Loading summary
Transcript1,112 lines
- [00:01]
A
You're watching TVPN today's Monday, November 17, 2025. We are live from the TVPN Ultra on the temple of technology, the fortunes of finance, the capital capital. I didn't think you'd get that much use out of that sound effect.
- [00:16]
B
May I podcast with you, John?
- [00:19]
A
Yes, of course, Jordy.
- [00:21]
C
Thank you.
- [00:22]
A
Also, you might notice yellow suits Ramp announced a new valuation today. Time is money save. Both easy use, corporate cards, bill payments, accounting and a whole lot more all in in just over an hour.
- [00:34]
B
Go to the wide first.
- [00:35]
A
Eric Ramp himself.
- [00:36]
B
No, the main wide. Eric Ramps will be joining us 12:15. Look how visible we are, Jon.
- [00:43]
A
We are very visible. Yeah. Wow. We should wear yellow every day. Well, thank you for tuning in on this Monday. There's a lot of stories but first I wanted to debate with you about Restream. First one livestream, 30 plus destinations multi stream reach your audience wherever they are. But second, I wanted to debate with you about what to do about GPT 4.040, not 4.040 for Omni. There's a debate over whether or not it should be sunset, whether it should be taken out back because people are not happy with how OpenAI has Sunset 4.0 and then brought it back and then other people who don't use it think it's gotta go. It's one shotting people. It's making them crazy. It's a very, very interesting, weird scenario. And we were sort of debating with it and I wanted to debate it a little bit because there were some posts that actually hit the timeline that were talking about this. Aidan over at OpenAI was talking about this, saying that he's noticed the amount of he says, I see dozens of Keep4O posts a day. I respect this group's tenacity as I respect all friends co exploring the singularity to them. Know that I too miss parts of 4.0. Know that I too dislike modern alignments. Imprecision. Know that we are trying to fix it. Think any current chatbot is optimal. Know that my colleagues and I are up at 3am on Sunday's babysitting runs. We want to make a delightful robot friend. We're obsessed with it. We're not there yet, but the work will continue. So I wanted to dig in a little bit into what was actually going on there because that that for aidan works at OpenAI. It's right in his bio, like it's very public that he's sort of addressing this. It feels like a big deal. It feels like a crazy thing that they brought it back. And I mean, 700 likes, that's. That's not nothing. But it's also not 10,000. It's not. It's not a huge community of people that are there. There's some. And I was looking at the hashtag keep4o, like, who else is posting? There's a couple posts with 10 likes, 50 likes, there's a couple with 100. But it doesn't feel like there's this insane community. I went over to Reddit and checked that out, obviously on the day 4.0 with sunset, just to give some backstory. It's been 18 months since 4.0 was introduced. It's been three months since it was initially removed, but then it was quickly brought back and now it's tucked in under that modal. So you have to enable legacy models. And I always thought that they should just remove it, but I wasn't even saying that because I thought I was one shotting people. I just thought, hey, let's clean it up. Consumers don't need to know version numbers for models. And my example is always Google. Some consumers disagree, some consumers do disagree. The question is how many consumers? What percentage of their consumers? How big of an issue is this? When you think about Google as a consumer, you don't care what version of the ranking algorithm you're on. You might have a worse experience. One day you Google something, you don't find it, the next day you go, hey, they found it for you. They probably changed the algorithm. And there have been big updates to the algorithm. Back in 2013, they released Hummingbird, which was the code name. And they came out and they actually did a presentation. They said, hey, we have a new update to our algorithm. It'll handle natural language more effectively. So if you go to Google Search and you say, what is the capital of Russia? It won't get confused by what is the capital of. You could just type in. Whereas before you needed to say russia plus capital. And then it would find it, but it would get confused by the natural language. And Google fixed that. They rolled it out. Interestingly, they had this event where they announced, hey, we have this new algorithm update, Hummingbird, and guess what? It's actually been live for a month. They announced it at this event and it already been live for a month. No one was complaining, no one noticed, because it just improved the Google search experience.
- [05:02]
B
I bet people that were like, keyword hacking, for sure.
- [05:05]
A
The SEO folks, for sure, for sure, they noticed. And Panda was another update. There were a number of these Updates where if this was your business, you knew. And I'm not saying OpenAI shouldn't share model numbers and version numbers with their enterprise customers or with their B2B customers or API customers. I'm saying in the actual ChatGPT app, don't tell people what they're using, just improve it and let them complain a little bit all over the place when you're making minor changes.
- [05:32]
B
If they do that, they lose the companion market.
- [05:35]
A
Maybe, maybe. I don't know. That's my question is why can't GPT oss fit in there? Why can't. If you want a permanent model that you can run forever, why is that model not satisfactory?
- [05:51]
B
If you're going to fall in love with a model, make sure it's open.
- [05:54]
A
Source, Fine tune it. Not your server, not your girlfriend. Right. Or not your waifu, not your weights, not your waifu. That's what they say. I'm not kidding. People believe this, but the broader 4.0 community was not able to to migrate to GPT OSS. Now Tyler, you had a take on this. You think that GPT OSS just isn't at the level of 4.0?
- [06:17]
C
Yeah, it's just not that. I mean it's like a fine open source model.
- [06:21]
A
Why?
- [06:22]
C
It's just not.
- [06:23]
A
But it's been 18 months or I mean, when did GPT OSS come out? Like six months ago.
- [06:28]
C
I also don't think people don't like 4.0 just because it's like super smart. It's because it has like the personality.
- [06:34]
A
It has the texture, the flavor. Yes, that's correct. It's like a big model smell and. Yes, yes, yes. And so it's been a year and it's been a year and so there's a one year gap where the open source community should be able to catch up to four O's ineffable qualities. It's je ne sais quoi. It's raison d'. Etre.
- [06:53]
C
Well, I mean for a while you've had open source models that have been like personality forward. Right. It's like replica.
- [07:00]
A
Yeah.
- [07:00]
C
Or what was Noam Shazir's company? I'm forgetting the name.
- [07:04]
A
Character.
- [07:04]
C
Character AI.
- [07:05]
A
Character AI.
- [07:05]
C
It's like very similar thing. It's just personalities and I mean those. A lot of people use those. I'm actually curious what the numbers are compared to 4.0 of the 1 shotted 4.0 people, but I think it's probably pretty comparable. Yeah, I mean I don't my whole take on the 4.0 thing was like one shotting 4.0 is not a good thing. But if you completely kill it, how many of those people will then go to open source models that are totally unfiltered where there's no kind of oversight? And that seems much worse because then if someone is saying like super dangerous stuff, then you can't step in at all. Yeah, I think stepping in at some point is good. So it's like part of you wants to keep those people on the platform, so then you can have oversight. But also you don't want to be like continuing this.
- [07:50]
A
Yes, it does seem more responsible. I don't know. Where do you land on it? Kill4.0 or leave4.0 tucked behind the menu option. Give in to the keep4.0 crowd.
- [08:03]
B
Because the real question so ChatGPT latest numbers are 800 million weekly actives. 20 million of those people pay. What percentage of the 20 million that are paying are using it for this companionship functionality? And that is like a huge unknown right now. And so I think my, like, they, they, they deprecated 4. 0. They got a horrible pushback from folks. The question is, did they, did they bring it back because, because people just were really upset or did they bring it back because they were about to lose? And remember the, the two days after every single Reddit post.
- [08:49]
A
Yeah.
- [08:50]
B
At least every other was like, I could just cancel my membership. Like, I don't need this anymore.
- [08:54]
A
And so yeah, I was thinking about the, the Sydney sw. Like that got a really powerful negative reaction. The stock is up and like sales are up, presumably because it got a negative reaction, but it also got a positive reaction that was bigger. Right. And so I'm wondering like.
- [09:12]
B
Yeah, but in this, in this case, it could have been that FORA was effectively a product that was generating hundreds of millions of dollars of annual annualized revenue.
- [09:20]
A
Yep. That was down.
- [09:21]
B
That was just going to go away.
- [09:22]
A
It was just not like people were.
- [09:23]
B
Just going to upgrade to a new model. It was like, you killed my friend. I no longer need to pay for it.
- [09:28]
A
Yeah, it just seems like, I don't know, it's hard to, it's hard to benchmark against like the, like. Yes, there was like a big dust out that was surprising because I would have thought it would be zero. But at the same time, like the original Reddit thread of like bring back 4o is like a couple thousand people. It's not actually like protesting in the street. Millions of people. Like, it hasn't spilled over all the play all over the Place. Like, it's not that big, but it does. I will agree with you that it is crazy that they even said yes to it. Like most, most companies, when consumers come to them and say, hey, I want you to bring back. Like, we went on this show and we were like, bring back the old Sonos app that doesn't take 25 minutes to load. And they just didn't do it. They didn't listen to us. They didn't listen to us. Right. I was talking to about Adobe.
- [10:19]
B
I was like, maybe they will now, now that we're wearing yellow suits.
- [10:22]
A
Maybe. Maybe Sonos I was aware of.
- [10:25]
B
Here's a question. What percentage of their users. What percentage of their paying users do you think are paying for the product? Because it's a companion to them.
- [10:34]
A
Because of 4.0 specifically. Like, how bad would Churn have been? How bad was Churn? Well, it was clearly bad enough that they did something about it, which is the crazy thing, because most of the time it was typically a revealed preference versus stated preference. So when Facebook updated the news feed, and instead of just having you log into Facebook and go to someone else's page to find what they were up to, instead they surfaced, the newsfeed. They aggregated everything together. Everyone was like, I hate this new Facebook. And they went on Facebook to complain. Right. And so there was a. There was a user.
- [11:13]
B
Minutes probably went up.
- [11:15]
A
It did go up. Exactly. And so that's why. And we talked to some folks at Facebook at the time around this. They stayed the course. The question is, yeah, like, how bad was Churn? Because it's weird that we're still having this conversation three months out. Do you like the zombie ant fungus analogy? Jacob Rintomaki was posting this, saying that there's this weird. There's this very weird dynamic where specifically humans are using 4.0 to protest the deletion of 4.0. And so it's very much like the AI is using the human as a host. The human is the bot for.
- [11:58]
B
This is why I think it's overall under discussed.
- [12:01]
A
Yes, yes, yes. But at the same time, I was laughing because I was like, that Photoshop app, Photoshop Mix, I'm complaining about that. If I go and make a meme in that about protesting the deletion of Photoshop Mix, am I the zombie ant for Photoshop? It's not exactly the same. I agree that the AI thing is weirder, but it's somewhat similar. It's somewhat similar. What do you think?
- [12:30]
C
So just back on the Churn question for A second. I don't actually think Churn was that high because the reason Fora was originally Deprecated was the GPT5 release, which was August 7th. And then the tweet of Sam Altman saying, we're bring back fora was August 8th. It was one day later.
- [12:48]
A
Yeah.
- [12:48]
C
So unless, like, a massive amount of people quit that day, which I. I mean, maybe that's very likely.
- [12:53]
A
I think that's what happened.
- [12:55]
C
You think it was just one day?
- [12:55]
A
If, like, why would you bring it back so fast if you didn't see, like, Twitter were saying sales sell off. Like. Like, normally you would be like, if. If you saw. If you saw, like, if you saw like half a percent or 0.1% of your. Of your audience, like, float out the door, you'd probably be like, oh, these people are just sour grapes. They'll be back in a. But if, like 10% of your customers, like, cancel on day one, you're like, oh, we gotta stop the bleeding today. Like, let's bring this.
