. I wish we had it on camera because it was a really special moment, John.

1:29:57Leaked, leaked, leaked. I'll be right back. Anyway, our next guest is coming in the studio. But first, let me also tell you about Figma. Think bigger, build faster. Figma helps design and development teams build great products together. Go to Figma.com to get started. excited um we have our next guest coming into the studio uh factory ai matan welcome to the stream how are you thank you so much for having me i'm good how are you i'm great uh thanks so much for hopping by uh can you kick us off a little introduction on yourself and also uh the company and the news i know that there's big news coming up yeah absolutely so uh i'm matan ceo at factory here to share a bit about our latest launch.

1:30:42We released Droids. Droids are not just your regular everyday coding agent. They are full end-to-end software development lifecycle agents built for the enterprise, big enterprise, not just startups. Although, you know, we've in the last 24 hours seen a pretty big explosion of usage with startups, which has been really cool. Yeah, that's, it's It's been fun. Talk to me about the evolution of AI agents and coding agents. I remember there was a company a couple of years ago that was training their own foundation model. Then the regime seemed to shift to, okay, we're not going to do any pre-training on code specifically.

1:31:25The foundation model companies got that covered, but we will do a bunch of post-training. Then it became more, maybe there's some RL fine tuning. Maybe there's just some reasoning that we're doing. Now it feels like a lot of the coding agent companies are just kind of like, we're rappers, but we're still printing money and rappers and unnecessary pejorative. It's actually the best way to build this business and it's helping us. So where do you fall and what is your take on the different paradigms? I just got here. I got to compliment Matana. I know, I know, I know. Fantastic suit. It looks great.

1:31:57Thank you for respecting. When you come to the temple, when you come to the temple, you must be dressed appropriately. Yes, yes. Thank you. But yeah, great question. I think it's been interesting how in the last two years, there have been so many trends within the kind of like sub sector of AI for software development. You're spot on. There was a really big trend, especially if you measure it by the capital that was put into it in terms of companies coming out. They're saying that they're going to train models in particular for code or fine tune models for code. There's been, I mean, there's been quite a few who raised to the tune of like half a billion dollars to train their own models.

1:32:38And I think something that a lot of people were saying at the time or maybe whispering because they don't want to offend those who put half a billion dollars in was code is a core competency for the foundation models. They will not allow a startup to fine tune their way to having the best models. As you see, in general with code, it's like alternating between OpenAI, Anthropic, Google, XAI in terms of the number one spot in code. And so I think it – I guess it took a lot more money than it should have. But I think people come to the realization that fine-tuning and RL for code-specific models will be won by the foundation models.

1:33:16Is that also just a function of like code is the internet? The internet is code. Code is on the internet. And so if you train an LLM on the internet, like you're going to learn like out of the box, if you do a robust, large scale training run on web text, you're going to get Reddit answers. And you're also going to get a pretty good model that can code. And so unless OpenAI says, let's not train it on code, it's just you're going to lose to them because they're putting the most, you know, energy, the most cycles into training the big model. And so you just get that as a byproduct. Yeah, that's, that's spot on.

1:33:53And it's actually, I would go even further to say that, and I know, there were papers on this in around like 2022. I'm sure there have been since but there is a direct correlation between how good a model is at code, and how good it is at other general purpose tasks. That makes sense. So it is just it is like it is, you know, table stakes for the foundation models to be the best at like raw coding. Yep. So exactly right. So how do you build a business without getting rolled? Yeah. How have you guys navigated just partnerships with the labs in general where it's your kind of frenemies? You know, it's like, hey, we can work together.

1:34:31But at some point, somebody is going to try to kill the other one. Yeah. No, this is a really good question. And it's very top of mind. I mean, I think the reality is the closer you are to the zero to one or like the vibe coding or the like less technical your audience and the smaller the company, that's really where the foundation models are like about to sweep it up or already have. So it's like, you know, building apps from scratch, building websites from scratch. That's something where the model providers will basically get it for free, like plus or minus some deployment details in terms. Yeah, you remember with pre-ChatGPT, there were a bunch of players that would generate copy for you based on the OpenAI API, and their revenue rocketed.

1:35:18And I remember it was the same behavior of posting screenshots of the revenue dashboards, and then those screenshots stopped being shared post-ChatGPT because all that revenue just went away. As Shakespeare said, these violent delights have violent ends. there was a really there was a really great great tweet I saw the other day I can't remember who posted it but it was you know a lot of people are talking about these like record-breaking like runs to like you know large ARRs and what's going to be coming soon is some record-breaking churn for a lot of this like monthly these monthly subscriptions so that'll certainly be interesting to see but yeah so that's the part of the spectrum where the foundation models are best poised to tackle is that like zero to one less technical use case.

1:36:07For us, we've been focused on the enterprise from day one, because one, it's really not sexy. And so it's not like conducive to like viral demos and all that hype. But that's where a majority of developers are getting a majority of their pain, like dealing with like COBOL, dealing with these like 20 year old code bases, migrations, refactors. And in order to handle that well, you can't just like one shot with like a chat GPT or a codex or a cloud, you need to have deep integrations with their code base. Oftentimes it'll be like multi repo integrations. You'll need to understand not just where the code is now.

1:36:44Nice love the little thumbs up guy. Not just not just where the code is now, but how it got there, what the best practices of that org are integrating with tools like Jira, Google Drive, Century, Datadog. The kind of first principles thinking there is that in order to produce the quality that an engineer who's been at that company for like 10 years would produce, you need to have access to the same information. And that information sometimes requires some really ugly integrations. You need to kind of get the workflows that are a little bit more specific to these large archaic orgs, which is a little bit further afield from what the foundation models are working on right now.

1:37:25Talk to me about synchronicity, asynchronous coding agents versus synchronous, co-pilots versus autonomous agents. Where do you think it's going? Do these lines blur? Are they distinct product? It feels like OpenAI has three coding products now. I can go to O3 and I don't even ask it to write code and it just does. I imagine that the amount of code that's being written for non-technical people who don't even know that code would be useful in giving them an answer to a question is just skyrocketing right now. in that product. Then there's Codex. They also have Windsurf now. And so you can imagine a few different bites of the apple there.

1:38:09Are they indexing the market? Are these distinct different markets? Will there be one power law outcome in the coding space in terms of paradigm? Yeah, great question. I think the first order bifurcation that'll happen is between the non-technical audience and technical audience. So for non-technical, being able to spin up apps on the fly is like cool and fun and that'll continue to be something that's useful they're never gonna yeah exactly right and they're never gonna really be that concerned with like going too deep into the code itself they'll just like i want a to-do list app make it for me okay great um for the technical user i think that's where it gets a lot more interesting um basically and maybe maybe this is just a quick like kind of philosophy that we have at factory every transformation shift has a very clear behavior change associated with it right so like internet had people uh getting most of their information from like tv newspapers books to then like going on this console on their desk and like you know getting all their information there um mobile had people you know walking around like heads up to now like you know walking around on tick tock subway servers all the time right these are very visceral obvious behavior changes yeah and yet everyone talks about ai as it's you know the the largest one to come it's going to put these other ones to shame.

1:39:22And yet, if you look at the most used products, the most used AI products right now, there is no new behavior. Like chat GPT perplexity, that's just Google with better results. The silhouette of that behavior doesn't actually look that different. Similarly with a tool like Copilot or Cursor, the way that software development is looking, the behavior is still the same. You're still in this IDE, which was built for the world where humans wrote 100 % of their code we're quickly going to a world where humans will write zero percent of their code and our take is that their behavior will fundamentally have to shift then and not like in some iterative approach where you like iterate your way from an ide to whatever this new thing is but our take is instead you need to build that from the ground up and this is kind of a long-winded way of answering your question about the like async versus sync the reality is in this future developers are going to be natively working with agents and so they'll need to like dynamically adjust between if they're collaborating with an agent to like look through their code base and understand how they should plan some new feature and then having a good plan for it and now like firing it off delegating it to these to these agents um and at the same time it's going to put more emphasis on the testing because if you just shoot from the hip a ton of agents or in our case droids um then you're going to have a ton of code to review before you release it to production because ideally you're going to review it right but that's kind of a depressing world where okay hey, we don't write any of our code, but now you just need to read like thousands and thousands of lines before you can ship anything.

1:40:51So if you as the developer have better tests, and you know, hey, if it passed these tests, I don't even need to review it, because I like expressed all of the constraints that I had through these. So if it passes, great, let's ship it. There's kind of this new emphasis on testing to kind of enable that more delegative workflow. What's the secret to avoiding churn in the enterprise? Are you trying to Inc. multi-year contracts up front? Are you just playing the same game as everyone else and getting a bunch of experimental budgets because money is money and that seems to be the meta right now, but I imagine that you're thinking about this or at least messaging this to the community.

1:41:34When a competitor gets to a customer first, are you kind of hovering with the understanding that like, hey, you know, maybe they're, you know, if we can really show that we're significantly better, we could win. And is there, is there like a little bit of a price war here? Because you imagine that you come into an enterprise and you say, hey, this would cost you so much to do with like Accenture. And we're going to do this, this replatforming of your COBOL application to.NET or something or Python. And, and, and you would spend 10 million on that. We'll do it for 8 million. And that's all of a sudden, all of a sudden, that's like 90 % margin for you instead of like 50 % margin.

1:42:14But then your competitor comes in and says like, well, we'll do it for 7 million. We'll do it for 6 million. We'll do it for 5 million. And so it seems like there's some sort of price war dynamic that might happen. Walk me through all of that of like winning in the enterprise financially. Yeah. Yeah. That's a great question. So first of all, yes, we do year-long contracts just because it's important to have that mutual commitment. And in particular, because adopting the tools is not enough. Like I cannot tell you how many like CIOs and CTOs I've spoken to who have adopted like the hot new AI IDE.

1:42:42And then you ask them the question that they don't want to answer, which is how many people are actually using it. And it ends up being like 10, 20%. And of those 10 to 20%, a lot of them are just using it like the IDEs of old. So they're kind of like adopting these new tools, patting themselves on the back, being like, look, CEO, we did it. We adopted AI, job's done. Give me the big bonus. but the reality is they're actually not getting any productivity improvements. And so part of why we do these longer term commitments is because one of our core competencies is not just having the best agents in the game, but also helping them.

1:43:15Droid. I love it. I love it. There we go. We're also helping them adopt their behavior patterns. Because that's the thing is like you could have a tool that's 50 % as good, but if you have twice the adoption, then you're now at parity. And so I think it's just a lot less sexy because all the like, you know, great engineers who are coming out of Stanford want to work on spinning GPUs and all that stuff talking about like behavior change in the enterprise that's like you know like you know they don't want to think about that but that is where the ROI is going to come and also John to your question like the way we actually get in and do these deals is focusing on those deliverables about this was scoped out to be four months and we did it in two months or one month or two weeks that's ROI that actually matters to like the c-suite when you talk about like we shipped tests 10 % faster, it's just so like, are we 20 % more lines of code?

