Speaker unverified: prediction

Transcript context

…ng before there was a crypto thesis and, you know, it was a primary driver of some pretty significant returns at at that prior firm. Um, and so luckily he knows the feeling and we actually have a term internally that we use called uh skin rippy which is really gross but like you want to rip your skin off. You are so >> like you could not be more passionate about this. You want to literally rip your skin off and then run around in a muscular skeletal suit apparently. And so >> there there's a visual for us. Paige, >> there is investing in Uber and >> it's now a horror film. Correct. Correct. >> That was a slasher film. >> Biotech. And >> you want to go full Hannibal Lectar on this investment. I get it. Kately passionate about it. Like a serial killer. >> Exactly. Deeply deeply obsessed. >> Right. >> And so because of that, you know, we we use the term iron sharpens iron all the time here. And you know, having the conversation when I came, you know, it was it was that night. I put it in in Slack and I was like, "Guys, I just I I have seen the future. I have been invited behind the curtain. Philip and the team are going to do something phenomenal. And I really, like I said, he he really made me work for it because that's how you get better. That's how you get smarter. That's how you get sharper. So, um, I was able to make a good enough case. Um, I I visited their offices up in, uh, Seattle where we have a few other companies like Stoke Space. And even touring the facility, uh, which was post check writing, um, I was like like, thank goodness because I I really had to go all in, right? And after uh after being able to convince the the court that that it was the right decision to make, um I I did not expect to see the public and and kind of big big names in the space kind of do this really 180 on it, you know. I mean, at the time of investment, there were there were people there were um very very smart people that had actually created websites to like take it down like to say like StarCloud is the dumbest thing. You know, orbital data centers are the dumbest thing ever known. And then it was like every that feeling of okay. So I wasn't totally crazy uh so fast. Usually it takes a longer time when you're especially when you're making a non-conensus bet like that. But I I I feel very um grateful to be a part of it. >> Yeah. And and thank you for taking us inside the curtain there. I think it's like really interesting to to hear how partnerships get these non-conensus bets over the line. Uh and Paige, you know, sometimes there are um data points that lead to opportunities. And one of the data points is, hey, the cost of putting a kilogram in space has gotten unbelievably cheap compared to historical costs. uh you know here we're looking at you know almost 10k back in let's say 2014 to put a kilogram in space and now um you know it's it's come down this year down to 4k and you know projection is in 2030 and then 2040 going down to 1500 and maybe even $273. This is uh obviously because of space x. Oh, I was amazed by how large the market actually was when you dug into the details. So I think that was a big lesson is when there are questions be even $273. This is uh obviously because of space x. Oh, I was amazed by how large the market actually was when you dug into the details. So I think that was a big lesson is when there are questions around it like what is the opportunity if this company is wildly successful in the next 10 or 20 years and focus on that rather than what other folks may be saying about the opportunity >> and >> other folks who are not talking to the founder and who are not doing the primary research and are not having the the blade sharpening blade uh research. Caitlyn, have you ever looked at the um a trend and tried to find a startup based on a trend? I.e., hey, GP, everybody's going to have GPS in their pocket. What does this enable? Or kilograms of space going down. What does that enable? Have you have you done that? Or can you give us an example of doing that? >> Yeah. No, this is actually that that's one of my favorite ways to to sharpen my own thinking on something. um is if you can develop a thesis, especially as a generalist, I think that we we have a huge opportunity to to really, you know, the the world is our playground, right? And so to be able to develop a thesis independently based on things that you've read and thought and feel um or just generally want to see in the world, then you can reverse engineer that back into hey um you know I you know Paige you you said it Twitter is still a very or excuse me um is is a very very fertile place for a lot of folks building um you know without political commentary like it is a really great place to source deals and have dialogue. Um but what what we've done in the past and something that I've done there's a few thesis that I've worked backwards from. Uh one is infrastructure for the inevitable. And so if I if I believe for example launch cost is solved. If I believe for example that um you know there's going to be this entirely new race um within the stars and for humanity like I can say okay great um or or AI for example if we believe that AI is inevitable um how does the world work well I know that I don't want to be downstream of a lab for example I know for a fact I don't want to be downstream of a lab I want to be uh investing where the labs are reliant on me right I need to be their their supplier for something and so I think a lot about supply chain I think a lot about um being upstream so that that there's reliance not just from a revenue standpoint but also a learning standpoint because that is how you get your best data is talking to customers. So there there is this infrastructure for the inevitable where I've been able to reverse engineer into companies like HGEN uh who are solving um you know heavy industry the it's on-site hydrogen production for all heavy industry. Um they're they're building down in Southern California and just have a phenomenal opportunity to take out uh natural gas in a way that I think is is just absolutely unheard of. page, we should connect on Knox and Hgn and if there's any amazing synergies there. But uh that that's a good example of saying like, oh, I I do believe that that industry and you know, this whole trend around the re-industrialization of gn and if there's any amazing synergies there. But uh that that's a good example of saying like, oh, I I do believe that that industry and you know, this whole trend around the re-industrialization of America and American dynamism is important and so what what are some of the flaws and bottlenecks within our existing systems? >> I've been thinking about this specifically with unmetered tokens. I don't know if you either of you have experienced unmetered tokens in your life, but somebody gave me like a um an API key to um they gave me an API key to a bit tensor subnet that had put like GLM51 from Z.AI. And I was like, okay, I'm not being metered. Great. instead of doing this daily this cron job to do research on something I was like give it to me every hour >> and it starts giving it to me every hour and then I'm out in Hawaii on my vacation with the family and I'm like getting these like incredible nuggets of wisdom and then I was like okay and every four hours I want you to look at your last four and then uh you know drops to me and tell me what you could have done better and then rewrite your skill every day to be better at finding this type of information and I was like, "Oh my lord." When you have unmetered, then this becomes like a free human. Like I'm I'm getting a free human that doesn't sleep that I can put on a task until I tell them to stop doing the task. Whoa. And that is going to be possible, I believe, when everybody has a $20,000 desktop, a 10 to $20,000 desktop. I don't know if you saw the Perplexity Computer announcement page, but they now with Perplexity Computer are going to automatically load um on your Apple silicon a and I'm going to predict right here. Perplexity is getting bought. the the first thing the new CEO of Apple's gonna do is buy Perplexity because Perplexity now allows you to load it allows you to load page a open- source model in their harness and when you do a job it'll go to the local model first and if it gets you the response it thinks is good enough fine and then it will go to the cloud and do through your open router another cheap open source one and then fall back to a uh finally fall back to maybe a frontier model. Those that concept of everybody buying a 10 to 20k desktop to me page seems the future if you could become unmetered. So thoughts on either working backwards to a thesis. You heard a couple of different ways to do that or working backwards from a thesis. Mine is unmetered tokens. Um, do you have one of those that you've done recently or just your thoughts on my my idea of Apple buying Perplexity and having their own search engine and having their own harness that they could put into iPhones and Mac Studios and MacBooks with 128 gigs of RAM? >> I mean, that sounds pretty incredible. We should uh call the new CEO of Apple and let him let him know. Um, I guess one of the thesis we developed in the early part of 2024 was around multimodal AI. So, I'd gone to a conference and I'd seen one of my LPs who helped build the wearables business at Qualcomm. um basically discussing the incredible opportunity in application layer AI and specifically in multimodal AI which if…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.