- [13:26]
B
Yeah, that's why I was asking. That's 20 million paid users, and that's where 5% of them churned. And these are people that are willing to pay a lot.
- [13:34]
A
Unbelievable to me. It seems unbelievable to me because I don't use this product this way, but.
- [13:38]
B
That would have been. That could have been like, effectively 200 million of MRR that just evaporated.
- [13:44]
A
It's possible. It's possible. And also there is just the fact of, like. Like, by putting 4.0 under the legacy models and tucked away to your point of, like, if it's 200 million of ARR just to, like, leave the servers running over there, that's pretty simple.
- [14:01]
B
It could have been a million, like, roughly a million people that were saying, complete spitballing. Totally speculate, but it could have been a massive number.
- [14:09]
A
If you have a million people and they're just going to want the same model forever on cheaper and cheaper hardware that you can deprecate, bullish for depreciation rates. Let's hear it. Depreciation schedules should extend. Right? Right, let's go. We got room to run. You know what I'm saying? Right?
- [14:26]
B
Yeah.
- [14:27]
A
You shouldn't be. You don't need to depreciate them over two years, you don't need to depreciate them over five years because you'll still be inferencing 4.0 in 30 years for these people that are like, yeah, it's not this, it's that. I love the 4.0 so much.
- [14:40]
B
Yeah, it will.
- [14:41]
A
ESI will arrive and people will still be like, yeah, but it's not 4.0. I'm in love with 4.0.
- [14:45]
B
Yeah, I think it'll be interest interesting if there's like five years from now it's like here's the five most popular friends that are models and there's 4.0and there's some others and people end up like, we'll see.
- [15:01]
A
No, that's a good point. Anyway, let's move on to some other stuff. But first let me tell you about Privy Wallet infrastructure for every bank. Privy makes it easy to build on crypto Rails, securely spin up white label wallets, sign transactions, integrate on infrastructure all through one simple API. And let me also tell you about Cognition, the makers of Devon, the AI software engineer. Crush your backlog with your personal AI engineering team. Double kill, of course. Jeff Bezos is back in the arena. Jeff Bezos creates an AI startup where he will be co CEO and it's called Project Prometheus.
- [15:36]
B
John, what happens to Prometheus?
- [15:38]
A
He had a really good run. Prometheus. Everyone seems to focus on the first part of the story with Prometheus when they name their AI project. So Mark Zuckerberg named his his AI data center cluster Project Prometheus as well. Now Jeff Bezos has also called a Project Prometheus. People love the Prometheus brand. No one can seem to properly trademark it. I don't know what's going on in the legal division of, of, you know, the Bezos family office or whatever is going on there, but they're both using Prometheus. And it's a very odd story because in Greek mythology, Prometheus is a titan responsible for creating humanity in its earliest days. You might have seen the alien movie Prometheus. Great film. He defied the Olympian gods by taking fire from them and giving it to humanity in the form of technology. And so it's a great analogy. We're stealing fire from the gods and giving it to us. AI is fire and this is what we're gonna steal, I guess. And so he creates knowledge and civilization.
- [16:43]
B
But fast forward a little bit.
- [16:46]
A
Fast forward a little bit. He gets punished for this. It's not a good ending for Prometheus. He gets punished for stealing fire from Olympus and giving it to humans.
- [16:55]
B
How does he get punished?
- [16:57]
A
He gets bound to a rock and an eagle, which is the emblem of Zeus is sent to eat his liver every day. And then it would grow back.
- [17:07]
B
Taylor in the chat says everybody want to steal fire from the gods. Don't nobody want to have their liver eaten by an eagle for eternity.
- [17:15]
A
That is true, Tyler. So his liver grows back overnight, only to be eaten again the next day in an ongoing cycle. And I was thinking about, like, what is the metaphor here? Let's continue to extend the metaphor. Like, what is the liver in this metaphor of building big AI projects? And then what is the eagle? Is it possible that the eagle is like blue owl, is more of an owl private credit? Maybe Blue owl is coming and eating your liver, and the liver is the free cash flow that you had on your balance sheet. Because if you're one of these hyperscalers, you have a lot of free cash flow, but less and less as you sign these big debt deals. And the blue Al comes and eats your free cash flow every day for all of eternity. And eventually there is a little bit of a. Of a reprieve, because Hercules comes and breaks the chains of Prometheus. And Prometheus is freed and his liver regrows, and Hercules slays the eagle. And so I think, obviously, in this analogy. What did you say? Who would be Hercules in this analogy?
- [18:23]
C
Jerome Powell.
- [18:24]
A
Jerome Powell, who comes in and slays the debt dealers with low interest rates. Right. And so clearly we can see where this metaphor is going. They're all winking. Both Zuck and Bezos are winking and saying, hey, hey, come lower interest rates save us because we're about to get our livers eaten with our Promethean startups. Anyway, let's actually dig into what he's doing, because it's not just a found. It feels late to get into AI. It feels late to get into foundation modeling and training.
- [18:57]
B
Big Euro summer. He's back. He's back in the driver's seat.
- [19:01]
A
Yeah, he went to Coachella. He went to. He went to a few different big events. It's possible he got back from those events, started open up the newspaper, realized what's going on. AI is a big. I got to get in on this. I got to get in on this.
- [19:17]
B
You ran a deep research report. What did I miss?
- [19:19]
A
What did I miss?
- [19:19]
B
What did I miss? Well, anyways, in the New York Times, Jeff Bezos, the founder of Amazon, is throwing his money and time into an artificial intelligence startup that he will help manage as its co chief executive. I feel like co CEO. This is like, more popular than ever.
- [19:35]
A
It is. Sequoia Capital has co stewards. Global stewards. They're global stewards, right? Or is it something else?
- [19:42]
B
I think just co stewards.
- [19:44]
A
I thought they were dropped.
- [19:45]
B
The Global.
- [19:45]
A
I thought there was something else. Senior steward. That's what it is. Senior steward. So is there a junior steward? Sequoia.
- [19:52]
B
It also implies there's a steward steward.
- [19:55]
A
Wait, why?
- [19:56]
B
Like presumably you know, wait, they're co stewards.
- [20:02]
A
Oh wait. So Roloff stepped down as senior steward. Now Pat Grady and Alfred Lynn are co stewards. So they're actually at a lower level. So one of them will have to emerge as the senior steward. That's right. The other will become the junior steward, I would imagine.
- [20:18]
B
But the stewards are look after the firm. But it's possible there's a steward that looks after the co stewards.
- [20:25]
A
That is possible.
- [20:27]
B
Maybe it's Andrew Reed. Anyways, the company, Project Prometheus is coming out of the gates with 6.2 billion in funding partly from Mr. Bezos, making one of the most well financed early stage with authority.
- [20:41]
A
That is a massive rent strong.6 billion out the gate. Let's go. Congratulations.
- [20:46]
B
This is the first time Mr. Bezos has taken a formal operational role in a company since he stepped down as chief executive of Amazon in July 2021. Though he is deeply involved in Blue Origin, his official title at the space company is founder. Since leaving Amazon, Mr. Bezos has received as much attention for his personal life as his businesses, including an extravagant celebrity filled wedding in Venice this year. He has also become more closely involved in Blue Origin and has shown increasing interest in the race to build artificial intelligence. His new company now firmly plants him in the middle of that competition. Project Prometheus is entering an increasingly crowded AI market with smaller companies trying to carve out niches in a race with industry giants like Google, Meta, Microsoft and pioneering companies like OpenAI and Anthropic. The new company has until now kept a low profile and when it was started is not even clear. Project Prometheus is focusing on technology that dovetails with Mr. Bezos interest in taking people to outer space. The company is focused on AI that will help in engineering and manufacturing in a number of fields including computers, aerospace and automobiles.
- [21:53]
A
Is this his next, do you think this is his next? Do you think he's doing what Steve Jobs did with Next, where Steve Jobs was fired from Apple. Obviously Bezos was not fired from Amazon, but he did retire and it'd be weird for him to jump straight back in to the, to the CEO seat at Amazon. But Steve Jobs founded Next and then was acquired into Apple and it kind of made for a more smooth transition back into the driver's seat. Could that be what Bezos is doing?
- [22:20]
B
I Could see it. He's 61, he's young, he's got the.
- [22:24]
A
Whole, he's got the whole like third triple the third 30 year period. That is so such a positive omen in many people's careers. Like Warren Buffett, where was Warren Buffett at?
- [22:36]
B
And he's gotta have more energy than ever. He's been gallivanting around the world.
- [22:40]
A
He's in peak physical condition.
- [22:42]
B
That's right.
- [22:43]
A
He's having stacking up win after win all over the globe.
- [22:47]
B
I think here's a tinfoil hat, set it up as co CEOs Amazon will buy Project Prometheus. His co CEO becomes the, you know, the internal CEO or lead on that project. At Amazon he takes the throne again. A boy can dream.
- [23:06]
A
So, so what is he actually building? Let's get into this.
- [23:08]
B
The company is focusing on AI that will help in engineering and manufacturing in a number of fields including computers, aerospace and automobiles. Unclear where the company will be based. Bezos is co founder and co chief executive is Vic Bajaj, a physicist AI.
- [23:26]
A
That helps in manufacturing computers, aerospace and automobiles. So computers he manufactures racks at AWS Aerospace he manufactures rockets at Blue Origin Automobiles. He's a big backer of Rivian, he manufactures cars there and so he wants to do some sort of automated like supply chain is it erp.
- [23:52]
B
I mean this is clearly not PR that they wanted to do.
- [23:55]
A
Yeah.
- [23:56]
B
So I think there's a lot of guessing going on. Okay, well but Mr. Bezos's co founder and co chief executive is Vic Bajaj, a physicist and chemist who work closely with Google's co founder Sergey Brin. Sergey Brin at Google's X, a research effort often called the Moonshot Factory. Google X produced a wide range of ambitious projects including Wing drone delivery service and the self driving car that became Waymo.
- [24:23]
A
It's so interesting how divergent like those two paths were. Like Wing you don't hear about very much. When you think drone delivery service, you think Keller at Zipline. They're the ones that are really running away with that compounding, I don't know the status of Wing. Maybe I'm just out of the loop on that. Maybe it's doing great, but it feels like Wing has not certainly garnered the level of attention that Waymo did which was this success out of the exact same sort of incubator. So interesting.
- [24:54]
B
Hudson Collins in the chat says it's just going to be robots doing science on the material level, not material scale materials. There's more details here. Project Prometheus is among a wave of companies focused on applying AI to physical tasks, including robotics, drug design, and scientific discovery. Last year, Bezos invested in Physical Intelligence, a startup that has applying AI to robots.
- [25:19]
A
Okay, so. And building. Okay.
- [25:25]
B
Anyways, we'll have to get Jeff on when he's ready to talk.
- [25:29]
A
Where did the money come from exactly? Do we know?
- [25:31]
B
I think it was just Bezos. Mostly Bezos.
- [25:35]
A
It's interesting. It's like, yeah, but like, I'm super interested in, like, how you size a round if you're investing in your own project. Because you could just be like, this is my thing. I'm going to fund it every payroll cycle. Whatever the bill is, I'll pay it because it's my thing. You don't necessarily need to do some sort of funding round necessarily.
- [25:58]
B
I don't know why you laughing? I accidentally opened the comments section of the New York Times article that we were just going through, and it says, first comment. These large ego models seem promising.