1:44:04It's just so amorphous and so not tied to real business outcomes that if we can come in and actually tie things to like things that are shipped per quarter or, you know, pulling in dates for certain deliverables, that's where it's just like, it's so frictionless because it's not really existent elsewhere in the market. I was texting with Adam and Ben from genius, one of your investors, and they said to ask you about a fateful walk that you had with Sean McGuire. Yes, it does. This is this is in like the founding history of factory. Basically, two years ago or two and a half years ago, I was doing a PhD in theoretical physics at Berkeley, which is what I was doing prior to factory.

1:44:47And you just thrown it in. you want to kind of take the easy path in school? Yeah, exactly. Yeah. You know, just some fun string theory, which to be fair, it is really fun and beautiful, but decided it wasn't a path for me in particular, because, you know, to be a good physicist, you kind of need to thrive in isolation, just like in your room alone, like reading papers. Did that for like 10 years, was really stubborn. It's kind of a long story, but ended up reaching out to the Sequoia partner, Sean McGuire, because he also used to be a string theorist and uh he ended up you know going into entrepreneurship sold a company for a billion dollars joined sequoia saw like a random podcast with him and i was just like i've never seen like another physicist who has somewhat social skills like let me hit him up and get some get some life advice from him ended up going on a walk together on this walk he said matan you need to drop out of your phd and you should either join one of my portfolio companies and just like work on glue not a thumbs up let's go or you should join you should join X because Elon just took over and you'd have to be a badass to voluntarily go there or you should start a company and so that was eight days later dropped out of the PhD and started factory so congratulations very cool and yeah thanks for stopping by this was fantastic and thank you for giving agents a cooler name I love droids we gotta get ourselves some droids Yeah, I've got to get you guys some drones.

1:46:14Thank you, guys. Great, chairman. Have a good one, Matan. To be of automation, let's tell you about Vanta. Automate compliance, manage risk, prove trust continuously. Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation, whether you're pursuing your first framework or managing a complex program. And next up, we have Johnny from Muon Space. We're talking about data centers with Chase Lockmiller in Abilene, Texas. Now we're going to space and we're putting the data centers in space. Very excited to talk to Johnny.

1:46:44So welcome to the studio. Johnny, break it down for us. And I'm going to let Jordy do the intro on this. And if you're on your phone, would you mind rotating it 90 degrees so it's widescreen? Thank you. There we go. Jordy, do you mind kicking us off the intro? What's going on? Great to have you. Where are you calling in from? Randomly, I'm actually down in Yael's neck of the woods in Long Beach. I'm hanging out with a friend at Bass today. Nice. Nice. Nice. So, yeah, why don't you give a quick intro, background on the company, yourself, all that good stuff? Yeah, sure. So, Mewon Space, we're a 150-person startup in the Bay Area.

1:47:19We're building a platform to deploy large numbers of satellites in Constellation format for a lot of different mission types. We were founded in 2021. It's been a pretty big rocket ship ride so far. It's been really fun. And my background going back a ways, I mean, I've been in space for, you know, well, longer than I care to admit, including I was an intern back at SpaceX in the very early days, like 2003 and 2004. I was part of I was the chief engineer at Skybox Imaging, which was the kind of first venture backed satellite company kind of space before space got cool. Very cool. And so, you know, taking a lot of the lessons learned and kind of things I've seen from those experiences to the new company.

1:48:01Awesome. And then break down the new company. Yeah. So, I mean, I think the way to think about this is, you know, a lot's changed in the last decade of space. Traditionally, if you wanted to go do something in space, it required a lot of sophistication. You had to have it literally a team of rocket scientists. You often had to have a lot of deep expertise in everything from, you know, ground stations to launch to avionics software. where barriers to entry have come down in a lot of ways, but a lot of that complexity remains. And our kind of goal is to really extract a lot of that away from a customer that wants to do something in space and doesn't want to have to go build that full vertical technology stack for every new use case that comes up.

1:48:42And I think that was a big lesson we took out of Skybox, is we kind of built two companies under one roof, a satellite company and a data company. And the future should really be about a data company being able to go build a data business without having to deal with all the complexity on the space, the ground, the operations, the hardware, the integration side. And that's really what we're trying to solve. I want to know about the potential of data centers in space. It sounds like a crazy idea. I know some folks are working on it. Just kind of like what's your high-level take of the progress? We've been tracking, you know, a dollar per kilo to orbit.

1:49:22It's been falling. but recently there's been setbacks in different programs and there's more competition and there's a lot of different dynamics going on. What do you think the key milestones are to get us to a future where we're really doing, you know, mass manufacturing, big, you know, mega scale projects in space? Yeah, well, let me start off. I mean, I think you guys are hitting on the right metric, right? It's like the hardest part of this is what does it cost to get a kilogram in space because everything else kind of derives from that. Ultimately, if you want a certain amount of power, if you want to be able to put an aperture in space to do communications, whatever it is, like that's really driven by what it costs to get it there in the first place.

1:50:02And I think it's important context to kind of see how far we've come down that path. I mean, largely driven by SpaceX over the last decade. And, you know, I like to tell the story of, you know, about 15 years ago at Skybox when we were trying to launch these small satellites, we were literally going to Southeast Russia and launching satellites on converted Russian ICBMs. I mean, they were popping out of the ground and putting satellites into space. And that was the only way for like a Silicon Valley venture funded startup to go put something in space. Wait, wait, hold on. Did I hear you correctly?

1:50:31Like the ICBM actually launches from one of those missile silos in the ground and makes it to orbit? Pops out of the ground and goes to orbit. I mean, there's videos on YouTube. That's crazy. If you search the NEPR on YouTube, you can see this. And it's crazy. I mean, you know, and that's that's what it took before SpaceX kind of like revolutionized the launch business And even at that we were spending something like ten times as much Per satellite or per kilo is what we can go buy on a transporter launch today So there's been at least one and arguably two orders of magnitude improvement in the last call it decade on Kind of what it means to launch things to orbit.

1:51:08I think if you imagine that happening again another order of magnitude it, you know, again, it's going to dramatically change the way you think about feasibility of some of these things, like putting very, very large power hungry things in orbit. We know how to do the solar, we know how to do the structures, we know how to get, you know, we know how to make electronics work in the radiation environment, which is always a concern and do it reliably. And so really, ultimately, I think it's going to come down to unit cost. And what does it actually take to do that. The great thing about space is you have virtually limitless power, the sun, you know, you can, you can go into orbits where you're in the sun all the time.

1:51:45So it's not like solar on earth where you're going in and out of eclipse or in and out of night, like you can be on all the time. And then you have this cosmic background of three Kelvin cold sink that you can go dissipate all the thermal energy you need from running your electronics and stuff. So it's actually in a lot of ways, like sort of an ideal environment to do this if you can actually get there. So yeah, that's the kind of way I think about it. It's interesting. It feels like SpaceX has commoditized launch and is kind of the power law winner there. Obviously, there's other companies that are competing, but they've standardized launch.

1:52:19They've also standardized Starlink. And then we talked to the Endurosat founder about standardizing a satellite bus platform and kind of getting to the mass manufacturing, less bespoke, less customized, less hand-built pieces. What else are you seeing as kind of critical pieces of the space supply chain that need to be standardized or you might be trying to standardize? Where are the pieces in the supply chain to do the things that you want to do where it's still like, oh, that has a long lead time or that's hand-built or that's way too customized. I get what I want, but it's an exquisite system.

1:52:56It's too expensive. Talk to me about the supply chain. Yeah, and I'm actually going to give you the answer you asked for and then an answer I want to answer, so I'll do both. On the supply chain side, you know, I think there's a lot of progress that's been made. Like our satellites look very much like, you know, especially things like the electronics look very much like what goes into an EV right now. The batteries, a lot of the electronics, a lot of the individual kind of semiconductor parts heavily leveraging other commercial industry. So it doesn't look like a traditional space supply chain.

1:53:25I think that problem more and more is solved. I think Endurosat obviously is doing very similar things in that way. Some of the places where we really see that that has not yet happened, one is in more on the payload side. So things where like, you know, you think about the spacecraft bus, Endurosat's building buses. You know, I do think there's a path to that becoming very standardized. But every mission you put in space has a bus and then it has a moneymaker. It has something that's actually doing what you need it to do, whether it's communications or remote sensing or beaming power. If you're trying to be power, if you're putting a data center in space, it's got a payload of compute or whatever.

1:54:01And what we see in most of the kind of traditional space missions is that that now has become kind of a hardware bottleneck. It's like, how do you actually get the payload you need to solve your ultimate business problem built in quantities large enough to deploy a constellation? It costs low enough that you can do it. And I think there's a lot of movement in the right direction in that way, but we're not there yet. And the other question that I kind of answer that you didn't ask is, I think the other big bottleneck outside of hardware is still software. So like, you know, we spent, there's a lot of talk right now on hardware supply chains across a lot of industries, including space.

1:54:38I think there's paths to address that. What is still very true in space is that these are very complex autonomous robots with global networks that are having to communicate in inside communication, outside of communication. And software is really the glue that holds all that together. And in a lot of ways, the integration of all these things and the software integration, the hardware integration is still the hardest part. It's really making these large, complex systems work as a whole. And I really think that's a place that there's not being enough emphasis put in the industry right now. It's something that we're really trying very hard to solve with kind of the kind of core hardware software stack that we're building.

1:55:13so that instead of, you know, for every new use case you're trying to solve in space, you're going in from the beginning having to do this crappy new integration of a bunch of parts of hardware and software that aren't actually designed to talk together. You have a platform that, much like a data center rack today, everything from the lowest levels of hardware to the high-level application software is designed to interoperate. It's designed to be pluggable. And so, you know, when Google's putting data centers in the data center, they can go take a bunch of stuff off the shelf. Every rack doesn't look the same, but they have a bunch of parts that are interchangeable.

1:55:45They can go throw that in, plug them in. The software works, the hardware works, et cetera. We've got to get to that with space, too. Yeah, I mean, we've done that a ton in data center, like Cat5 cables, Cat6 cables. These are all standardized. Even Facebook open source their blade design for their rack mounts. And there's a lot of other ecosystem tools. Last question from me. are you tracking the SpaceX Starship progress? There was news today that Elon Musk is winding down his relationship with the government as a special government employee. That was a 130-day mission, probably spending fewer nights at Mar-a-Lago, more nights in the SpaceX factories.

1:56:26Starbase. Are you tracking that? Are you excited about that? And I guess the big meta question is, Elon has always seemed interested in getting to Mars, but he's willing to go to Mar-a-Lago if that's what gets him to Mars. He's willing to go to Starbase, Texas if that's what gets him to Mars. So it feels like this might be a moment where he's shifting his focus. The mission's the same, but he's shifting his focus from regulation to engineering challenges. Mars is back on the menu, boys. Mars is back on the menu. But what has your take been on the recent Starship or SpaceX news? Yeah, I mean, I guess I would start by saying look it you know building rockets is really fucking hard Like I mean, I think anybody that says it's not is crazy.