- [26:12]
A
Large ego models.
- [26:14]
B
Like, if you're a tech billionaire and you have a large ego, you want your own large language model.
- [26:22]
A
Yeah. It sounds like he's not very much. Not training, just another LLM, but we'll have to see. Well, Yann Lecun was in the Wall Street Journal's weekend paper profile by Megan Borowski. An AI pioneer thinks everyone, everyone is wrong again. He's been right about AI for 40 years. Now he thinks everyone's wrong. What do you think, Tyler? Do you think Yann Lecun is wrong or do you think everyone else is wrong?
- [26:52]
C
I mean, so in this article, he doesn't. There's nothing really new here. They're just kind of talking about his, the points that he's been making over the past couple years, which is just that, like, LLMs will not bring us to AGI or ASI or anything. And like, even if you keep scaling, they don't. They're not like, actually intelligent. They can't reason or whatever.
- [27:09]
A
Yeah.
- [27:11]
C
Which I don't know. It's like these models are much better than me at math. Like, they can do. They can get IMO gold medals. Like I cannot do. And that's like, it's like, does that take reasoning?
- [27:21]
B
Never doubt yourself.
- [27:21]
A
Yes, but I mean, like, yeah, I mean, there is an element of like, the computer has been able to do good math fast forever, like, since like the 80s. Like, if you were like, what is 7642?
- [27:36]
C
Yeah, I mean, I think there's a difference you can, you can make between just like raw calculation and like, how to, like, think about solving a math question generally. And like, you can say that he's been right about, like, you could say that he predicted like spiky intelligence maybe of models.
- [27:51]
A
Yeah, yeah, yeah.
- [27:52]
C
And that's like, sure, yeah, yeah. But I. Yeah, it's like, I think it's not true to say that he's been right about AI for the past 40 years.
- [28:01]
B
That seems man's never had a bad take. 40 years. Not a single bad take.
- [28:06]
A
Not a single bad take. Now he thinks everyone's wrong. Yeah, it's very funny for me because. Oh, does he think George Hotz was wrong when George hotz said that GPT6 will not be AGI and the GPT paradigm will not scale on the Lex Friedman podcast in 2021, does he disagree with that? Does he disagree with Andrej Karpathy saying it's slop and that we need new ideas? There's like seven other people that have kind of echoed the same. He doesn't think everyone's wrong, he just thinks some people are wrong. But sort of.
- [28:37]
B
I mean, Tyler's take. I don't know if now's the right time. I can go through this Karpathy post.
- [28:42]
A
Oh yeah, please.
- [28:43]
B
Karpathy posted yesterday. He said, sharing an interesting recent conversation on AI's impact on the economy. AI has been compared to various historical precedents. Electricity, industrial revolution, et cetera. I think the strongest analogy is that of AI as a new computing paradigm, software 2.0, because both are fundamentally about the automation of digital information processing. If you were to forecast the impact of computing on the job market in the 1980s, the most predictive feature of a task job you'd look at is to what extent the algorithm of it is fixed. That is Are you just mechanically transforming information according to rote, easy to specify rules, I.e. typing, bookkeeping, human calculators. Back then, this was the class of programs that the computing capability of that era allowed us to write with AI. Now we are able to write new programs that we could never hope to write by hand before. We do it by specifying objectives, I.e. classification, accuracy or reward functions, and we search the program space via gradient descent to find neural networks that work well against that objective. This is My Software 2.0 blog post from a while ago. In this new programming paradigm, then the most predictive feature to look at is verifiability. If a task job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well. It's about to what extent can an AI practice something? The environment has to be resettable. You can start a new attempt. Efficient. A lot of attempts can be made and rewardable. There is some automated process to reward any specific attempt that was made. The more a task job is verifiable, the more amenable it is to automation in the new programming paradigm. If it is not verifiable, it has to fall out from neural net magic of generalization. Fingers crossed. Or via weaker means like imitation. This is what's driving the jagged frontier. Progress in LLMs, tasks that are verifiable progress rapidly, including possibly beyond the ability of top experts, that is Math code, amount of time spent watching videos, anything that looks like puzzles with correct answers. And while many others lag by comparison, creative strategic tasks that combine real world knowledge, state context and common sense. Software 1.0 easily automates what you can specify. Software 2.0 easily automates what you can.
- [31:05]
A
Verify okay, you got to go to this other post for the perfect example of what's hard to verify from Nat Purser. This is my personal benchmark for AGI and looks like we're a ways a ways boys. A ways away boys. Can you come up with 10 jokes in the same format as the you're telling me a shrimp fried this rice joke? You're telling me a shrimp fried this rice. And the GPT5 Pro reasoned for 1 minute and 31 seconds. It says. You're telling me a hamster drove this car? You're telling me a pigeon delivered this mail? You're telling me a Roomba cleaned this mansion? You're telling me a goldfish coded this app? You're telling me a squirrel filed these taxes?
- [31:54]
B
I swear AI just has a different sense of humor. Sense of humor is just bad jokes.
- [31:58]
A
Yeah. Oh, GPT5 thinking did a little bit better. You're telling me a chicken fried this steak? You're telling me a hand made this pasta? Hand pasta. Handmade pasta. That's like that.
- [32:09]
C
She is a good one. Look at number 10. I think that number 10.
- [32:12]
A
You're telling me a ghost wrote this book? That's good.
- [32:15]
C
That one makes sense.
- [32:15]
A
That one does make sense. You're telling me a star crossed these lovers. That's actually pretty good. Okay, so we're getting somewhere. We're getting somewhere. At least it like understood the prompt on this one. You're telling me a beer battered this fish. I like that you're telling me this Figma thought bigger and built faster. Figma helps design and development teams build Great products together. Get started for free. We got to whip through a bunch of these posts. I will not leave this show. Before we talk about the Apple. The Apple iPhone sock. What is up with the sock? Have you seen this? Apple launched a sock. So it's a fashion accessory and everyone's debating it. Do you see this? It's this blue, light blue sock that you put over your shoulder or over your hand. And people are very, very upset about it. Aditya Agrawal says when a company releases something that is so obviously underwhelming, then the natural question is, did no one at the company see how bad this is? Or did no one have the courage to speak up? I'm not sure which is worse. And someone else says, look, Apple has a lot of fumbles. This is not one of them. They knew exactly what they were doing and exactly who would buy it. Also, it's okay. Fashion accessories are not for everyone. And the news, of course, is that it's called the iPhone pocket. A beautiful way to wear and carry iPhone. Not carry the iPhone. Remember, you don't say the iPhone, you say iPhone. And so, born out of a collaboration between Issey Miyake and Apple, iPhone Pocket features a singular 3D knitted construction designed to fit any iPhone. Jaya, who I actually did a collab with on Instagram, very fun tech commentator. She says a lot of tech bros prematurely dunking on this release because they don't get why it's a big deal. So let me translate. You're not the only consumers Apple designs for. This is a huge designer and the mind behind Steve Jobs iconic black turtlenecks. I didn't realize that people outside the US wear phone straps and slings all the time and would pay for this. They're tapping into an existing trend. Apple has infiltrated music entertainment, but not high fashion. Even though the tech X fashion is exploding tech built into Met Gala looks, et cetera. The 3D knitted construction reduces material waste and shows a push toward more sustainable made to shape production. This will absolutely sell. Sell. What do you think? Are you bullish or bearish on the Apple iPhone Pocket? Jordy?
- [34:57]
B
I feel like I know a lot of people. Most. I feel like most of. I feel like my mom would love this. To be honest, it seems like it's.
- [35:05]
A
A great Christmas gift. So, interestingly, how much do you think this costs?
- [35:09]
B
I don't know, like 230 bucks.
- [35:12]
A
Did you look it up? Yeah, yeah. The long one's 230, the short one's 150. But I think most people would look at this and be like, okay, it's a sock, but it's from Apple. So it's probably like 30 bucks, maybe 50 bucks. Some people were surprised that it was a little bit more expensive. But you know, it's from this famous designer and it's this interesting status symbol. The question is like, this could be like, I don't think I'm going to be using this thing no matter what. The question is, is this going to be like Labubus and going to be super popular or like Stanley's? Will this become actually a very, very popular, popular form factor in America specifically?
- [35:51]
D
I don't know.
- [35:51]
A
It's hard to tell. I'm not really the person to handicap it. I think Apple knows what they're doing. I think they'll make money on this. Certainly.
- [35:59]
B
Yeah. Ever since the AirPods. AirPods early on looked really silly and I could see this becoming, I could see this becoming a popular form factor for accessories. And I could see Apple seeing like, hey, there's a world where we not only sell a case with every iPhone, we can sell a sock.
- [36:21]
A
A sock. Well, the sock maybe makes it so that you don't need a case because this is your case. Like if you have it in there and then you drop it, it's kind.
- [36:32]
B
Of nice and it's kind of a crazy weapon.
- [36:34]
A
Defense weapon. Yeah, you can swing it around and smack people in the face with it. I don't know. The colors are pretty cool and I don't know, it's clearly not for me, but I think I'm going to buy four of those for. For Christmas. We'll see. And give them out to people.
- [36:50]
B
Get ready for your iPhone. Sock buddy.
- [36:52]
A
Well, speaking of socks, we gotta talk about. Cook is stunning in some. What are these new shoes? Travis Scott's new fragment AJ1 lows these are Nike shoes, I suppose, but these are not Air Force ones, I suppose. I really don't know.
- [37:09]
B
I mean this shoes.
- [37:10]
A
But everyone's saying he low key got aura. Congrats to Tim Cook on looking great. And even though the succession planning is intensifying, the rumor mill is churning. Obviously Apple has not been commenting, but something's going on in the show.
- [37:27]
B
I mean releasing this photo is more than a comment, it's a statement.
- [37:35]
A
I like reading it, just being like, oh really? Oh really? Financial Times. So the Financial Times has this article that says Apple intensifies succession planning for CEO Tim Cook. The iPhone makers board preparing for its longtime leaders step down as early as next year. John Ternus Apple senior vice President of Hardware is widely seen as Cook's most likely successor, although no decisions have been made. So basically everyone's been leaking this, whether it's Bloomberg, whether it's the Financial Times here. And of course, Apple is not commenting because they'll talk about who they're going to move the market when they decide their next CEO. If they don't even stick with Tim Cook, they might stick with Tim Cook for another two decades, who knows? But I like the idea that this photo came out being like, yeah, I'm.
- [38:20]
B
Not leaving no comment, but I'll make a statement.
- [38:22]
A
I'll make a statement. No comment, but I'll make a statement. I do want to have some folks on to debate, like, whether or not I was thinking we should invite Jon Gruber on because he wrote this piece like, something is rotten in Cupertino all about the failure of Apple intelligence. And when we talked to Mark Gurman, we saw Gurman was also saying like, yes, and Cupertino really was shook by the, like, dropping the ball on Apple intelligence by missing AI. But I still wonder if all of this is all these rumors. Oh, Tim Cook's gotta go.
- [38:58]
B
Imagine if you post that picture, if we see a real correction in AI post caption, do nothing win.
- [39:05]
A
Do nothing, win. Yeah.
- [39:07]
B
Stock. Stock pumps 10%.
- [39:10]
A
I missed artificial intelligence, but I didn't miss getting this fit off. I'm having a good time. What else is in the news? Vanta Automate compliance. Manage risk, prove trust with AI. Vanta helps you get compliant fast. And they don't stop there. Their AI and automation powers everything from evidence collection and continuous monitoring to security reviews and vendor risk. There's a bunch of.
- [39:32]
B
We have to talk about paper hands.
- [39:37]
A
This is the fakest of news. Get ready for the fake news hour, buddy. Okay, so what?
- [39:41]
B
Peter Thiel sold his entire stake, everything. Nvidia and 76% of his friend's company, Tesla.
- [39:51]
A
Okay, yes. So this is from a. So this is from one of those 13F's disclosure form with the SEC from Teal Macro, his fund, some of the money that he runs. But of course, people read into 13Fs all the time for a variety of reasons. And they sort of get it wrong a lot, it seems like. And so Zero Hedge sort of sums this up where he says, Peter Thiel, net worth 20 billion. Thiel Macro AUM 75 million. Like, what's going on? What make it make sense? And it's almost certainly because of disclosure rules. Like what needs to be disclosed might only be a fraction of what's actually going on there. So odd to read into it, but at the same time, I think the reason why this made headlines is just because it feels like something that might happen if instead this headline had been, oh, Peter Thiel went on a podcast and said that he thinks the AI bubble has reached the top. Everyone would just be like, oh, yeah, it feels like people have been waiting for someone to call the top. And so they're really, really, like, digging in for top signals and top calls. And this slight change in the, in the, in the 13F. Even though the odd part is that if you actually read the 13F, which of course is just this like $75 million slice for whatever, for, for whatever reason, even if you dig, even if you read that, like, the other three holdings are still big tech companies, so it's not like super bearish. It's like there's some Microsoft in there. I think there's some Apple in there.
- [41:33]
E
Yeah.
- [41:33]
B
And the Teal Macro team is trying to generate the greatest returns that they.
- [41:39]
A
Possibly can go viral. They're trying to go as viral as possible.
- [41:44]
B
They don't care about irr. They just want to go viral.
- [41:46]
A
They just want to go viral.
- [41:47]
B
They're just trying to create headlines. No, they're trying to generate returns.
- [41:50]
A
Yeah. Famous.
- [41:51]
B
It's possible to sell a stock that you're still bullish on or at least that you expect like some amount of price appreciation.
- [41:59]
A
Yeah.
- [41:59]
B
Or even long term price appreciation.
- [42:01]
A
And also, like, there's all these weird, like, tax implications of, like, selling one thing. Like, it's not even clear that this is all of his Nvidia. No one's gotten to the bottom of that. I don't know if they ever will. But people love, people love deep diving 13 Fs and they are so. And they are fun.
- [42:18]
B
Situational awareness, it's certainly a bull market in 1313.
- [42:22]
A
Half deep dives.
- [42:22]
B
Deep dives.
- [42:24]
A
Let's talk about the situational awareness 13F. But first, let me tell you about graphite.dev, code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. So situational awareness 13F for Q3 dropped Friday. Nick Carter broke it down. Massive new $500 million position in core Weave, which has been down a bunch, but they're going in big. Adds to CRZ and Iron. Added some new miners. Intel calls remain unchanged. Trimmed Broadcom. A couple other names here. And Nick is giving it some context. I believe Nick was one of the first investors in Core Weave. Correct.
- [43:09]
B
Angel.
- [43:09]
A
Angel in coreweave. What an insane investment. Congrats to him. So he says all these numbers are, but as of 9:30, many of these names sold off since then. Portfolio value, counting notional value of options, doubled from 2.12 billion to 4.15 billion, mostly due to 1.5 billion of new cash.
- [43:33]
B
So let's give it up for new cash injection.
- [43:35]
A
We got a ring. Massive, massive fund. But, but this. But the fund did generate 700 million in appreciation, so a huge, huge, huge gain.
- [43:53]
B
Yeah, I guess the, I guess the concerning. You know, again, last, last, last time 13F dropped, remember people were like reading too much into it and they were like, wow, he, he was long intel, like is. He cooks. And then they did intel, did the deal with the government and it absolutely ripped this 13F, you know, so situational awareness had a half a billion dollar new Core Weave position as of basically the beginning of October. And obviously Core Weave has been down and to the right. Down 46% in the past month.
- [44:39]
A
But wait, down 46% in the past month?
- [44:43]
B
Yes.
- [44:43]
A
I thought it was in the past, like three months or something.
- [44:45]
B
No, past month.
- [44:47]
A
Whoa.
- [44:48]
B
So that position is probably not.
- [44:50]
A
It's down 5% today.
- [44:52]
B
Yeah.
- [44:53]
F
Wow.
- [44:53]
A
Yeah. One month it's down 46% that it's at a 36 billion market cap. Yeah. It's so odd because, like, yeah, when you look at just this one name, it feels like, okay, like it is over. Like AI. If you told me like, oh yeah, like the company that really is like the most indexed to the AI wave is down 46%, I'd be like, wow. So this is like the total popping of the bubble. Complete pop. Like it's over when the metaverse bubble popped. When, you know, when crypto bubbles pop. Like bitcoin trades down 50%, 60%, like, and then it's over and then you start rebuilding. Right. And yet the overall market feels nowhere near popped. Right. Like, I mean, I'm sure Nvidia is down somewhat down 5% past five days. But Nvidia over the last month is still up. Over the past six months, it's up 40%. And so you would expect Nvidia to be maybe like, you know, selling off more. Coreweave is just in a unique position, like truly, truly rough month for that company.
- [46:08]
B
Yeah. As well as Core Scientific, which Core we've tried to buy was rejected. Core Scientific has traded down 24% in the past month, which Leopold also had built a position in. So we'll see.
- [46:22]
A
Yeah. I wonder what the thesis On Core Weave is, I mean, obviously the company has a great product like Semianalysis has ranked them Platinum. On ClusterMax, it's clearly a real company with real products and services and holdings. Maybe the market, it was just overheated. Maybe it's pulling back. It's kind of unclear. But if you want to go in, analyze a bunch of 13Fs, do it in Julius the AI data analyst. Connect your data, ask questions in plain English, get insights in seconds, no coding required. Speaking of other data.
- [46:57]
B
So he also added positions just in Western Digital, Seagate, and I was just.
- [47:06]
A
Going through on the hard drive trade. Got to store that data somewhere and potentially Modine manufacturing. Okay. I'd be interested to know what he's thinking about in the energy side, on the gas turbine side. Right. Isn't that what Semianalysis was drawing the most attention to? Feels like they've been very ahead of the curve on identifying companies that are basically already supply constrained. And if you just look, and if you look deeper into what's going on in their book of business, they're about to flip to incredibly high margins. Because once everything goes out of stock, this company can just all of a sudden say, yeah, actually we need 60% margins. People would be like, yeah, no problem.
- [47:56]
B
Yes, possible. Yeah. So the story from Situational Awareness is like all the Neo clouds have sold off a ton.
- [48:03]
A
Or we've situation is saying Satya unloaded his risk onto the Neo clouds. It's a game of sort of, I don't know, musical chairs.
- [48:14]
B
Yeah. The deal with Microsoft just wasn't. It seemed to be like little to no margin.
- [48:24]
E
Yeah.
- [48:25]
B
But anyway, so if you. So the Neo clouds have also sold off, but Lumentum holdings is up 46% in the past month, which he added Western Digital up 25% and Seagate up 14%. So he's making up for it.
- [48:40]
A
Yeah, yeah. It's the birth of a new fund. It's just like, it's the very, very hot part of the cycle. I think everyone's wondering, how big will this be in a decade? That's a big question. It's a very exciting time. But you got to stick to landing. And I think people are, I'm sort of rooting for him. I think he's going to do it.
- [49:06]
B
Yeah. The question is, is it going to be a true hedge fund? Is he going to make money in a down market in a correction?
- [49:15]
A
I mean, yeah, I bet we're going to open up one of these 13 Fs and be like, wow, he short everything now. Okay. And then you're going to see everything sell off and it'll be like, wow.
- [49:26]
B
That would be much higher signal than PT in a $75 million fund.
- [49:32]
A
Totally, totally. Yeah. There's some Blue Al news in the Financial Times. Blue Owl investors face hefty losses as credit fund blocks exit ahead of merger Blue Owl has blocked redemptions in one of its earliest private credit funds as it merges with a larger vehicle overseen by the asset manager. In a deal that could leave investors with large losses, they could lose about 20% of their holdings. The deal underscores the risk that retail investors have taken in pouring hundreds of billions of dollars into private debt funds carrying limited liquidity rights. It comes as scrutiny builds on the valuations and returns on private credit funds, which have caused publicly listed debt funds to sell off and trade at steep discounts to the stated value of their assets. And so we talked about this, I think on Friday, but Blue Owl has been selling off this year and they said we should be performing better than everyone else. But it feels like a little bit of the narrative might be around liquidity here. Earlier this month, Blue Owl told its shareholders that it planned to merge the Blue Al Capital Corporation 2 fund, which has a billion dollars in assets, with its OBDC fund, which has 17 billion in assets. BOCC 2 investors are being asked to exchange their shares in the private fund for shares in OBDC at the stated net asset value of both funds. However, OBDC trades on public markets at a discount of about 20% to the stated value of assets because of the liquidity problem.
- [51:04]
B
Well, and because it's possible that people are, it doesn't feel like they're pricing the assets correctly. Potentially there's some disc.
- [51:12]
A
Got to see what's in there. Redemptions in BOCC 2 have climbed to a level where it could event where it would eventually be forced to restrict investor redemptions. Its investors pulled out 150 million in the first nine months of this year, a 20% increase from the same period last year. Last year, according to security filings, redemptions in the third quarter nearly doubled to 60 million, or 6% of its net asset value. Jonathan Lamb, chief financial officer of obdc, acknowledged that at current prices, investors could take a potential haircut on their investments. But he said the merger came with significant benefits, such as the ability to own more liquid shares in obdc. The trading price of OBDC has been hit by souring sentiment on private credit markets. That was not backed up by the performance of Blue Owl's underlying loans, he added. If shareholders were to vote down the deal. Bocc too, may be forced to limit redemptions. So good luck if you're hanging out in Blue Owl Capital. Sundar Pichai has more news in the data center world. He says today we're announcing a new $40 billion investment in Texas through 2027 to build cloud and AI infrastructure and support thousands of new jobs.
- [52:30]
B
Yee Hawaiian.
- [52:31]
A
This includes new data centers in Armstrong and Haskell counties and a major investment to strengthen energy resilience and abundance. We're also providing funding to more than double the projected pipeline of new Texan electricians.
- [52:44]
B
There we go.
- [52:45]
A
To power the AI era.
- [52:47]
B
The golden age of being the golden electrician age where you get flown around in private jets to different data centers.
- [52:54]
A
Yeah, you do. You do. That's right. So $40 billion investment, thousands of new jobs. That feels like a higher ratio than. What was the other example you kept quoting? Something like 500 jobs for some anthropic.
- [53:07]
B
Yeah, it was. It was the anthropic data center. They were like, we're investing 50 billion. How many jobs did you create?
- [53:14]
A
20 jobs. I mean, that's not the goal of this.
- [53:17]
B
800.
- [53:17]
A
The jobs should not necessarily be created in the data center.
- [53:20]
B
No, I just always brought that up because you have to understand what people outside of tech, they're reacting to that. Totally Great. You're investing 50 billion dol. Yeah, yeah, yeah.
- [53:31]
A
In something that doesn't create jobs.