1:57:10It is rocket science I think we've almost been lulled in complacency by how successful SpaceX has made it and like how easy they make it look But I mean Starship is a crazy complex complex rocket unlike any that's ever been built So like at some level I feel like these setbacks should be expected and people shouldn't act so surprised I think there's I don't have enough inside information I know if like forward progress is being made or sideways sideways progress, there's more focus, whatever. But like this stuff's really hard. So I don't think it's that surprising that there's setbacks. You know, I think that the the key thing for the launch unit economics, which I think ultimately, you know, Starship, we're all hoping it's going to get us another order of magnitude on unit economics.

1:57:49So much of it is not even about scale or size or technical performance or capability. It's about launch cadence. It's about rate. It's about flying. it's like airlines, right? If you bought a 737 and flew it five times a year, nobody would be able to afford to fly. But because it flies 10 times a day, all of a sudden that the amortization of that fixed asset over those flights makes a ton of sense. So I think for Starship to really go, what we need is we need applications that require Starship, like Starlink has for Falcon 9, that drive us to launch it three, four or five times a day, like you fly 737s.

1:58:25And I think if that happens, like it will, we will get another order of magnitude in the unit economics. And I mean, maybe tying the loop back to the original question, you know, you start thinking about applications like deploying data centers in space where you're putting thousands of things that are each many, many tons each into space. And there's like a really core first order economic driver requiring that needing that to scale AI training and these types of workflows. That's the kind of thing that I think could drive the demand that you need for something like Starship to really get another sort of order of magnitude in the unit economics.

1:58:58I don't know if I really answered your question, but that's the way I think about it. No, man, no, it makes sense. Well, thank you so much for stopping by. This is great. Come back on again soon. Yeah. Yeah, you bet. It was great talking to you guys. We'll talk to you soon. Bye. And while we're waiting for our next guest, let me tell you about Linear. Linear is a purpose-built tool for planning and building products. Meet the system for modern software development, streamline issues, projects, and product roadmaps. And Linear has agents. I'm lucky to use for over a decade now. Over a decade on linear?

1:59:27That's crazy. I'm surprised the company's even a decade old. No, no, no. Sorry, half a decade. Half a decade. Half a decade. Because a decade ago, you were in middle school? I was about to say the better part of a decade. Yeah, yeah. Okay. Well, we have our next guest. David from Retool is coming in the studio. I'm very excited to finally talk to you. I was at your YC Demo Day, Alumni Demo Day. And no one's going to believe this, but it's true, so I just have to say it. And you were the one company that stuck out, and I was like, that company's going to be amazing. And I didn't even think of myself as an angel investor.

2:00:02I should have just been like, please, let me invest. Anyway, it's been fantastic to watch the arc of Retool and everything that you've built. So congratulations, and thanks for joining the show. Thank you. Great to be here. Huge trainer of TPPN. So excited to be here. It's great to have you. Can you give us a little update on the company, kind of define the different eras, what the product is and where it's going? I know, obviously, we're in a period of transformation with artificial intelligence. I'm sure we'll go into all of that. But what has the bread and butter been for the last couple of years?

2:00:36Yeah. So for the first few years of Retool, Retool is like Legos for code. which is we allow you to build a sort of higher level building blocks and you can code faster. You can piece them together and you can build software. And the weird thing about Retool was that we always focused on this category that we called internal software, which is extremely not sexy because no one ever thinks about internal tools. But surprisingly, it's something like 50 to 60 % of all the software in the world is actually internal facing. And no one thinks about it, actually. And so that's where Rital started. And the idea behind that.

2:01:10And engineers are not like, oh, I want to work on the internal tool. It's typically like, you know, you put the engineer that's not cracked, you know, and be like, yeah, just focus on this dashboard. Or just someone who's just like grinding away. But yeah, I mean, this was the big, we were talking to Joe Weisenthal about this with the question of like meta training llama. And there's a bunch of places where LLMs will instantiate themselves in consumer facing products across meta. but also they have a massive amount of internal tooling that can benefit from AI. And so not having to fork over endless boatloads of cash to just check, does this have profanity in it?

2:01:49One billion times a second, right? It's like that could be a very big open AI bill if they don't train that internally. And so, yeah, I think the dark clouds over the internal tooling world is something that most people might not be aware of, but you've obviously been living in that. So talk to us about the growth of the company. What's the mix? Is this all enterprise-driven? Is there small and medium businesses that benefit from Retool? Kind of what's the bread and butter on the customer side? Well, if you look at the homepage, it's like in the logos. It's like, oh, so you guys work with every company.

2:02:23Every company. Yeah, it's pretty cool. I've worked with companies ranging from the U.S. Army to the U.S. Navy, the state of Utah, all the way to small startups. Two-person companies, five-person companies. You don't get put on the American Dynamism market maps. I'm going to start demanding. Yeah, we've got to put them on the American Dynamism. Put some respect on Retool. It is the backbone of the U.S. military. No, seriously. I mean, there's internal software everywhere. But yeah, maybe it'd be good to shift into kind of like the AI moment. It feels like Retool was kind of vibe coding without the vibe coding meme or without the – It allowed you to vibe code without the actual instantiation of, you know, you're not in an IDE, but you're effectively vibe coding or building something that's very quick.

2:03:11How have you been processing the AI boom? When did you first think about implementing LLMs and other AI technology into Retool and how has it been going so far? Yeah. So the really cool thing about Retool is that once you have these Lego blocks built, you can actually give them to AIs to actually use, which is really interesting. What do you believe is missing now, which is today, if you look at how many dollars have been invested in the US in AI, I think it's around a trillion, maybe a trillion and a half or so. But if you actually look at how much revenue there is from all AI products across all companies, I think it's like 20, 30 billion.

2:03:55That's pretty crappy ROI. That's a 3 % hole. That's terrible. And the question is, is it all a bubble? Yeah. Can we figure out some use case for this LM beyond just chat? Yep. And if you look at where we are today in the consumer market, in the enterprise market, it pretty much is all just chat. I think if you look at that$20,$30 billion of revenue, I want to say something like 80 % of it is chat GPT revenue plus cloud revenue, which is awesome. But chat is pretty limited. I mean, is chat going to grow 30x where we are now? It's hard to say. I think probably no is the answer. And so what we think is really missing is a way to actually leverage and use LLMs to actually go automate labor.

2:04:41the weird thing about this is that LLMs are actually plenty smart already if you, you know, I use chat a lot I'm sure YouTube both use chat quite a bit as well it's basically AGI at this point, I mean, it was the classic Turing test thing of, you know blue past that way past that yet AIs aren't doing anything yet so it feels like this is a big disconnect what we're here to really solve is can we actually allow AIs or LMs to actually do things in your business. So right now, it's just chatting back and forth. Can we actually allow it? Can we allow the US Navy, for example, to say, hey, it's not just using ChatGPT to answer some questions, but actually ChatGPT actually, or LMS, actually do things in the US Navy, whether it's approving orders, whether it's approving plans, whatever it might be.

2:05:29That, I think, is the next frontier for AI. It's AI that actually does things. I think we're almost It's pretty cool. That's what we're working on. Can you help me work through this question of, there's like this AGI, ASI narrative, and then like, how does that not destroy every software company? Because I'm seeing MCP servers spun up left and right. And my question keeps being like, if the AI is so smart, why does it need a server? Why can't it just use HTML and UI like any other human? right? At a certain point, is there a world where I say, I want to sell a t-shirt online, and instead of spinning up a Shopify store, it just writes payment interop code.

2:06:18It doesn't even use Stripe. It just builds Stripe from scratch at a moment's notice if it's so smart, and it just works for a billion human hours added up at 160 IQ, and it just builds me Stripe for this one t-shirt that I want to sell. Like that feels like it aligns with this ASI. That's also maybe not even the best example because of the regulatory component. Sure, sure, sure. But we also had Steven on from the co-founder and CEO of Lambda Labs. Yeah, yeah. His point of view was like broadly that you're just going to generate the software that you need and you might generate 500. And I'm sure this is stuff that Retool is already doing to some degree.

2:06:54Generate a bunch of different versions of what a tool could look like. Yep. Rank them, allow you to sort of try different versions of it before landing on. So there's like this one, there's this one world where like having retool primitives that LLMs can interact with is amazing in terms of like making sure that there's robust performance and everything gets up to speed really quickly. At the same time, you know, in the really long term, do we even need these primitives and can we just do everything from scratch? What's kind of your long term view of how this plays out? so this is i think a secret that we've actually discovered is in what cases do you want determinism versus what cases do you not want determinism that's interesting and we actually just announced yesterday that uh we have automated 130 million hours of work for our customers over the past 12 months if you divide out the math that's around It's a lot of hours.

2:07:52And I think the secret is exactly what both of you just pointed out there, which is in what cases do you want AI and what cases do you not want AI? And to give an example, OpenAI has a product called Operator. It's an agent-like thing that does things on your computer. And actually for consumer use cases, Operator is really good. But what Operator does is basically an LLM with one tool, and that one tool is use the computer. Yep. And if you want to go buy a shirt, if you want to go buy a pair of socks, that agent is really good. You know, what it'll do is you say, hey, I want, you know, socks, size medium, a navy blue color.

2:08:31It will open up the browser. It'll Google. It'll find the sock. It'll buy it. It'll use a credit card. It's done. And that's fully non-deterministic. You know, it's kind of making it up on the spot. And that's pretty good for a consumer use case. Whereas for an enterprise use case or a federal use case, for example, you actually don't want it doing that. And so I'll give you an example. One of our customers, a large company, actually uses Retool for employee onboarding. And so what they say is, hey, every time a new employee starts, you've got to go do all these tasks. Maybe you have to go send them a laptop, you want to send them a key fob, you want to do this, you want to do that, whatever.

2:09:06and if you ask operator to go do that operator says i got one tool and it's web search so how do i onboard an employee i'm going to open my browser i'm going to google how do i onboard employees how article it reads the wiki how article it's like that's not at all what you want business has very specific ways of onboarding an employee yeah and the ai actually calling those specific things because you actually don't want it reinventing the wheel all the time. And a lot of those might be gated and private and maybe not even on the open web and also very high risk from a regulatory perspective even.

2:09:46So you don't actually want the AI reinventing it all the time. Sure. So that's kind of what we mean by the building blocks. You almost want to give AI these building blocks, these tools, these MCP servers, and have AIs call them as opposed to AI reinventing it all the time. because you don't actually want AI rewriting your security policy or how do I send laptops to employees every day? Instead, you want it to say, hey, oh, let me think. There's a big storm happening in the Southeast. So for me to get the laptop in on time, I have to ship via Express versus ground. That's something you want the AI reinventing and reasoning about, but you don't actually want the AI reinventing, you know, do I use FedEx today or do I use UPS?

2:10:23What do I feel? You know, you have a contract with FedEx. You want to use the FedEx one. Sure, sure, sure. Yeah. And that's something that a good office manager, a good person in HR who's onboarding would actually think about and reason about. And you could inject that with a reasoning LLM. That makes a ton of sense. Talk to me about the evolution of the business model. Consumption versus seat-based pricing. And then going into the future, are we going to be looking at outcome-based pricing like what we're seeing Mark Benioff talk about in terms of Salesforce, where it's more like resolution-based?