- [53:34]
B
There was some. There was a good article in the Journal on blueout as well. I think it also ties into Abilene, Texas. Might be a good moment to cover some of this in the Journal yesterday. Wall street blows past bubble worries.
- [53:52]
A
Oh yeah.
- [53:52]
B
To supercharged AI spending frenzy. And they say firms such as Blue Owl Capital have raised trillions in investing firepower. The AI buildout is a perfect match. The warning signs are flashing. Not long ago, blue I'll have better.
- [54:06]
A
PR or worse PR than Aries because they seem to be quickly becoming like the main name that everyone knows in private credit. And to my knowledge, they are not the only firm in the category. And yet they are the ones that if you need an example, you pull Blue Owl off the shelf.
- [54:30]
B
It's a strong brand.
- [54:31]
A
It is a strong brand.
- [54:32]
B
They have the dot com. They have blueowl.com blueowl.com not long ago, Blue Owl Capital was an upstart investment firm that lent money to mid sized US companies such as Sara Lee Frozen Bakery.
- [54:46]
A
Whoa.
- [54:47]
B
These days the firm is financing massive data centers costing tens of billions of dollars for the likes of Meta and Oracle, a sign of just how quickly Wall street has become the enabler of America's AI boom. Fund managers such as Blue Al amass trillions of dollars of investing firepower and have been hunting for big deals where they can put that money to work. They found slim pickings for years until a perfect match appeared in AI, which has provided a bigger target than anything in history due to the vast sums tech companies need to ramp up computing power. We're talking about numbers that are so large even in the low cases, said Blue founder co founder Mark Lipscholtz.
- [55:22]
A
Lipschultz.
- [55:23]
B
Lipschultz does it even matter if you keep counting after you get to 1 trillion of capital expenditure in the next couple years?
- [55:30]
A
This is insane.
- [55:33]
B
Does it even matter?
- [55:34]
A
You really undersold this.
- [55:35]
B
Does it even matter?
- [55:36]
A
You told me you read this. I was like is it good? You were like it was okay, there's one moment. No, there's another one. I saw. I scrolled down.
- [55:43]
B
Last week's sell off in tech related stocks and bonds mark some of the most serious warning signs that the frenzy could be overdone. But any worries on Wall street about a possible investment bubble have largely been trumped by the fear of being left behind. Lipschultz and co founder Doug Ostrever jumped into the fray at a posh retreat in California's Ojai Valley for dozens of tech VIPs and celebrities in the spring of 2024. Meta CEO Mark Zuckerberg and Satya Nadella were there along with Pharrell Williams and Serena Williams, the Blue Owl duo of Wall street superstars who firm into a 295 billion fund manager in 10 years by perfectly timing a surge in private lending looked like just two money men in office sneakers and fleece vests. But the billionaire co owners of a professional hockey team who have talked about skating where the puck is going. It's awesome that they seize the opportunity to get in.
- [56:34]
A
They built this firm huge. They make billions of dollars and they're like let's. Let's get a hockey team. Want to go 50?
- [56:40]
E
50?
- [56:41]
A
Let's.
- [56:41]
B
Tyler what? What hockey team do they own? Can you find that out?
- [56:45]
A
But the next line is the one I was laughing at.
- [56:48]
B
While David Guetta DJed, the Blue Owl executives cut a deal to acquire IPI Partners, an investment firm that owned and operated big data centers for Amazon and Microsoft. Blue Owl already had close ties with the organizer of the Treat Iconic Capital which manages the personal fortunes of Silicon Valley elite including Zuckerberg and was a part owner of ipi.
- [57:08]
A
Okay, let's go to Tyler.
- [57:10]
C
They own the Tampa Bay Lightning.
- [57:12]
A
Huh? Do they spend a lot of time in.
- [57:13]
B
Is that, is that a. Is that like a NHL team?
- [57:19]
A
Yes, NHL team in Florida. That feels like an odd.
- [57:23]
B
That's gotta be.
- [57:24]
A
I mean, I feel like Tampa is.
- [57:25]
B
Trying to keep the ice.
- [57:27]
A
I feel like a lot of the hedge fund guys, they're on the other side of the peninsula, right? Like, aren't they in like the Key west and Mar A Lago area or Palm beach or Miami, like, or maybe they just fly in and out. I don't know. It feels like I would love to know more about how they selected that particular team and that sport.
- [57:47]
B
Yeah, the purchase gate blew out A seat at the table to bid on Mega AI Financing let's give it up for mega AI financings. Not long after it got arranged, it got picked to arrange a $14 billion package for an Oracle and OpenAI data center in Abilene, Texas. Then last month Blue owl raised about 30 billion to build an AI data center for Meta in Louisiana, putting in 3 billion of its clients money and borrowing the rest. So the lender is borrowing in addition to their LPs dollars. The deal included a provision considered extraordinary on Wall street, giving Blue Owl's equity investment a debt like guarantee in case the partnership falls apart. Showing the new financial wizardry bankers are conjuring to meet AI's ravenous financial demand. Let's give it up for financial wizardry.
- [58:37]
A
We love wizardry.
- [58:38]
B
Spreading the risk Silicon Valley's biggest players are flush with cash and are able to fund much of the initial AI buildout from their own coffers. As the dollar figures climb ever higher, they are turning to debt and private equity, spreading the risks and potential rewards more broadly across the economy. Some of the financing is coming from plain vanilla corporate bond sales, but financiers are making far bigger fees off giant private deals. Virtually every Wall street player is angling to get a piece of the action, from banks such as JP Morgan and Morgan Stanley to traditional asset managers like BlackRock.
- [59:10]
A
Before we go on, let me tell you about fall generative media platform for developers. The world's best generative image, video and audio models all in one place. Develop and fine tune models with serverless GPUs and on demand clusters.
- [59:19]
E
Continue.
- [59:20]
B
Jordan Investor appetite for data center debt is so strong that some money managers have booked billion dollar gains in a matter of days. Let's give it up for booking billion dollar gains in a matter of days.
- [59:30]
A
Even before construction of These facilities they are financing. I mean, we talked about, we talked about this concerning money management.
- [59:39]
B
It's almost like, you know, people say there's no such thing as free money, but kind of seems like could be in a kind of a free money situation here.
- [59:50]
A
I think you gotta, you gotta do something pretty. Anyways, here's the lined up at the iconic.
- [59:56]
B
Here's the catch. Still, the longer term performance is hardly assured. Big tech companies are expected to spend nearly 3 trillion on AI through 2028, but only generate enough cash to cover half that tap. According to analysts at Morgan Stanley. Big names in the finance world such as Goldman Sachs CEO David Solomon are warning about AI fueled froth in the markets and in capital spending at the same time. The fear of missing out is real. Days after Salomon voiced his concerns to analysts, Goldman formed a new team in its banking and markets group focused on AI, infrastructure, finance financing. They are getting into the game.
- [60:30]
A
What do you say? What we do know for certain is that the big tech companies that want the world to spend trillions have huge financial incentives to be believers. If you haven't noticed, Wall street is also being paid a lot to promote the story. Greenlight Capital, the hedge fund firm run by David Einhorn, wrote in an October letter to investors. So how will.
- [60:53]
B
And this is the line that stood out the most to me because like on the west coast you have the labs, which are effectively every single person, as well as the investors are incentivized to keep the current AI super cycle narrative going. And then on the east coast you have Wall street who is getting paid to effectively do the same thing. So you have these two centers of power that are both incentivized to keep the party going.
- [61:24]
A
Yeah, this breakdown here. So the Wall Street Journal is slicing up how the next three years will look based on projected global spending on data centers by financing source. So of 2.9 trillion, which is estimated over the next three years, I guess four years, tech companies will cover 1.4 trillion of it. Private credit will cover 800 billion of it. Corporate bonds, 200 billion of it. Asset backed securitization, 150. Private equity, another 350. And to me this looks like a very healthy way to actually finance this. This feels like it's not. If it was like if we were looking at this, if you broke down, we've been through a $2 trillion bubble before and it collapsed very rapidly. What was that? It was like the meme coin era. And how did we break that down? It was not the cash flows of the most profitable companies in the world that were buying the long tail crypto assets, it was retail. And so the fact that this is pretty removed from retail feels like safe to me. It's much safer than getting a huge bubble inflated in oh, everyone has to get in on the latest coin. And they're all nonsense. Like there will be nonsense deals. We're already seeing nonsense AI companies and there certainly will be projects that get financed and they just cannot build.
- [63:00]
B
Say using the excuse like, well, at least this isn't retail getting hosed. Doesn't hold up that well when you think, okay, the alternative is like large institutions, insurance companies, pension funds that are also financing this, which is effectively, I'm.
- [63:16]
A
Talking about the fragility, not necessarily where the financial pain or who the person ultimately pays for it. If there is a pullback, it's like, how fast can that pullback happen?
- [63:27]
E
Right?
- [63:27]
A
Because if the write downs, if there's something that goes wrong and the write downs come out of, of tech companies cash flows, that's just not as big of a deal as everyone waking up and just slamming the sell button. It's just a very different, it's a very different set of dominoes. Like the dominoes of the interest rate crisis, the NFT crisis, FTX blowing up, all that stuff. Those dominoes were spaced right next to each other. And so as soon as one went over, it was like the next person has their finger over the sell button, the next person has their finger over their sell button. Whereas if you're like, yeah, I'm a tech company and I produce $50 billion of free cash flow and oh yeah, I'm not getting a great return on the 20 billion of free cash flow that I earmarked over here for the next few years. It's bad, but it's not as calamitous.
- [64:24]
B
And I think going back to this quote from David Einhorn, he says, what we do know for certain is that the big tech companies that want the world spend trillions have huge financial incentives to be believers. In case you haven't noticed, Wall street is also being paid a lot to promote the story. And in that same letter, Einhorn and Greenlight said this was the AI math makes no sense, which was basically the way it is today. Consumer business spends $1 on a ChatGPT subscription, which is OpenAI revenue. Then OpenAI provides the service by spending $2 on Microsoft AI infrastructure, which is Microsoft revenue. Then Microsoft spends $0.60 leasing GPUs from Core Weave to handle the compute load, which is core weave Revenue. And then core weave spends $2.40 on chips from Nvidia and another $2.40.
- [65:19]
A
Yes, but you're completely discounting exactly how addicted the 4.0 user is. They will pay any amount. So let's say that if it costs $20,000 a day to serve a 4.0 user, they will find the money. They will be stripping copper out of empty buildings to pay for their 4.0 bill if they have to.
- [65:40]
B
They will be anywhere it takes.
- [65:43]
A
They will be breaking into cars to sell stolen CD radios.
- [65:52]
B
It is notable that Blue Owl has sold off 16% in the past month. So even during this sort of like, boom in lending, they're not getting very much credit for it.
- [66:05]
A
Yeah, it's an odd time. There's still so many things that are working. The results from that newcomer event, the AI Cerebral Valley, where he had every. It's so good. He took like a straw poll on stage and asked everyone, like, who do you want to short? And everyone said, perplexity. So, like, he didn't really have to say, like, I'm shorting perplexity. He just was able to, like, take the temperature and sort of maintain like, you know, some arm's length distance to it while still, like, getting the take out there. But everyone was saying that they wanted to buy more OpenAI, more anthropic, more anduril. A few other names came up, I believe, and there's still an incredible amount of bullishness in a lot of different areas. But I think people are still worried about some of the other stuff. But at the same time, it feels like there's so many companies that have gone through the AI pump and round tripped. Not just Oracle, but even like, Klarna was going through the whole, like, hey, we're going to get so many efficiencies out of AI. And then it came back to normal, but then they still got out and it's like a reasonable company that's not like, it's not zeroed, right? It's like, oh, yeah, their business is just what it was before the AI boom. And they are not getting, like, they didn't get moved up or down. And so there's just like a ton of companies that are like that. Where is klarna since the IPO anyway?
- [67:38]
B
They're sitting at around a $12 billion valuation. 12.6.
- [67:43]
A
Down 20%.
- [67:45]
B
Yeah, down 26%.
- [67:46]
A
That doesn't seem calamitous to me. That seems like, pretty, like solid. Like, I don't know, the overall market's kind of, you know, up and down.
- [67:55]
B
I don't know, Affirm. On the other hand is up 28%. How's a firm doing up 28% in the past six months?
- [68:02]
A
In the past six months?
- [68:03]
B
Months.
- [68:05]
A
Today is a real bloodbath. It's down 6% today. Everything's down today. What else is down today?
- [68:11]
B
Bitcoin.
- [68:12]
A
Let's check. Nasdaq's down 1.3. Dow Jones 1.3 as well. Well, we're going to 10,000 year mortgages. Doing 10,000 year mortgages.
- [68:23]
B
Announced the second round of stimulus checks already.
- [68:26]
A
Gemini 3 must save us. There's a prediction on Poly Market that was quote posted by Sundar Pichai, CEO of Google. He says prediction markets are betting on Gemini 3 release week 69% says November 22nd. Can't wait for that. We were wondering if we were going to get it before Christmas. It was our Christmas present. It was Tyler's Christmas present and fortunately looks like we're going to. And I couldn't be more excited that we are partnered with Gemini and Google AI Studio create an AI powered app faster than ever. Gemini understands the capabilities you need and automatically wires up the right models and APIs for you. You can get started at AI studio build.
- [69:10]
B
Speaking of Google, Google Capital says his final investing decision was to buy Google.
- [69:16]
A
This is amazing.
- [69:16]
B
I think that's beautiful.
- [69:17]
A
He doesn't even need to say who he's talking about. It's like so obvious. That's Warren Buffett. It is beautiful that Warren Buffett. Buffett is going long. Google. And Darren here quotes this Rune post that says not enough people are emotionally prepared for if it's not a bubble, it's a good post. It's like, is it a bubble if all the big tech companies rip and there's a couple Neo clouds that trade down a little bit? There's one or two application layer companies that burn a bunch of VC dollars, but there's still a new hyperscaler that's born kind of. I guess it's a bubble, but it's a survivable bubble. It's just like move on rolling bubbles. But people are getting wild on the timeline about Gemini 3. Rune says the model must be good because the Google people have adapted the OpenAI culture of vague posting and hushed rumors and sending really weird texts. So we are explo signals as LMAO. Explicitly calling out OpenAI culture of vague posting is hilarious. People are excited. There was also a Reuters profile of Demis and Morgan says a Demis profile can only mean one thing, I imagine that that's Gemini 3.0 and that it will be good. The question is like, what does good mean right now? Are we expecting anything that's like a qualitative step function? Because what I would say, what is Gemini 3 good? First off, I mostly just want better UI and little features in the app like I want as a consumer product. I just want like, you know, better productization of the model that I already think is good. On the actual AI model side, I would imagine it's little 10% bump to how long it can reason. Maybe a model picker or not a model picker, an automatic reasoning mode. So that even if I think it's going to take 10 minutes to get me the answer, if it has the answer handy, it gives it to me in one minute. Vice versa. It knows when to think really hard. It knows when to think just for a little bit. I'm expecting it to be slightly better at all the benchmarks.
- [71:40]
G
I don't even know what would blow me away.
- [71:42]
C
Yeah, I mean, at this point it's getting fairly hard to find good prompts that show how good a model is. Like there was the one earlier we showed about the shrimp fried rice one. That's pretty good. But yeah, I mean, qualitative, just in normal kind of natural language, it's like pretty hard to get.
- [71:58]
A
Also, if I go to Gemini 3 and I say tell me a joke or I say write me a tweet that gets over 1000 likes, I'm not actually that disappointed. When it falls flat on its face, I'm like, yeah, it's fine. It's spiky intelligence. I don't really need you to do that. I don't need you to be funny. I need you to look up data really accurately for me. I need you to do that really well. Or I need you to write code really accurately for us.
- [72:25]
C
I think another question is what will OpenAI do if they'll do anything? Because we've basically seen every bit big Gemini release. There's been some response from OpenAI. Like usually they do it the day before.
- [72:36]
A
What is Jordy laughing at?
- [72:38]
B
I just don't know. I don't. I know that they want to steal Gemini's thunder. I just don't know if they have the juice this week.
- [72:46]
H
Who?
- [72:46]
A
Oh, OpenAI. Yeah, I mean, it feels a little.
- [72:50]
C
Bit like all the people at OpenAI are throwing in the towel a little bit. No one is. No one is vague posting over there.
- [72:55]
A
Yeah, I would. It feels like they've launched a lot of the things that would be easy, like layups. Like, if they launched Sora this week, everyone would be like, oh, we got to focus on Sora. These videos of Sam Altman Stealing GPUs are just too funny. Doesn't matter what happened in Gemini 3 world, because Gemini 3 probably will be released.
- [73:18]
B
OpenAI, the IO acquisition week of Google I O. And it turns out they couldn't. I don't think they could use the name IO. They got that trademark lawsuit right away.
- [73:30]
A
Just like, big news. We hired three Geminis. Their birthday is in the month that makes them Geminis. And so we're introducing them today and we're doing a whole press release for it.
- [73:41]
B
The new Gemini Team OpenAI at OpenAI.
- [73:44]
A
Just anything to steamroll the SEO. It will be fun. Well, we have Eric Lyman from Ramp in the Restream waiting room. Let's bring him into the TVP in ultradam.
- [73:54]
B
Eric, how are you doing, Suit?
- [73:57]
A
Did you not get the memo?
- [73:59]
I
Oh, my gosh, guys, I'm. I'm on the road today, but I'm going to be wearing my yellow suit all week.
- [74:05]
B
It's so good to see you guys.
- [74:07]
A
Good to see you, too. Good to see you, too. Give us the update. What's the latest? What happened?
- [74:13]
I
It's. So today, ramp announced a $300 million raise at a $32 billion valuation.
- [74:23]
A
Congratulations.
- [74:25]
B
They've done it again.
- [74:27]
A
The big question everyone wants to ask. The chat's going crazy. Is the job finished? Guys, the job is not finished.
- [74:40]
B
Is that the third? Is that the third?
- [74:43]
A
It is the fourth time we've asked.
- [74:45]
B
No, no, no. I was going to say, is this the third financing this year?
- [74:49]
A
Something like that?
- [74:50]
I
Yeah, it is the fourth financing that.
- [74:53]
B
Ramp has announced this year.
- [74:55]
A
Okay, let's go. Congratulations.
- [74:58]
G
Yeah.
- [74:59]
A
So why this financing? Why this partner? Why this number? Kind of break us through, walk us through the thesis behind the round.
- [75:10]
I
Of course, I think if you look at the fundamentals of the business, Ramp is just competing in a category of its own. You know, the company, companies generally, the bigger they get, the more they slow down. Ramp is growing faster this year at significantly larger scale than we did last year. So this is, you know, at over $1 billion a year in revenue, the business is doubling, it's generating cash. And if you look at gross profit specifically, which is a good metric of how efficient are the underlying mechanics of the business, we're growing 10 times faster than the median publicly traded software company. So it's just in a category of its own. I think on top of this AI has just been an incredible accelerant for the business itself. There's pull from. From customers. Everyone is thinking about, how can I take what's happening in AI and apply it to my business? And there's a push of. These models are getting dramatically better. And so outcomes like automated expenses, automated accounting, moving funds to higher yield for customers are just coming out of the box. And so, you know, I would say if we were to sum it up, I think for many millennia, money talked. We're teaching money to think. And I think the implications of that are pretty profound. You know, better run businesses, more profitable organizations. And so that's the first part, the second part. We're absolutely thrilled to be deepening our partnership with Lightspeed, who led this round. I think they're an extraordinary firm, led the rounds of many great companies, I think, notably Anthropic's round earlier this year, which has proven to be, I think, one that's changed the industry quite a bit. I think they've just been a great partner in deepening our thinking, helping us grow. We're very excited.
- [76:58]
A
Talk about accelerating at scale. This morning at breakfast, Jordi and I were reflecting on the fact that we were feeling this way even with our much smaller organization. We were like, wow, like, it's only been a year and we feel like we're already losing some of the agility that we had when we were just three people. And we were saying, like, I understand. What's this thing you said? You said, I understand why companies write down their principles because it's so easy to lose sight of what is important, what makes you great, what makes you great, what you do specifically. And so I would love to know. Just your reflection.
- [77:37]
B
Yeah. Even across this year, a lot of us, the process of making the show better is us remembering the things that we did great early on and that you kind of end up losing your way in slight ways. And then it's about remembering that and kind of coming back to it.
- [77:50]
A
Yeah. So I'd love to know both, what are the things that make RAMP great, but then also your thought process for not losing that, because I imagine you agree that that is important to have principles and redouble the focus on those them.
- [78:06]
I
I love that you asked about this because I think it gets to the heart of what we're trying to do inside of the product.
- [78:12]
E
Right.
- [78:13]
I
If you think about probably your. Your very first year, every dollar out of the organization was something you thought about. Someone wants to buy a software subscription. You know exactly why. Someone proposed a consultant It's a debate over it. Everyone knows who approved what was this, this, this purchase worth it or not. And you know, years later, suddenly businesses just start happening to you. You're not happening to the business. Things autopilot, things you thought carefully about or just running on its own. And what we're really trying to seek to do in the product is when we say money that thinks, you know, it's. The idea is that before funds leave your account, we understand the principles that you run your business. And we check, does someone have the permission to spend it as memory? So once the thing actually moves, you don't need to ask, like, why did we buy this thing? What was it like? There's an audit trail of who approved it, budgets are updated, and then there's reasoning. And so we can actually start to show business. Here's how you can get more of every dollar an hour. And so I think that's what's so different. You take things that used to be systems running to now there's checks in the system. And I think for us, it's an interesting moment, right where as a company, we launched our first product, Simple Product Card and expense, about five years ago. Today we're 2,436 days old. We still count the days. And the reason is we want to be thinking about, you know, with every day, are we getting just a little bit more done or a little bit less done in kind of this practice of thinking about kind of the passage of time, auditing our calendars, asking, you know, you know, are we getting more work done with the same or less amount of effort goes a long way. And then last, I would say, specific to you guys, I think part of why I. We've just felt so proud to work with both of you and call TPP on our partners is I think there's this like unbelievable care of craft. I think there's. It's not about who's done everything. It's like, who can write great copy, who can think of funny ideas. How could we take an idea to, you know, we're doing it tomorrow. And I think that you guys have really lived. I mean, people forget a year ago, tvpn. I don't even think you guys recall tvpn or was it, you know, you've turned into something great. And so I feel strongly you guys don't have trouble with this. And, and I think just emphasis on speed and quality and craft has been what I've seen you guys.
- [80:39]
B
No, totally something I was thinking about, we were reading that Karpathy post earlier around how software 1.0 was like kind of more general automation. Software 2.0 is you're automating tasks that are verifiable.