2:10:55You want a job done. our agents, our tools can get that done and you're going to pay for results. Yeah. So this is a fun, I think a birdie secret too, is that I think outcome based pricing is basically designed to rip customers off. Okay. The reason why I believe that is when Mark Benioff tries to charge you per outcome, he's like, well, what's the value of what I deliver? and great, pay me that. Whereas in reality, it costs a lot less for him to actually go deliver that value. This is software though. We want 90 % margins. Come on. What do you have against high margins? I hear you. Yeah, no, it creates a, but couldn't you argue with SaaS?

2:11:40The SaaS vendor is basically trying to charge as much as possible without the person saying, oh, we're going to build this ourselves or oh, that's actually, you know, we can just hire two more people to do this. You know, it's always this dance between you want to be, you know, companies should be trying to capture all the value that they create. Right. Yeah. Cause no one would buy a product. Yeah. So the way that we're pricing is pretty interesting, which is that we price based on inputs, which is we just say our agents work for$3 an hour and that's it. If you want a smarter agent, you can hire smarter agents.

2:12:13So actually that$3 agent is a deep seek agent. So it's a Chinese agent for$3 an hour. You could go buy O3, which is a quite smart, probably the smartest right now reasoning agent. I think that's something like maybe$120 an hour or something. So it basically depends on what kind of task do you want to do. If you want to do a simple task, actually deep seek is quite good at that. And at$3 an hour, you're getting a really good deal compared to hiring human labor. Whereas if you want something really knowledge worky or something really nuanced done, maybe O3 is better. actually. But I think the innovation here is that we're charging on a per runtime hour basis.

2:12:50And we've discovered that an hour of runtime for 03, for example, is actually worth something like 40, 50 hours of human labor. If you try to do research, you can do it an hour. I mean, I probably could have been doing it honestly. So that I think is really cool. And then what happens is you actually look at how much the customer pays per task actually completed. If you compare you put our pricing to Salesforce's, for example, per ticket, resolve pricing, our pricing ends up something like 95 % cheaper, which is really, really cool. I mean, it'd be like an AWS charged you not on compute or storage.

2:13:25If they charged you on, how much money does your app make? Let's charge 90 % of that. That's a total ripple, right? So that's kind of how we think. We're trying to be almost like an AWS of labor, if you will. Yeah. How do you evaluate as the businesses evolve? I'm sure people early on, it was pretty easy to communicate even to investors. Hey, people spend a lot of time and energy and engineering resources on internal tools. We can be a big company just doing this. now I imagine as you look at the business evolving, you're like threatening a lot of different categories of software because you're saying we're going to enable teams to more easily make the decision, do we buy this or build it ourselves?

2:14:09And like maybe retool is like some middle ground. I don't know if that's the right way to think of it. But how do you evaluate kind of the scale of the opportunity now? So the internal goal that we have, and it's pretty early, but our internal goal is to go automate the equivalent of, if not actually, 10 % of the labor in the U.S. by 2030 is our goal. And the idea is that... Let's go. I love an ambitious goal. I love it. I love it. It's amazing. Here's the TAM. Yeah. U.S. labor. 10%. 10%. Yeah. And actually, the cool thing is we're on track, actually. That's amazing. Over the last few months, if you draw that now, 2030 is a long time away, but if you draw the line out, it comes to, I think, 100%, actually.

2:15:00That's incredible. It's pretty cool. But 130 million hours automated is no joke. Yeah, that's serious. Yeah, that is absolutely wild. Well, 15 minutes was not enough time. We have three more minutes. We have three more minutes. Our next guest is at 120. I have one more question. Talk to me about the knockout dragout fight happening in the foundation model space. You mentioned DeepSeek at$3 an hour versus O3 at $120 an hour. How do the other foundation labs compare? What do you do to benchmark them internally? Are the benchmarks cooked? What's the take on Meta and Llama? There seem to be a ton of energy there.

2:15:40In terms of your use case, having an open source model that you can inference for cheap or at cost seems incredible, and yet it seems like they're a little bit behind on the reasoning models. Are you optimistic there? Are you kind of model agnostic? What's your overall take on the foundation model space? It's pretty cool to see this play out, especially with our customers, because, yeah, so for us, we're theoretically agnostic. You can use whatever model you want. But it's really cool seeing what customers prefer, actually. And so, for example, with a large federal customer, you might think, oh, actually on-prem is really important, and so they actually prefer to use Lama or something like that.

2:16:21That's not the case. Actually, OpenAI is selling a lot of OpenAI, it turns out. Actually, there's a lot of traction even in sort of big federal agencies for something like OpenAI, and they actually have contracts already, and so they plug it into Redual and it just works. That makes sense. I think I would have thought, if you asked me three years ago, is the federal government happy giving all their data to LLM? No, hell no. It seems very unlikely, but it's happening, which is really cool. But you also see this obviously happening in other countries too, whether it's Saudi Arabia or Europe with mistrial, for example.

2:16:51People are getting kind of nervous about sending data to other countries' LLMs. And so it's going to be cool to see how the world plays out. It's very much not a meritocracy is maybe one way of putting it, which is we thought that when we build our agent's product with an evals framework, people would just say, let's see who does best, who does it for cheapest, and let's go. And actually that's not the case, which is i don't know how i feel about that honestly is that because like there are sdrs that are buying steak dinners for people and swinging them over or is it more like there are qualitative metrics that matter more like the nature of the contract matters matters more than just benchmark price etc like the quantitative metrics there's that but maybe another way of putting it is lms have been stickier than i thought which is maybe the branding is so important or something like that.

2:17:41Yeah, everyone just says, oh, it's just one line of code to swap out. But if you're used to the certain, like, what does UnderCarpathy call it? Like spiky intelligence. There's like certain models that spike in certain ways and you have expected behaviors. And they, yeah, they might all hallucinate at the same 3 % rate, but if they hallucinate in a specific way, you build around that, that type of thing. Okay. Exactly. Yeah. Very cool. Cool. But excited to see how the space develops. So. Well, thanks for coming on. This is fantastic. Yes. We'd love to have you back and just chat about AI and automating 10 % of labor.

2:18:11Good luck. Give us access to the internal tool that's just the labor automation tracker. I'd love to just follow along. We can put it up as a ticker on the screen. It's just like progress, you know. No, I love the ambition and excited to follow it. Come back on again soon. Yeah, thanks so much for stopping by. We'll talk to you soon. Bye. Let's talk about sales tax AGI. Numeral, put your sales tax on autopilot. Spend less than five minutes per month on sales tax compliance. Go to numeralhq.com to get started. If you're selling anything, software, hardware, e-commerce, whatever you're selling, get on Numeral.

2:18:47Makes it really easy. Monitors, where you're selling, how much you're selling, whether you're approaching that threshold. Start filing and it does it for you automatically. Cuts it down to five minutes per month. It's brilliant. Also, get on Public, public.com. Investing for those who take it seriously. Multi-asset investing, industry-leading yields. They're trusted by millions, folks. Double kill. Double kill. I love a double kill. Thank you to public. Anyway, our next guest is from Pallet. We have Shushant in the studio. Let's bring him in and hear the latest update. How are you doing? Doing well.

2:19:22How are both of you? Fantastic. Would you mind kicking us off with a little introduction on yourself and the company? How are you doing? Yeah, sure. So I'm Shushant. I'm the founder and CEO of Pallet. So what we do at Palette is we're an augmented AI platform built for the logistics industry. So we automate away all the back office tasks you do in running a logistics business, right? Like data entry, quoting, appointment scheduling, tracking where your ocean containers are. So there's about like$11 trillion spent on logistics. But what's surprising is about a trillion dollars of that just goes on these back office tasks that you have to do.

2:19:55So what we saw was here is like a real prime application of using AI agents to go and automate away these tasks. But we took an augmented AI approach in that we always have humans QA the work that agents are doing. That way, the completion of the task is near guaranteed to very high degrees of accuracy. So that's why a lot of the leading logistics providers really trust us to automate away what basically eats up about 10 % of their optics. What's the wedge product that most companies start with? Do you think of it that way, or are you trying to sell a whole suite of products on day one? So what's interesting is that the technology we built is pretty generalizable, and you can customize it per use case.

2:20:40So customers have started with a wide range of use cases. One of the largest cold chain storage companies started off doing data entry. another very well-known shipper has started off automating how they do how they basically bid for freight with different transportation providers other folks have looked at using us for automating spot market coding other folks have started off with appointment scheduling so we always try to start off with one use case but that use case could be quite distinct and the agent we deploy could be quite distinct client by client when you say data entry are you migrating people from other systems primarily or excel or are there still paper-based workflows out there like what are you actually seeing in the modern american economy because i feel like a lot of companies will give the pitch of like get off of paper but realistically most companies are on excel you guys have a humanoid robot that goes to a printer something out and then faxes machines still still in use of course humanoid robots will come in soon but uh the way it works actually is that like you go to one of these, like, even the largest logistics providers in the world, right?

2:21:44Like your DHL, your FedExes, they still get paper documents sent over email. Basically, the way it works is you have your email inbox, you have a bunch of emails coming in with documents attached to it. You look at it first, you look at this customer, someone I should be working with, is it someone I work with or not? Then I look at the document, does it have all the fields I want? Then I go type in literally all the fields off the document into my system. And I'm doing this like a hundred times every single day. Right. Like I think the whole reason I started this company was I went to one of our customers offices.

2:22:19It was like a Sunday afternoon. And he was like typing in data, like he was typing in a hundred orders into a system on a Sunday afternoon. And he's like, I can't watch the NFL playoffs. I have to type data manually in from my inbox. That is just un-American. transportation management system and this is how i spend my sunday evenings and i was like that's crazy like there has to be something better we could do here what is the what is the super bowl of the industries that you sell into like are there specific conferences where where everyone gets together in terms of like logistics that are super important for you to sponsor or like what does the top of funnel look like for you or or is it is it very much just like cold outreach sales wraps, dinner, steak dinners, that type of thing?

2:23:04Or what's the actual go to market motion? So the go to market motion, I think we rely on a wide range of channels. We rely on conferences, cold dialing, believe it or not, is still really popular. Like when I got our first customers, like I would actually do something even more extreme. Like I would drive around East Bay and Stockton, literally knock on warehouse doors, get kicked out until I could get our first customer in. So I went a bit extreme, but a lot of our SDRs still cold dial. And then there's quite a bit that comes in inbound. But I think the key, the key though, it's not just about the method of the outreach.

2:23:39Like all these companies are a bit skeptical of Silicon Valley technology companies naturally. Right? Yeah. So the thing I tell, like the thing that we've done is about 30 % of pallet has come from the industry. They've worked at places like Siva, they've worked places like Uber Freight, they've worked at places like Flexport, and they understand all the use cases in the company. So I always tell everyone on our sales team, you have to earn the right to talk about the technology. First, you need to show the customer on that cold call that you understand their business really well. That like, if I'm talking to a freight forwarder, I understand terminology, like how they deal with containers, what are incoterms and all this industry specific jargon.