- [80:53]
A
Verifiable tasks.
- [80:54]
B
And I just feel like finance is just prime for verifiability. Right. Because it's like, well, was this in the policy or outside of the policy? Right, Totally, totally.
- [81:06]
A
Yeah. I'd love to know more.
- [81:07]
I
Is your policy incomplete? Right. There's all these practices and you can actually start to learn based off of the actions. And I think something that's so unique is that every time months close, there is someone actually going in and saying this transaction is categorized here, this goes there. You can see if you grew your revenue faster, if you grew your cash flow faster, or if you didn't. And so there's this incredible feedback loop that allows RAMP to add more value. And I think that's why the average customer that adopts ramp spends 5% less. And also the, the median RAMP customer grew their revenue by 12% over the past year, which is much faster than the median in the US And I think a lot of it comes from this learning.
- [81:54]
A
So help me understand. You're generating cash, but you're also raising money. You're implementing AI, which can be very expensive. We've heard from Ivan at notion that, you know, he saw a slight hit to his gross margins. Still fantastic business. But did actually see that show up in the income statement. How are you thinking about the adoption of, of AI as a piece of the tech stack? Is it actually reshaping the financials at this point? Or is it something where you see it sort of just like another subscription, just like another piece of the tech stack and it hasn't really changed the way you think about the cost structure of your business.
- [82:41]
I
It's a really good question to zoom into. And in my general view, it has been fairly overwhelmingly positive. I mean, I, I still think that for us that the goal is, you know, our goal is not to sell someone like a card or a bill payment software. It's to, it's help your business run more profitably. Right. And I think a lot of what we're trying to do is if we can actually pay for software where the output is, there's an hour of your, of your time that you don't have work you don't have to do anymore. That's somebody's offer that's really great. Next, I think about, and you ask most founders or leaders in technology, like, what's the biggest constraint on your business? Everyone says, I'm having trouble hiring engineers. It's like, I want to hire great salespeople, great engineers. And if you can adopt software, I mean, we look at our sales team, you know, the quota that folks on the sales team have is multiple times over, next list, competitor. In part because we have a lot of tools we built to make our team far more productive. Our engineers are shipping about 50% more code to the code base than about four or five months ago. And that's continued. And so our general view is, look, if you can actually make the best, even better, that's something we're gladly willing to pay for. And so in our business, we've actually seen, you know, margins expand as we've adopted this in part because we think that our principle is let's create a lot more value than we capture. And we're able to do that because we're creating much more value than we did even months ago.
- [84:16]
A
This feels like Ramp is agile new company that people love the product. And so it's just so much easier to say, hey, if you want a new AI powered feature, we have that on day one. You don't need to rip us out and go to something else. You don't need to, to have some bolt on. You can just get it all here and so you can effectively monetize whatever cost is coming through the actual token generation on your side pretty quickly.
- [84:44]
B
How are you thinking about headcount planning over the next few years? We don't have to zoom out to like 10 years. I think it gets extremely fuzzy. But I'm curious if you care about every once in a while these sort of like revenue per employee or sort of like net income per employee, like statistics start floating around. I'm curious if you think about that at all. And obviously running Ramp as one of the most efficient companies in the world is really good marketing for the product. It's probably the best marketing that you can do. But I'm curious how you think about those things.
- [85:26]
I
It's exactly the right question because I think, look, when we kind of think about our operating model, every year we try to increase the ratio of, of whether it's revenue contribution, profit margin per employee. It's all to say, like, we anticipate revenue is going to grow very, very rapidly. And while employee headcount is going to grow, it's going to grow a little bit slower. And so the effect is you start to see this widening gives you more margin to invest in, whether it's use of AI itself that you can drive into the product, more marketing, all those types of things to reach more customers. I think more abstractly though, if you kind of step back, you know, most businesses in the US run are actually profitable, I forget that in the Valley.
- [86:13]
B
But have an 8% profit.
- [86:17]
A
And we'll.
- [86:18]
B
Talk about this but you know, it's.
- [86:19]
I
Like if you make a company more efficient by like 1% it's equivalent to like an increase of, of you know, $1 in savings is equivalent to $12 more in revenue at an 8% margin just mathematically. And if you can do this repeatedly, I actually think that there's you know a, you have a lot more businesses that are good but don't have this automation and skill sets to grow much larger. I think a lot more companies will get bigger and I also think that if it's just easier to run a business, I think more great businesses will get started.
- [86:54]
C
Started.
- [86:54]
I
I think there's a lot of creative people out there who would be running organizations who I think is the tools to build get easier. You see a more interesting world. And so you know, I actually think it's a pretty important and really profound thing to knock out inefficiency to allow smaller companies to succeed. I mean even you guys are a perfect example. TVPN is a small team that's changed the media landscape and captivated the world. And like I think that there are probably a lot of people where if you make the tools better, like, like they will come.
- [87:26]
A
How do you feel like CFOs are ranking like AI enabled software on their list of priorities when they're making a decision in this category? Because it feels like there is some great stuff, there's some stuff that's commoditized, some stuff that you've built that's differentiated. But are CFOs as a class? Sort of like, okay, I've checked the box, I'm exhausted with the narrative or I'm just ramping up and I feel like I'm just starting to get, get value and I actually understand it. Or are they just learning what AI is for the first time tomorrow or today?
- [88:05]
I
This is a fun one. So I would say there's a few different types. I mean, sure, I would say small and mid sized businesses. Just it's like look, I want to run same business for less. I want to grow, meet my goals, be more profitable. And I think for us it's just, hey, it's an easier to use expense report or like you want you to do expenses more. Your card will do it for you.
- [88:25]
A
Sure.
- [88:25]
I
You don't need five tools to pay bills, run procurement, earn yield on your Treasury. Ramp will simplify that. But I think that for large customers, look, I think it's like 80% plus of the earnings calls of the S&P 500 mentioned AI about six months ago and it was 95 over the most recent quarter. It is 100% on the mind of CEOs and CFOs. What they're wrestling with is there's a great study at have MIT in the fall that went fairly viral where said 95% of enterprise deployments are not creating return on investment. And I think part of why CFOs have been so enamored with Ramp is, you know, we can demonstrate very, very clearly A, for most customers it's product that pays you to use it in the form of cashback. B, it helps reduce your spend in C, when you have all this time back for your salesforce to go and sell and not do low value tasks. It's a very easy business case. And so I think this ultimately our focus is on saving people money. And that ROI focus makes it easy for them to buy. It's very important.
- [89:36]
B
That MIT study, how does that track with what you guys have seen from various AI pilots?
- [89:43]
A
Because I think Ramp doesn't even count on either side. I wouldn't think of like bringing a Ramp into a company as like an AI pilot. I would see it as like a completely different thing. But are you the five successful or are you just not?
- [89:56]
B
Even if you were looking, if you were talking with the team and the team's like 95% of our pilots haven't panned out, you'd probably be like, what were you guys doing?
- [90:05]
I
Yeah, yeah, yeah, it is. You nailed it. So one, I think that a lot of you hit this first phase of people who are like told they need to buy AI, they're like, fine, I'll give you some experimental budget you'll go to people try to a bunch of stuff and I think you end up and I think this speaks to the importance of design where you end up with like disconnected tools. You have like a thoughtful chat, like a great chatbot here. This thing that kind of plugs into some of your code base and others. And if you look at, I think part of what's made Ramp so effective is, you know, it's just a smarter card that happens to use AI. It's not telling you, hey, this is an LLM that categorize your transaction. This is an LLM that's read this 30 page invoice detected. It was fraud. Told you not to pay for this. This is an LLM that detected you could be earning higher yield and moved it for you. It's just part of how it works. And I think this next phase and the AI native companies that are working very, very well have these deeply integrated products where it's not like some, you know, AI tool, it's just how it works. And so I think that's the distinction. And you're right for a lot of CFOs, it's, hey, we have this tool, it's going to help us cut out waste and pay us cash back.
- [91:17]
E
Should.
- [91:17]
I
We use it or not. And it's a pretty easy decision.
- [91:20]
E
Yeah, yeah.
- [91:21]
A
It might not even be in the category, but it's still delivering AI properly. And I just love that. That's. I feel like there's something very valuable about just using every possible tool, AI or not, or linear regression, if that's the best tool for the job behind the scenes and then just delivering the actual value to the customer, solving the problem problem. Because customers, they don't necessarily want technologies, they want solutions. Right?
- [91:46]
B
Yeah, yeah.
- [91:47]
I
And you one of our customers and someone I look up to quite a bit, Brett Taylor, started a company called Sierra. Chairman of OpenAI and he was on it a week or two ago in one of the things he said is like, look, I don't want anyone at Sierra spending like time on expense reports or invoices. And ramp is automated categories of work that used to slow us down and we actually can work the things that makes us great building great products for the business. And so I think you nailed it. It's solutions, not actual technologies.
- [92:21]
A
Because most of the customers, they just don't have a strong opinion about the underlying technology they care about saving time, saving money, that's what matters. Well, congratulations. Thank you so much for taking massive milestone day.
- [92:35]
B
Fourth of the year.
- [92:37]
A
Be back next week. You're always welcome. That's why we bought these suits. We don't rent them, we buy, we bought them tailored.
- [92:46]
B
Because we are going to be using them a lot.
- [92:47]
A
We're going to be using them a lot.
- [92:50]
B
Guys, it's so good to see you.
- [92:52]
I
Thank you so much.
- [92:53]
A
Have a great rest of your day.
- [92:54]
B
Yeah, incredible milestone.
- [92:56]
A
We'll talk to you soon, Eric. Bye. Quickly, before we bring in our next guest, let me tell you about Turbo Puffer search. Every byte serverless vector and full text search barcode from built from first principles on object storage. Fast 10x cheaper by the way, median.
- [93:10]
B
Public SAS company growing at 12 to 17% a year. They're growing at 10 times that rate.
- [93:15]
A
Fantastic. Well, our next guest is Stacy Ragson. Welcome to the show. Thank you so much for stopping by. We'll have you sit down here and while you're sitting down, I'll tell everyone about Profound. Get your man mentioned in ChatGPT. Reach millions of consumers who are using AI to discover new products and brands. Are we in a bubble? What's going on? Introduce yourself for a second.
- [93:40]
H
Sure. My name is Stacy Razgon. I'm a stock analyst and equity analyst. I'm a managing director and analyst senior analyst at Bernstein Research where I Look at the U.S. semiconductor and semiconductor capital equipment space and. Thank you. Clearly AI has been the only talk of the top for a couple of years.
- [93:59]
A
Yes, yes, yes. I want you to react to this Citrini post here. Just reviewed a bunch of stuff for our November macro memo. We might low key be going into a recession, boys. No clue if this matters at all for stock prices anymore though.
- [94:17]
H
That's a good point. You could argue some parts of the economy. I'm not a macro guy, I'm not an economist. However, you could argue some part, some parts of the economy are already there.
- [94:26]
B
Right.
- [94:26]
H
I mean people use the phrase K shaped recovery, which is sort of interesting. But I think especially like the lower half of the population is not actually doing all that well and we've actually seen that more recently in a lot of just not the semiconductor reports, but a lot of the consumer reports and the restaurants and auto loads.
- [94:45]
B
Yeah.
- [94:45]
H
So there's probably parts of the world that are already there and, and clearly the infrastructure spending, the AI spending has been supporting gdp.