2:24:15And if I can show that, the person on the other end is more receptive to hearing me out about how I can utilize technology. Because if you understand their business processes, you understand their workflows they're like this is not just another company that's in soma that just like looking at logistics terms on chat gpt these are people that actually just get me how much has ai been just an accelerant of the product that you're already selling versus something that you can upsell and bolt on as like a new product i think ai has been a huge game changer where it's like been like the primary focus of the company right like if these foundation models that didn't exist or value prop of our product wouldn't be as strong as it is today.

2:24:56So I'd say it's very critical. And I think, in fact, one thing that's made it even more critical is what's happening with the recent tariffs, right? Like the volatility, like what tariffs have really created is like this extreme sense of like volatility where like April container volumes in Long Beach are spiking. Now they're down and the courts reversed the ruling and it gets reversed back in the last hour. So people have no idea, like my volumes keep going up and down. And I just can't scale up the human capacity to support it. And now AI really just allows me to kind of come in like, if I'm getting more emails into my inbox, I can deploy these agents, and then they can go and handle this excess volume that I would otherwise have to hire contractors really quickly to go do.

2:25:37So it's a huge game changer, especially with what's happening right now. What was it like selling, maybe call it a month ago, when the trade war chaos was sort of peaking in some ways. And I imagine people at that point, you know, feel like they probably need better tools, but at the same time, they're like, hey, like, I can't really handle like onboarding. But I'm curious what what your experience was actually like. What I tell them is like, look, you're kind of like right now, the way we price this is on a per success outcome. Your cost of doing this, let's say is a dollar in house, it's going to be $1.30 on pallet.

2:26:17So if we do what we're going to say, you're going to have guaranteed ROI, and you're going to have immediate time value. The second thing is that we've built a very, the way our product works is we can adjust the customer's SOP and immediately spin up an agent that replicates their workflow. So we also really emphasize our time to value where we're like, look, you're not changing your tech stack. You're not changing anything about your operation. You're just sharing, you literally give us a video recording of what you do. We'll capture that. And we can then create the agent to go and automate it.

2:26:49And it requires minimal touch. You're not changing your system. I just have to call something out. So you were at Retool before Palette. You just had David on. And he said that he's just so funny. He was like charging on a per outcome basis is BS. That was his take. I don't fully necessarily agree with it. But it's amazing that you were just there. Yeah. I love David. He used to be my former boss. but I think that's one area we probably disagree on. I think outcome-based pricing is a game changer. Yeah. Yeah. Yeah. Talk about the round. It's only been seven months since you raised your $21 million Series A.

2:27:26Now there's a$27 million Series B. I'm sure there's a valuation markup. I don't know if you're sharing that, but it's not a huge step up in terms of total capital raise. So what does that tell us about the structure of the business and the burn rate and how you're thinking about capital consumption. You mentioned that AI has kind of infiltrated everything. Is that a cost center now? Are you actually spending significant amounts of capital on inferencing AI models? Or is this more just, we have something that's working, so we're going to hire more, write more code, more sales reps, more dinners, more steak dinners, more champagne, more hitting the size gong when you get a big client?

2:28:05So actually, most of the capital from the last round, we hadn't even spent. We barely tapped into that. So this round was entirely opportunistic. And part of the reason it came together was the folks at General Catalyst, I knew them really well. We got to know each other really well, and they helped build Samsara. And Samsara was probably the most iconic logistic software company. So a lot of it was Heymont, Mark, and the team there really knew how to build a big business. The second part was, I kind of felt like the industry was going for a big change, and I just really wanted accelerate our momentum on our product and our enterprise sales motion where what I saw was like, I was talking to the CIO and the CEOs of companies like DHL, Kuna Nagel, Estee Lauder, and all these corporations.

2:28:49And they were telling me like, we're trying to figure out how to use AI. We have no idea which model to pick for which use case. It's like these models keep evolving at a rapid pace. We need someone to be that translation layer to kind of make that happen. So that was very obvious from the conferences that I was going to. And the second thing was when these tariffs kicked in, I was like, there's so much volatility that the value prop of this stuff is like a no-brainer and we just really need to accelerate at this time. So that's why it was completely opportunistic where we had the right partner.

2:29:19The market felt like it was just getting ready for the technology and we were like, we know exactly what we want to do. So let's go and accelerate. Yeah, you mentioned some of those really big companies and kind of them not being able to decide between different models. have you bumped into any of the large consulting firms, the McKinsey's of the world? Because we heard this narrative early in the AI boom that they're making more money than the startups and the tech companies because they're selling pitch decks at extremely high margin, talking about AI transformation, and then just saying, hey, if you're a big company, you can go build it.

2:29:55Here's here's our take on AI? Or are they more of a partner that could recommend you? Do you have any insight into how the big management consulting firms are confronting the AI issue and plugging into like the Fortune 500? So what we've generally seen is that the consulting firms are like very much open to partnering with companies. I think the way that you think about it is like the executives at these businesses, they're like, there's a lot that's happening. There's a lot of noise, I'm not even sure who the right partner is. Do I go with a horizontal solution like Microsoft Copilot? Do I go with one of these newer players?

2:30:31Where do I allocate my capital and who do I pick as my partner? And think of the consulting firms as a way to help these companies filter out all that noise on who is the right partner for me to work with. So we very much think of them as critical partners that help us earn the trust of these Fortune 500 corporations. And that's how we view it. Actually, the CEO of one of those firms is actually someone who I've gotten to know well. And he actually has helped us actually open a couple of doors. Very cool. Just got to say, you're exceptional at sales. I'm ready to sign up for pallet right now. And I have nothing to do with your industry.

2:31:08Hey, we're going to be shipping a lot of hats, a lot of jackets. He's like, well, a lot of people think that our solution might not be right for them. Yeah, I'm like, I want it. actually try yeah yeah yeah yeah the the anti-sell sells better than anything else for sure yeah um anything else jordy no congratulations on the new round yeah this is fantastic progress yeah i mean grateful to be here um one last thing by the way is that one thing we've uh we've really kind of pushed for a palette is this concept called hybrid work sure where what we believe in is actually that like there's only two places you could work at palette either in the office or at the customer site.

2:31:45So every week, every month, we make our team spend a week at the customer site. So last month, our entire team was out in Medellin at a customer's BPO, learning their operations and learning how they can like, like every single aspect of their processes. And I think this is really important because in a world where software costs are treading to zero, like if you don't understand the customer's workflows, like in excruciating detail you might not be like building the right thing and they don't feel good documentation you can't just go chat gpt the right answer on how does container track and trace work you actually have to sit there observe learn and then uh yeah i love that the hiring the hiring process candidates like so do you guys support a hybrid work and you're like of course we do you know about a week out of the month you'll be you know in south america you know um and you and you built out that function at retool right this sort of concept of this deployed engineer is that correct correct yeah yeah i i was probably the first forward deployed engineer there very cool it's cool i love i love that so you're pretty bullish on the forward deployed engineering meme kind of just growing and growing in the future broadly for like every company or is it specific to b2b software or something about ai automation and uh something that needs to be customized i imagine that there have to be some organizations that they don't need it because the product is so standardized, right?

2:33:13Yeah, I think it's not necessary for every type of business. I think in our case, we're dealing with a company that has a lot of internal systems and processes that you have to hook onto. And we're trying to make change management, to your point, what you said earlier was like, hey, I don't want to change my operations too much. I want to stick with what I do so if we want to make change management simple we need to hook them with your existing infrastructure so at least for pallet we think the forward deployed engineering function is critical to make change management easier but there's obviously tons of companies that could succeed without that function yeah yeah that makes sense uh well thank you so much for stopping by yeah congrats on all the progress i'm super bullish yeah we'll talk to you soon cheers bye how'd you sleep last night jordy i have historic numbers actually had kind of a brutal watch it We were on so much caffeine yesterday, John.

2:34:04Did you track how much caffeine you were on? I was on a lot of... A fair amount of caffeine. A couple of year remontes. I only got six and a half hours of sleep, which is abysmal for me. Almost seven. I got an 82. What did you put up? What was your final number? Did I beat you? No, you didn't. Even on my worst night, John. My worst night this month. I still beat you. I got an 84. Anyway, go get yourself an eighth sleep. Five-year warranty. 30-night risk-free trial. Free returns and free shipping. It's what? Charles Leclerc. Sleeps on, that is crazy. When he's crying himself to sleep. Yeah. There's a timeline post I wanted to highlight here from Andy Nexuist because it ties into our next advertisement.

2:34:42Stripe put up a digital billboard in Grand Central and it just says, Stripe Billing, usage-based billing to scale your business faster. And the meme is, is everyone there having the same thought bubble? What a silly niche ad. Nobody at Grand Central has even heard of Stripe, let alone implemented it in prod like I have. And it's so funny, but it's so true. Like I have actually implemented Stripe and I'm sure so many people that go through Grand Central have. And I think it's my interesting takeaway from this is that there is a desire for, like we talk about out of home, obviously. And we're going to read you an ad quick ad.

2:35:28Don't worry. It's coming. But I think that there are companies that feel like they need to abstract their message into even broader terms when, in fact, I think just being clear about what the product is and just knowing that, yes, implementing Stripe is a common thing at this point. And those people walk through Grand Central. Yeah. A lot of people do start yelling about their company without being clear about the product marketing effectively. And that can just make your life so hard because people are like aware of your brand or your company. But when the moment comes that they would convert, they don't actually think of you in that context.

2:36:18Yeah. Like there is a different version of this Stripe ad that is like a photo of Patrick Collison talking about increasing the GDP of the internet and stuff. And it's very like, like high fidelity and stuff. I would like to see him going like one of the classic Arnold houses. Hitting the ziz. The ZYZZ guy. Yes, absolutely. But, but, but aside from that, it's like, it's like, it's okay to just send the message up front. Just be straight up with it. Like AdQuick, out-of-home advertising made easy and measurable. Say goodbye to the headaches of out-of-home advertising. Only AdQuick combines technology, out-of-home expertise, and data to enable efficient, seamless ad buying across the globe.

2:37:02And, yeah, I'm still super long on out-of-home advertising. You can see Stripe is doing it in Grand Central. And it's the most fun. It's the most real. Yep. seeing your ad on on meta x etc is cool yeah seeing it in the real world nothing like it nothing like it it's different uh did you see nathan fielder piloted a full boeing 737 plane yeah so i didn't finish the show yet but i finished it it was kind of a challenge because i kept falling asleep yep like two-thirds of the way through every episode so i feel i've seen the entire season but like i actually should probably watch it back again awkward but um but yeah it was it was absolutely wild and especially the context like the timing of when of when the show actually premiered and the episode started rolling out yeah it really is insane but he's one of the best ever do it and he's he's a he's a real icon yeah and uh we have to i would like to be in the next his next uh season well speaking of icons there's nothing more iconic than a Rolex, and you can get one on Bezel.

2:38:13Go to getbezel.com. Your Bezel concierge is available to source you any watch on the planet, seriously, any watch. They have 26 ,000 luxury watches available, and so go check it out at getbezel.com. And next up, we have Rob Taves, a good friend of mine. I grew up with him. I went to middle school and high school with him. Now he's a venture capitalist at Radical Ventures, and he's a great columnist and writes about artificial intelligence and has been writing about artificial intelligence. since before it was cool, which is fantastic. So we'll bring in Rob and ask him a bunch of questions about artificial intelligence.

2:38:46How are you doing? What's going on? I'm doing great. Thanks for having me, guys. Welcome to the show. What is top of mind for you right now in AI? What's the biggest thing that you've gotten right over the last few years? I know you have that scorecard, 10 predictions, and then you rank yourself. As you look back on all of your predictions, what's the big one you got right? What's the big one maybe you missed on? It's a great question, yeah. Every year I write a column of 10 predictions for the year to come in the world of AI. And to keep myself intellectually honest, I go back at the end of the year and grade which ones were right, which ones were wrong.

2:39:20Let's see. So far this year, one of the predictions I made at the end of last year was that President Trump and Elon would have a messy falling out, which would have various implications for the world of AI, which looks like it may be playing out. We'll see how that all comes together. Yeah, we were reading that today because Elon is no longer a special government employee. But at the same time, messy seems like the key word. There might be a split, but if it's clean, you're getting a yellow light. It's like a clean split. We need satellite imagery of whether the red Tesla is parked at the White House.

2:39:55If there's a Tesla still at the White House by the end of the year, I think you got to fail that one. Yeah, not messy yet, for sure. Yeah, it's interesting because it is kind of a narrative violation because half of the world was like, they're going to be together forever. They're in love. They're going to be partnered for four years, maybe for 10 years. And then the other half was like, this cannot last. It's going to blow up and it's going to be super messy. And it feels like right now it might just be, hey, back to business, sleeping in the factory. Who knows? Yeah. Well, the book is not yet over.

2:40:25Yeah, yeah, yeah. Anything could happen. Anything could happen. um what what about um things that you feel like have played out as expected have you been uh have you been uh kind of reassured or shocked by uh the data wall the pre-training wall the shift of the focus to reinforcement learning uh the importance of tool use going forward the importance of productization uh versus this like this narrative around oh yeah just scale up the big transformer model gpt7 will be asi done yeah yeah no i think i think those trends are all very much playing out and it's been interesting to watch i think it was it became increasingly clear over the course of last year that the pre-training scaling laws really were plateauing and um this narrative emerged around towards the end of last year as opening i was increasingly teasing its reasoning models and it's O series of models that there is this kind of next frontier, next vista for scaling, which was inference time compute and scaling reasoning models and so forth.

2:41:31And the reasoning models, O1, O3, O4, DeepSeqs, R1, et cetera, they have been very powerful and have unlocked a lot of new capabilities and use cases and so forth. I think the jury is still out as to whether they really represent like this next massive Runway for scaling that will take us many years into the future I think it's still maybe the case that the fundamental underlying capabilities of the models are Not growing as quickly as they did in the like GPT 2 to GPT 3 to GPT 4 era But importantly that may not matter that much to your point John because I think The value and the activity is increasingly moving up the stack and even if all model capabilities is basically we're frozen in time today.

2:42:14There are no further advances. There's literally trillions of dollars of economic value to be created by just figuring out how to productize these models, build particular solutions for particular end markets. And even the big frontier labs, OpenAI, Anthropic, et cetera, are increasingly focused further up the stack. So I think that's where more and more the action will be. Yeah, what is your take on how the shift from pre-training to test time compute effects, like data center build out, the need for these like hyperscale, Stargate-like massive data centers. That stuff made a ton of sense in the Leopold, Aschenbrenner, we're just going to scale it up and get this massive data center going for like the biggest pre-training run ever.

2:42:57In a test time inference regime, it feels like maybe it's more on demand. Maybe you don't need as many big super clusters. But am I just thinking about that incorrectly because maybe we still need massive clusters because the rate of token production is just going uh going completely parabolic no i think that i think that narrative conceptually holds true and and i think a key insight as you touched on is inference can be done in a much more decentralized way you don't need to have like hundreds of thousands of gpus that are all really tightly interconnected like as close together physically as possible and so in a world where more and more of the compute is inference either like productionization of models or even inference time compute that can be done on gpus that are more spread out and you don't need to have these like matt like five gigawatt clusters the counter narrative is the like and i'm sure you guys got sick of like hearing jevin's paradox get referenced like a million times over the past few months we love jevin's paradox we think it should be taught in middle schools we almost got it tattooed right here never forget never forget Yeah, that guy has really been immortalized.

2:44:04Yeah. But I think there is a narrative that even as inference becomes more important, as compute gets cheaper and cheaper, there will still be bigger and bigger and bigger pre-training runs. And so that will justify the massive CapEx build-out that we're seeing. Yeah, I've always been interested to see if there's going to be like the big transformer training runs, like the GPT-4-5 type training run in image diffusion or in robotics or even in, I was thinking about like, you know, we've kind of solved chess engines, but like what happens if you scale that up seven more orders of magnitude? Like just do you get some weird outlier scenario?

2:44:43The question I always have is like, when can I see it on semi-analysis? When can I see it from a satellite image? Then I know that the big training round's happening. I'm wondering, are there any other areas or regimes of training that aren't just predict next word that we could see a huge data center run happen? Or is it really just text is the only option for that scale? Or will we see like a VO4 training run and we'll be hearing about like, oh, wow, it's at five gigawatt scale because VO4 just needs that much pre-training. Yeah, yeah. So this was one of my predictions for 2025 is that we would start to see more and more scaling laws in data modalities other than text, other than language.

2:45:27And I do think that we're starting to see that play out in robotics, for instance, in biology. I mean, there's like the whole premise of the whole justification for building such massive compute clusters is this notion of scaling. And as you increase compute and increase data, the model gets reliably better. And we are starting to see like nascent signs of that in some of these other data modalities. I think one important challenge or constraint though is there are very few other modalities that where there is as much raw data available as there is for text. Like there is not an internet for robotics training data or an internet for biology even.

2:46:05And so the training data sizes can limit it. Like it doesn't make sense to have a massively overparamitized model if there's just not enough training data. And so for the foreseeable future, at least, like I think the amount of training data available will cap like how big the biggest robotic foundation model can be for instance relative to the biggest general purpose language model do you buy the whole like uh robotics thesis of like but but transfer learning is really effective and like we can totally simulate this because we have unreal engine it feels like a little bit of potential cope being like well we know that you don't have the trove of the robotics data And so you're coming up with something that hasn't worked in other modalities, but at the same time Unreal Engine is pretty real and you can imagine walking a robot around and learning a bunch of stuff if you train So what is your take on on robotics data and transfer learning simulated learning that type of stuff?

2:46:59Yeah, I mean I think there's no question that robotics foundation models are getting increasingly generalized and like there's more and more signal that You can in fact build a general purpose robotics foundation model and you can expose it to some previously unseen scenario and it knows how to navigate the real world in a way that generalizes the same way that language models started doing a few years ago. I think this question around training data is a really, really important one and a really key one. And there are very strong opinions on both sides. There are robotics experts who swear you can use simulation and synthetic data to massively scale up the training data set and use that to kind of erupt these scaling laws.

2:47:39And there are other folks who believe that you know there's really no substitute for real-world data and you can you can supplement it a little bit with simulation but like the sim to real gap is still a very real like there's a meaningful gap there and so you can't get that much juice out of just simulation I think today I fall more on the side of the importance of real-world data like I think there's just so much nuance about the real world that that isn't adequately fully captured in Unreal Engine or whatever like sophisticated simulation engine you can build. But I do think as time goes on, that sim-derial gap will probably close.

2:48:16I would imagine it ends up looking not dissimilar from the way the autonomous vehicle industry involved. I was about to ask. Where real-world data is essential, right? And everyone remembers seeing Google's cars driving around for over a decade collecting data. but no major autonomous vehicle program today is not deeply based on simulation as well and so i think some mix ends up being necessary i think in robotics today it's still you still have to be heavily weighted toward real world data but hopefully just for the sake of leverage and technology advancement i think simulation will end up getting better and better is there an analogy between the pre-training to test time inference regimes in LLM training and chat models to what's happening in autonomous vehicles and the difference between pre-training on all the data and on policy training that you can do and then also add simulation on top of that?

2:49:14Is there an evolution of the mix of training paradigms that you would use in AV. And I guess like the bigger question is just like Waymo versus Tesla. Is there a meaningful data gap between those two companies? Because Tesla's obviously been tracking tons and tons of data, but Waymo seems to be working pretty well when you get in the back of one. And so it seems like both companies might just have enough data and we might wind up in a situation where, you know, like no one's talking about a data gap between Gemini, ChatGPT, Anthropic, Llama. They all just have all the data. Yeah, yeah, totally.

2:49:56Yeah, it's interesting. The way that autonomous vehicle AI stacks work today is like, because they were kind of crafted and developed in the pre-generative AI era, they're not like these massive pre-trained foundation models where they just fed all of the data in the world. Like, they are much more handcrafted. But it is an interesting thought experiment. And there are like more younger, next generation autonomous vehicle companies that are taking this approach. like let's build a foundation model for driving um and you know and can we do something you know similar to what companies like physical intelligence are trying to do for general purpose robotics can we just apply a model like that to autonomous vehicles how would a startup like that get data that that feels like the hardest thing uh because you can't just crawl the web right exactly yeah yeah it's yeah it's much harder to get training data for it to your question on google versus tesla waymo versus tesla i think this is like a fascinating age-old question And you're right that Tesla has way more training data in the sense that it has this fleet of personal vehicles driving around.

2:50:56The quality of the data is lower than Waymo's, though, for the specific concrete reason that they don't have LiDAR as a sensor modality. And so this is like the big debate. Before I got into BC, I worked in autonomous vehicles. And like even back then, this was a debate like, can you get to level four autonomy without LiDAR? and Tesla needs to believe that the answer is yes because their business model is selling cars to consumers and LiDAR is so expensive that it would completely wreck the economics of selling a car to an individual. Waymo can stomach the cost of a LiDAR because their model is a robo-taxi model.

2:51:33Honestly, I think the jury is still out. Tesla folks will tell you that we're very close to that and LiDAR was never necessary, but it's, I think it's not, it's not completely clear yet. And so in that sense, like the data set that Waymo has is more robust just because it's much more multimodal. Do you have an, do you have a sense for why LIDAR is so expensive? Like, I mean, Elon Musk was able to make rockets cheap. Like the iPhone is cheap. Like how can we not get LIDAR down? Like even just to like a couple thousand dollars, like that wouldn't break the Tesla paradigm. Right. But I assume that when we're talking about a lidar package on a waymo we're talking like five or six figures and that breaks the model but we historically humans have been really good at like mass manufacturing expensive stuff and make it cheap so uh i've always wondered about you know elon for a decade has been lidar is doomed we don't need it we don't need it but he also changes his mind and so i wouldn't be surprised if one day he's just like yeah we figured out how to do it we did we have lidar in the cars and like you know yeah too bad yeah yeah yeah and there have been rumors of Tesla like experimenting with LiDAR here and obviously under wraps.

2:52:42So I wouldn't, it wouldn't shock me. I think it's a great question. I think the short answer is just like, it's a complicated sensor. There's a bunch of lasers, they're moving around and so forth. But, but to your point, uh, the cost curve has been coming down on LiDAR and we'll continue to come. So like the original, like kind of like bucket shaped LiDAR that Valadine made were like $64 ,000 a pop on like the really old school Google self-driving vehicles. Now they're probably like a few thousand dollars each. You need several of them on a car so it does add up. But I think you're totally right.

2:53:13And like there are like research prototypes of LiDAR that are solid state, meaning they don't have pieces that move and that makes them a lot cheaper. And people talk about like$250 per unit LiDARs, which again are not like production ready yet. But I think you're certainly right that like the way this debate could be resolved is like it may just all converge because in a few years LiDAR gets cheap enough that like Tesla can use it and everyone can use it. I had a buddy, a friend of the show, share recently. I'll give you some context. He said Waymo will change the world. Walk down the street,$1 million in metal, sitting unused 20 hours a day on every street in America.

2:53:49Garages will be useless, turning into storage or ADUs. Parking lots, the same thing. Street parking will be more lanes. Driving offense revenue, DMV revenue, parking tickets all go to zero. Gas stations will be worthless. sounds insane right but uber went from nothing to everywhere in a decade waymo might take 15 years but the disruption will be nuts um how do you think about the downstream you know assuming that you believe some of that is is real and and i think sean makes some great points um how do you think about the investment opportunities that are downstream from ubiquitous autonomous driving and like is it even are there going to be vent as many venture opportunities or ai car washing startup pull the autonomous vehicle in it washes itself it's great that would be different than a regular no no roll up we're doing a roll up for humanoid humanoid yeah um yeah so i just to me this seems like a lot of opportunities on the real estate side is like hey can we turn this into more housing or you know whatever but parking lots aren't cheap though but yeah this is this is the context here is that Waymo monthly rides were sort of ticking up gradually and then just shot up to 708 and I guess they're now doing more rides than lift in the Bay Area as well it's crazy yeah it's been amazing to see how quickly Waymo has gone from like a novelty to just a piece of the fabric of life for people in San Francisco.

2:55:21And like most people that I know in the Bay Area, in San Francisco, use Waymo more often than they use Uber and Lyft. And it's like quickly spreading to beyond just being a Bay Area phenomenon. Like they're now live in LA, as you guys know, and in Phoenix and in Austin and so forth. And yeah, so I think it is, it's remarkable to see it after all these years finally become like a real business that's scaling quickly. And I really like the excerpt from your friend. I mean, I think I very much agree with that line of thinking. And one of the things that initially attracted me to the world of autonomous vehicles back like a decade ago, what was this fact that it's really fascinating technology?

2:56:00You know, it's difficult technology being able to get a car to drive itself. But it also, once you solve it and you start scaling it, it has so many broader implications, second and third order impacts on so much of society and the economy. Like so much of modern life is built around roads and cars and the way cities are designed and so forth. So I do think it will have dramatic implications on civilization, bigger picture. It will play out over a longer period of time because we're talking about the built world and it takes time to adjust. But yeah, I think especially in cities, at least to start, especially in urban areas, It's some crazy stat like a third of real estate in the average American city is devoted to parking.

2:56:43And so much of that can go away. And it's like such an inefficient use of space. So you can imagine cities being totally redesigned around humans rather than around cars, more pedestrian areas and so forth. You can also imagine there's a lot of people talking about the rise of exurbs, like being able to live further and further outside of an urban setting. because when you're commuting, you don't have to be driving. Like you can imagine the entire form factor of a car changes and maybe you have a desk. Yeah, I want it to be a desk and a couch. And I just want to, yeah. I mean, I've spent a bunch of time thinking about this because I kind of have a gnarly commute right now.

2:57:19And it will just become so much better within when I can get it. Jordy's in Malibu, I'm in Pasadena. You've spent time in both. We'll get you back here eventually. Yeah, yeah, yeah. Yeah, I'm excited for Waymo to expand the geofence in LI more and more. Yeah. Well, this is fantastic. We'd love to have you back. This is a lot of fun. We could talk about 25 other topics in AI, since this is the most fascinating industry right now. We didn't even get a chance to talk about deals and stuff, but I'm sure we'll go into all that the next time you're on. Yeah, that sounds great. Thanks for having me, guys.

2:57:52Thanks so much. We'll talk soon. Bye. Next up, we got to sing. We got to hit gongs. We got lots to do. We're going to say. Are we going to sing with us? It's always hard to sing with a remote guest, but maybe we should sing beforehand. Find your happy place. Find your happy place. Book a wander with inspiring views, hotel-grade amenities, dreamy beds, top-tier cleaning, 24-7 concierge service. It's a vacation home, but better, folks. And we have the founder of Wander in the studio coming in to announce a massive round of funding. Welcome to the show. How are you doing? hit on that wide hit that gong hit that gong hit that gong welcome to the studio john it's great to have you here it's it's great to be here i was hoping that i'd be able to sing with you guys so i'm a little i'm a little disappointed that it's really hard with the delay we got to figure this out somehow some latency mitigation or something we also need to make it a full song yeah because oftentimes I get to place.

2:58:55There needs to be choruses and verses. Yeah. We need to integrate the whole pitch into one big song, but congratulations, break it down for us. Ideally, when somebody walks into a wander for the first time, they're sort of serenaded by us. Yes. We can sort of, you know, two, three minutes on the whole home speaker system. So we'll work on that. We can make that happen for sure. That'd be great. But yeah, give us the business update, break down the fundraising round. What are you actually spending it on? Because I know it's an asset light model, but explain how the round came together, the progress of the business.

2:59:30Totally. So it's a$50 million Series B led by QED, FITWALL, alongside Red Point. Logan Barlinson. Okay. Red Point, Cookin. Let's go. Red Point, Starwood, bunch of really incredible folks. Fantastic. And yeah, in terms of capital, it's really, I mean, I'd love to get on here and act like we're going to do some like crazy stuff, but it's really scale. Like we have three core priorities, which is quality stays, quality customer support, and then quality homes. And that's really where we're, where we're laser focused. Very cool. What's the key to onboarding more, uh, wanders? I I've seen the numbers ticking up every week as we cover you guys.

3:00:10Um, what is the, what's the funnel look like? Yeah. It feels like you guys have a relentless pace and like a culture that really celebrates that and yeah break it down yeah wander wander definitely has a very high output culture as a as a startup i tell the the team that like culture is not um you know like uh happy hours and you know like that kind of stuff it's it's about winning and so you know as a company we we obviously very much focus on on that idea from from a growth perspective there's about 300 ,000 Wander worthy locations across North America and Europe. And so that's really our target.

3:00:50You know, from a systems perspective, we actually built out a fleet of agents that went and found each one of these homes and then enriched it with owner contact information. So we know exactly who we're going after. And so that's sort of the focus right now is really a sales driven model onboarding these homes onto the platform and then automating their operations with Wander OS and delivering that great experience to customers. Talk about kind of looking back a little bit. I mostly want to spend time looking forward, but like navigating through the Zerp era, lessons learned, like that era allowed for a very different type of business model that you guys have evolved.

3:01:31But I would love to hear kind of the backstory. Yeah. I mean, when Wander started, interest rates were pretty much zero. And so that allowed for us to have a very asset heavy model where we actually went out and bought those first few locations on balance sheet really with the idea of solving the cold star problem of the marketplace do things that don't scale and so as as we grew obviously that had to transition you had two crazy events happen at once you obviously had um sort of the massive you know rise of interest rates but wander actually at the time had a hundred million dollar credit facility with Credit Suisse as a six-month-old startup, which was incredibly hard to put together.

3:02:09And then to have this systemically important bank explode as a CEO trying to scale the company was pretty wild. Traumatic. That was a pretty quick transition. talk about uh i i feel like one of the one of the craziest things you can potentially do as an e o a category where there is a there is a there's an active startup even if you're scaled maintaining differentiation and really competing as Chesky goes on a run, building out different products and different strategies. It seems like there's more opportunity than ever to differentiate, but how do you think about it? Yeah. I mean, first of all, I think Airbnb is a great company and Brian's an incredible founder.

3:03:12So I have nothing negative to say there. I think for Wander, you know we we do deliver a little bit of a different experience so our net promoter score for q1 is you know 85 uh which is you know phenomenal thank you uh yeah like true true customer love yeah and that doesn't happen by accident i mean literally if anyone has a negative sentiment in the concierge chat when they're like talking with our support that literally gets flagged across the entire company if there's a stay that's below an eight out of ten then they're going to get a call from our COO. And like, I see that feedback and we're going to be hyper aggressive on fixing it.

3:03:53And that's across, you know, every single stay, every single home, every single customer. And so I think that that like relentless customer focus is just a very different model than, you know, Airbnb. Airbnb is sort of that unmanaged marketplace versus, you know, Wander, like we truly do care about the quality of our inventory, the quality of your customer experience, you know, to the point where literally like I will hop on the phone and like deal with whatever the issue is to ensure that it gets there. And I know that sounds obviously like very unscalable, but I think that just purely from a cultural perspective, it forces, you know, systems and automations to be built.

3:04:33And in a time where I think that, you know, given everything that's happened in AI, you actually do have this opportunity to deliver hospitality and like perfection at scale. And so that's, that's really where our focus is. Can you talk about advertising in this category? Um, I was, I, I was, we were talking to Keith Roy about whether or not Airbnb should have an advertising product because we've seen with Uber and, uh, and is it Instacart where Fiji Simo was, where she spun up advertising. There's, it's, it's not quite a marketplace business, but there's an element of that, But the take rate was low.

3:05:10And so the advertising product was very meaningful at those companies. Keith Rauwoy's take was that in the housing vacation stays market, it's less relevant because the take rate's a little higher. But do you agree with that? Or do you think that there is a future where just broadly the category is driven by advertising dollars in a meaningful way? Yeah. I mean, when you go on to like booking.com as an example, or even the RBO, like you will see ads. Sure. Yeah. Promoted to get to the, because if I have a house and I want it rented, I'm willing to sacrifice a little bit of my margin to try and rise to the top of the rankings.

3:05:46Right. You'll actually even see off platform ads. So like very typical, like, you know, go buy this product where they're actually taking people off platform, which is pretty interesting. Like I'm sure that it's a revenue source. I mean, these platforms spend a ton of money on getting traffic from a performance marketing perspective, from internal marketing, SEO, et cetera. And so it certainly makes sense to try and monetize a percentage of that traffic that's never going to end up converting to actually booking a home. But that being said, Keith is obviously correct and very smart that the take rate on vacation rentals is really high.

3:06:26Sure. For Wander, our average order value is about$5 ,400. And you're looking at like an average take of about 30%. And so like for us, that would have to be a lot of like random ads, you know, like make that up. And of course, the customer experience is not not great. We're really trying to market users on that, that, that booking. But yeah, I mean, I think for a company like Airbnb, it would totally make sense, especially to the, you know, on the avenue, you mentioned where hosts are just paying a little bit more to promote their house to the top of the feed. You know, they have a lot of inventory and, you know, sort of pulling, pulling yourself out of that is, you know, difficult for your mom and pop Airbnb host.

3:07:06Yeah. Talk about disintermediation in the context of marketplaces and platforms. We saw the canonical example of the dog walker, which is basically just lead gen company, because as soon as you find a good dog walker, you immediately disintermediate. And there's some things that those companies can do to prevent that um with some rentals it's very much like i'm going to be in this town for just this week i need this place i'm not going to bother disintermediating but if you fall in love with a place and you're going there every year we've heard about people kind of trying to google the place and find a different way to get around airbnb has an issue where property management companies will list a home and then the person that's booking the home just google stay there and then just reach out to the property manager and do a deal off platform whereas wander you guys are just full stack right so it's like you could even get you could look up the title or whatever and get to the owner and they'd be like okay if you want to book the house like it's still going through wander but yeah yeah how do you think about that what what what does that tech stack look like to prevent that or help that yeah i mean so i i do think that it is like the core, the core risk.

3:08:16And I think that's even something that like has been talked about on their earnings calls is like direct booking websites and there's ways for them to mitigate it. And you also have a ton of supply that isn't professionally managed where the operator isn't going to have their own direct booking website. They're just going to list on Airbnb. And so I think they're actually like positioned relatively fine. You know, I think it's like a, a travel hack that not many people use, you know, for, for wander like i am a huge fan of like having complete and total control over the business and the platform that i'm building like when i was a kid um my first little company i was like 13 14 um we were like hosting minecraft servers and whatever else and when minecraft got purchased by by microsoft they rolled out a eula totally killed my little business had to fire like four or five people and so i learned about like platform risk at like pretty you know pretty young age.

3:09:06Never again. Wow. And so, you know, for Wander, I knew I wanted it to be verticalized. I knew I wanted to have like my own booking engine. I knew I wanted to have my own property management software. I felt like that was like the most durable piece. And then you also have to ask yourself, like, you know, as a space and a brand matures, like that brand, that brand promise, you know, actually matters. And so if people look at Wander and associate it as a brand that says, hey this is a quality stay you know you don't get uh replaced quote unquote um and so i think that's also like a really important piece is the underlying brand and sort of that guarantee to the customer they're going to have a good trip yeah uh how do you think about taste in the context of of what you're doing wander's always felt like a just a very it's felt like a hospitality brand as much as it's felt like a technology company uh where did that come from from anywhere i mean you were you were building uh coding coding tools historically which which you know i guess it's you know important to have good design in that category but maybe when you started coder it wasn't even the case so i'm curious where it came from uh besides you know people like kyle crushing it yeah um you know my my journey as a founder has been like pretty you know pretty fascinating obviously like my first venture-backed company i started when i was 17 18 you know coder enterprise developer tools so radically different space than you know travel um you know however like i've always had a deep passion for you know design and as a kid i you know did high school online traveled you know 200 plus days a year all over the world um and you were you were race car you're driving yeah i don't really i don't really talk about it but yeah i used to i used to race formula four No way.

3:10:55And then, yeah, Formula Mazda. It was funny. John and I got lunch in Malibu a couple of years ago at this point. And I was like, oh, do you want to like, I knew he drove cars at a high level. I was like, do you want to go drive my Ferrari or whatever? And John gets behind it. And he like, you know, normally when people are like trying out a car, you know, like everybody's car, they're like taking it to him. And he's like, you know, really experiencing the naturally aspirated v12 that's amazing but i was confident i was i was okay i was confident that he was gonna take care of it so you have the experience for it that's awesome yeah it was it was a great it was a great time great meal um and so so yeah i think with wander i'm candidly speaking like wander is like my soul but like as a company like everything needs to be high quality the software needs to be on point i also don't think people realize like how much is truly automated with Wander.

3:11:50Like at the top, you have the booking platform, but underneath is literally this property management software that's running all of the vendor communication, coordination, payouts, preventative maintenance, task tracking. Like Wander doesn't actually employ any local property managers. That's all just through software, the underlying coordination of vendors, which I don't think that like anyone fully realizes because of course we don't market it that way. Like, you know, you want it to feel like a, like a magic trick. Um, and so I think like, I think that's probably where you're seeing the design come from is that like, if there's anything on the site that, that, that bothers us, like the entire team is just obsessed over this, this principle of quality.

3:12:31Last question for me, and we'll let you go. Um, Um, are experiences or local other services, uh, a true compliment to vacation bookings and housing, or, or is that kind of a round peg in a square hole? Yeah, the, the way that I look at it is that I think it's a feature of a platform. I don't think it's a platform in and of itself. And I think that you have this rare moment where you can effectively abstract the way that you connect to these service providers. So if you were to go back in time, let's look at OpenTable as an example. They had to build out integrations with the restaurants. The restaurants had to use OpenTable to manage their reservations or whatever else so that you could provide this online platform.

3:13:19Now with what exists from a technology perspective, you can have an AI agent call a restaurant and make the reservation for you. And so you no longer need to force these types of integrations. And so the way that I look at it from a services perspective is you just end up with this like a genic concierge that exists inside of your travel app that goes ahead and calls the you know local chef or the restaurant or whatever else and so you don't really end up building necessarily direct integrations you more build like an abstraction layer and a curation layer on the experiences it makes a lot of sense i mean i have one more question about uh seo is really big in the early airbnb story are you looking at any of these ai seo tools do you think that it's important to show up in chat GPT, for example?

3:14:02And are there any services or kind of best practices that you think relate to AI SEO? Or isn't it GEO, according to Andreessen Horowitz? Yeah, I don't know what the acronym is, but that it sounds like a good acronym. It's actually something that we have started focusing on pretty intensely. So the real key is sort of where these agents are referencing you know the materials and the facts that they're that they're getting and so what you end up with is like you basically need a source of truth strategy um which is is a really interesting phenomenon basically it's like how do you how do you sort of like uh become the system of record from a fact perspective or get the data onto you know a platform that that is is viewed as having the facts like for example chat gpt references Wikipedia a lot.

3:14:51Um, and so like that, candidly speaking, it's a little bit harder to like quote unquote hack versus like traditional SEO. Um, and there's definitely going to be like an entire push. And I actually think you're going to see a lot of value creation from platforms like Wikipedia, uh, come from the fact that they are viewed as like a source of truth. Very cool. anything else Jordy uh you said there's 300 ,000 homes that you guys have sort of soft circled as targets I'm assuming you're going to get all of them in the fullness of time what do you what do you want to do in the next uh what do you want to do by 2030 do you have a do you have a target yeah I mean hopefully by 2030 we've accomplished that for sure um that's the that's the pointing to the yeah pointing to the I like I I can't wait for there to be like five homes left and you're just following you're following around the owners like in a helicopter being like i got you i got you give us the keys we'll get it done and before i jump one thing i want to note is as a sponsor of of this show how incredible you guys are and how happy we are and so to anyone who's watching this thinking about where to spend their ad dollars or what podcast to sponsor i cannot encourage you more enough to work with these boys they are incredible and the roi is is through the charts thank you man that really means a lot you guys you guys bet uh you guys bet early you're one of our very first ones and uh we will never forget it yeah i think my favorite thing is that uh people might think that we ran the song by you before doing it live we did not live we didn't get any pushback and i think it's all for the better if there's less oversight so we've been having fun it's funny did we create did did we create we created singing we created jingles we created the jing we created the jingle from first principles we created the jingle no literally they didn't send us any copy we didn't ask for any copy we didn't ask like how do you do an ad read for your company we just went to your website and just sang the first tagline on there if it had said anything else it's so funny it's so drilled into my brain now that i just assume that you guys were we just we just went to wander.com and started singing find your happy place well we should take you out i want to make i want to make like a rate we should we should make like a radio oh absolutely yeah hey this is john and jordy anyway thank you so much for coming on the show this is fantastic and congratulations i'm super excited for you and the team and uh john and i are like basing our summer plans off of wander vacation uh wander location so for sure uh congratulations we'll talk to you soon bye cheers thank you guys so much later john uh one last news item I want to hit before we get out of here.

3:17:34We were talking about the poly market on Elon Musk out as Tesla CEO in 2025. It is lower than ever. It was around 20 % in March, crashed down to 18, 15%, was recently sitting around 13 % and now is down at 9%. And so So the news about Elon leaving the U.S. government has been a bull case for him staying as Tesla CEO, which makes sense because he has a lot of work to do, a lot of opportunity between self-driving and humanoid robots. Who better to run that company than Elon? And another one I've been tracking, Circle IPO in 2025 is up to 94 % chance. It drops to 47 % chance on May 21st, so about a week ago.

3:18:23and it has just rocketed up to 95. And I expect we'll see some news on that front. Yeah. I mean, the other markets, since we're in polymarket mode, these are fun. I love checking these. So there's, of course, our market on how much the iPhone 17 will cost. Over$1 ,000 is a 13 % chance. Over$1 ,500, 2 % chance. Over$2 ,000 is only a 2 % chance. I think Tim Cook's going to get it done. Keep iPhones cheap. The other interesting thing is which company has the best AI model at the end of May? Google's running away with it at 96.5%. Yeah, that's May. If you go out farther to the end of the year, Google's at 40%.

3:19:13Open AI is at 22%. XAI is at 23 % Anthropics at 7%. And so a little bit closer of a horse race towards the end of the year. What I want is I want benchmarks and evals for video models now because VO3 seems to be just completely running away with the game. But we haven't seen what the next iteration of Sora looks like. OpenAI clearly cares about that. They've been building a product in image VO. In fact, the first consumer product from OpenAI was not a chat model. It was Dolly. Dolly 2 was the first product they released. And then ChatGPT came out. And so it's obviously in their DNA. They're not just going to let Google run away with it.

3:19:57There's so much more that you can do in video. We see this with Runway. We see this with a bunch of other products in the space. It'll be interesting to track that. But we don't really have a great benchmark or eval for that. So we'll have to get one, and then we'll have to put it on Polymarket. Got it. Anyway, we will see you tomorrow. It's going to be a light show. Rora. Oh, yeah. We launched a new product today, a filtered showerhead. Oh, congratulations. If you go there, let's see it. Big use for Rora, filtering your showerhead, cleaning up the water. You're bathing. Yeah, so this is a product that was in the works for a couple years at this point.

3:20:33We were pretty strategic about when the right time would be to roll it out and made the best shower filter in the game on a bunch of different metrics. Flow rate, filtration, ergonomics. And yeah, go check it out. I made a code or I had Brian, CEO, make a code, TBPN, a little discount. And yeah, excited to see how this goes. So we will be back tomorrow. A little bit of an earlier show. I'm doing some traveling and it'll be no guests, just me and Jordy chopping up the timeline. Take a throwback episode. I think these are some of our best. It's going to be fantastic. I can't wait. So it'll probably be an hour, hour and a half, around 10 a.m.

3:21:16If you're looking to tune in, we will of course let everyone know and it'll be in your RSS feeds. And if you're listening to the RSS feed, please go leave us five stars on Apple Podcasts or Spotify. Do it. Ben's pointing a gel blaster at us right now. Thank you for watching. Take care, guys. We'll see you tomorrow.