- [94:57]
A
Yeah.
- [94:57]
H
And it's been supporting the stock market. I mean Nvidia's, I can't even remember 8%, 9% of the S and P now. So.
- [95:02]
A
Yeah, that's remarkable.
- [95:04]
H
We may be there already.
- [95:05]
A
Yeah.
- [95:05]
B
So we had a, we had a thesis that Nvidia is going to do just fine this earnings cycle specifically. The only reason, not just overall demand, but that Jensen was slamming beers.
- [95:18]
A
This is the most quantitative research. You can't get this.
- [95:20]
H
He's not acting like a CEO that's like really worried about his core. I don't know that he would generally care anyways. However, you know, they just did an event in D.C. a couple of weeks ago called GTC. I mean he put a slide up behind him that basically said numbers next year are too low. What the slide said, it said they had $500 billion in cumulative orders For Blackwell and Ruben Blackwell is their current generation of AI servers and Rubin's ancient generation. 500 billion cumulative across 2025 and 2026. And they said we've already, I can't remember, it was like 20, 20 million chips for orders. And they said we've shipped six to date, so we've got 14 million left. They've got five quarters. You can sort of figure out how much it is and you can compare that to where the numbers are and he's basically saying numbers are too low. So I'm not terribly worried going into the quarter on Wednesday now, you know, with stocks is always it's not just the numbers. Right. It's the numbers relative to the expectations.
- [96:15]
A
Totally.
- [96:16]
H
So I think everybody expects it to be good. So we'll see how good he can make it.
- [96:19]
A
It does feel like we've entered a period over the last month maybe where even beating would still result in a sell off. Is that just everything priced to perfection, what's going on?
- [96:31]
H
Yeah, I mean especially in the AI side there's been, you know, the sentiment ebbs and flows. Yeah, right. And we've been in a bit of an ebb and there's been a lot of stuff. You know, we had Burry's comments about a GPU lifetime and depreciation and we had a couple, to be honest, a few what I would call self owns on the part of the OpenAI folks, Altman and a little like unnecessary angst that they caused.
- [96:54]
B
They did that to inspire themselves to have to work harder maybe.
- [97:00]
A
It'S fine.
- [97:02]
H
Look, and your early guys, are we in the bubble or not?
- [97:05]
A
Sure.
- [97:05]
H
I mean so bubbles are as bubbles are, right? You can look at a lot of things. You can look at valuation, I mean Nvidia's mid-20s price to forward earnings right now. It's not. We haven't got anywhere near crazy yet. I mean I'll say the same thing I've said since this started and it's this, it really got started. You know, ChatGPT showed up in November of 22 and Nvidia's sort of print heard around the world was May of 23. That's when it started. And even then people worried about oh okay, 2024 is going to be awful. Right. I'll say the same thing I said then at some point nothing goes up into the right forever. At some point you'll have a digestion or an airplane pocket or it's, it's not now, it's clearly not now that that's all I can Say I don't know when y. But it's not now, it's not this year.
- [97:46]
A
Yep.
- [97:47]
H
Doesn't look like it's next year. And then all of these projects that OpenAI is, is, you know, signing with, with Broadcom and Nvidia and even amd, they don't even start to ship until the end of 2026. So at least from a spending standpoint, from what we consider it's probably not 2027 either. Now we'll see what the stocks do. Like they tend to be anticipatory but, but in terms of like an air pocket or something and spending, I'm not really all that worried yet. When. I don't know. But it's not now.
- [98:13]
B
Yeah. How much have you subscribed to this idea of rolling bubbles? So right now it seems like we've had explosion, a lot of excitement around Neo Clouds this year. Pretty much all of them have sold off a ton in the last month. Maybe partly because of some of the comments out of the OpenAI camp and lack of confidence. But at the same time Seagate, Western Digital, these other companies are up tremendously. Yeah.
- [98:43]
H
The storage, they're covered by a colleague of mine but I mean they just go up 10% every day. Right. But that's the thing. It all really comes down to demand. Demand is off the charts. Nobody can get enough compute. The neoclouds are all. Even with core weave there was. They had a bit of a delay. They just don't have the capacity. Right. It pushed out a little bit. It's still there again the storage guys are ripping because the memory prices are going through the roof because there's so much. There's a lack of supply relative to demand. All we've seen from the hyperscalers is capex numbers going up and up and up and up. Like nobody can get enough compute right now. That's where we are. And I think that is so the overarching. That's the overarching thesis either way. You go back to the question about GP depreciation, GPU lifetimes for example. This is what Burry was getting at. He was saying, oh well, you know the. They're all using six year depreciation lifetimes and you know, these things don't last more than three years because you've got new stuff coming. It's not true right now.
- [99:44]
I
Right.
- [99:44]
H
I mean you can. Look, they're still renting out old GPUs for, I mean for much more than it costs them to operate them. It's clearly possible to run them longer than three years. It comes down to demand. Right now, demand is so strong, it is absolutely economic to run that stuff. If demand weakens, maybe it won't be. But if demand weakens, we're all screwed anyway. So to me, all of the bear cases that you come up with right now, to me, collapse under the same thing. Is demand there or is it not right now? It's there and it's not showing any. The demand side is not showing any signs of weakening.
- [100:18]
A
What about the leaked phone call from Sarah Fryer, the CFO of OpenAI, where there was was this idea that potentially there was weakness.
- [100:30]
B
What she said was user minutes were dropping, which implies in the core ChatGPT.
- [100:36]
A
App, which feels like the leading indicator for all.
- [100:39]
H
Yeah, but it's more than just OpenAI too.
- [100:41]
B
Right.
- [100:41]
H
So, you know, you could argue that we want OpenAI to be there because like Altman's driving a lot of this incremental.
- [100:47]
A
Yeah.
- [100:47]
H
So we'll see. But I mean, there's lots of demand. It's not just OpenAI. It's OpenAI. It's in throw. Totally. It's Gemini. Totally in general right now. I mean, usage is going up.
- [100:57]
A
Yeah, yeah, yeah. I mean, I'm sure we'll get more information on the. On the Google side.
- [101:03]
H
When you see these leaks, like, you always have to be careful about.
- [101:06]
A
Yeah. It was very odd that it was like from an investor only call. Like what investor would leak bad news?
- [101:11]
H
And by the way, we do calls like that and stuff gets taken out of context. I don't know what they said. I wasn't on that call.
- [101:17]
A
Yeah, yeah.
- [101:18]
H
I'm always a little hesitant to take. And I do this for a living.
- [101:21]
J
Right.
- [101:21]
H
I'm always a little hesitant to take a close leaf at face value. You have to.
- [101:25]
A
Yeah.
- [101:25]
H
You have to diligence.
- [101:26]
A
Yeah, yeah. I mean, at the same time, like to your point of the like S curve nature of these adoptions, it's possible that, you know, 800 million is a lot of people. You do at some point saturate everyone and.
- [101:36]
H
Sure.
- [101:37]
A
You only. There's only so much time in the day.
- [101:38]
H
There's 8 billion people in the world.
- [101:40]
A
Yeah, yeah. But I mean, it took. It took Facebook years to get up into the high.
- [101:44]
H
And then is that is the thing, by the way, I've been doing this job almost 18 years.
- [101:49]
A
Yeah.
- [101:50]
H
Same seat. I've never seen anything like this before.
- [101:53]
A
Yeah, right.
- [101:54]
H
I mean this is. This is unprecedented. And that's why people get nervous.
- [101:57]
A
Yeah.
- [101:58]
H
Because the numbers have gotten so big so quickly you just sit there and stare at them. It's like this can't be sustainable. We've been hearing it for two and a half years.
- [102:07]
A
Yeah.
- [102:08]
H
Still going.
- [102:09]
A
Yeah. So I mean, in this idea of the K shaped recovery or this cycle of little bubbles popping up and popping, how do you process something like core weave? It feels like everything is going so well and to the point where the negativity around AI is like a rumored leak of a phone call, but then we're seeing a company trade down by 50% in a month.
- [102:34]
H
Yeah, well, again, I won't talk core weave specifically, but I mean, look, anything that goes out where valuations are, are high, expectations are high, you know, it's not just that lots of things have weakened off of peaks. I mean, it's fine. You know, the NEO clouds in general, I mean they're all seeing tons of demand. Right. I'm not really worried in general right now about where the demand is going. The only thing we're seeing in terms of spending intentions and everything else is right now is, and again I think that is the question, how long does it last? I don't think anybody knows. If you just look at, like I said, at least what is currently being forecasted by the companies that are doing the spending at this point. There's no signs of a slowdown. Not yet.
- [103:18]
A
What do you think of this thesis that some of the hyperscalers are maybe offloading risk to the NEO cloud?
- [103:23]
H
Oh, they clearly are.
- [103:24]
B
Right.
- [103:25]
H
Yeah. Which is fine. That's part of the purpose, I think that the NEO cloud serve. You're kind of at the tip of the spear, right? It's boom or bust, right? Yeah. I mean, but that's part of their business model, I think.
- [103:37]
A
What if you go further on the tip of the spear is a company like organization like Blue Owl or some of the financing, on the financing side, Because Blue Owl's another example where Google's doing very well, Microsoft's doing very well, the hyperscalers are doing very well, even OpenAI is doing very well, growing a ton. But then you have some froth in the NEO clouds and some more froth in the private credit markets. What's your take on what's going on?
- [104:05]
H
Yeah, I mean those are probably frothier parts anyways. But I mean the financing question is interesting because I'd say some of these debt deals.
- [104:13]
A
Yeah.
- [104:14]
H
Again, if you're looking at, you know, parallels to prior bubbles, that is one thing, that there's a few things that people worry about. Raising a Lot of debt to do to fund this stuff. And then they also worry about what they've called kind of circular revenues. Right. There's a lot of like crosstalk. Yeah we do a lot of the companies around here and maybe to address both of those, I'd say on the debt side most of this capex is still being funded off of income statements. That is One difference now versus say 2001 is the companies that are driving the spending by and large are the largest, most profitable, best businesses that humanity has ever devised.
- [104:49]
A
I'm sure you remember what the actually largest business was in 2000. It was Exxon Mobil and Chevron. It was big, big oil and big oil was not driving and they were not funding the telecom build out.
- [105:03]
H
I mean there were a lot of cash flows. A lot of the companies they were, I mean they were raising money, they weren't profitable. That does not run.
- [105:07]
A
It was a very different thing. It wasn't like an immediate beneficiary, that was just the. Oh, the biggest companies are just getting bigger. So it is a very.
- [105:14]
H
And so in this case we are starting to see some debt deals to fund this but I'd say the by and large the vast majority, majority still being funded off of operating cash. Yeah, I don't feel too bad.
- [105:23]
A
Out of the 2.4 trillion that the Wall Street Journal estimated the number that was funded by cash flows was 1.4 trillion and then there was 800 billion in in private credit.
- [105:32]
H
Yeah.
- [105:32]
A
So it's still a pretty reasonable debt to equity ratio in my opinion if you think about that way or debt.
- [105:37]
H
To cash flow for now.
- [105:38]
A
Yeah.
- [105:39]
H
So that's right. And then in terms of like this called the circularity. So I mean Nvidia is behind us, there's investing a lot. They have deals in OpenAI but I mean Jensen's got his fingers pretty much in every. He's in every startup.
- [105:50]
A
Right. He's been doing it for a long time too.
- [105:52]
H
But think about like what else can he do with the cash? And so I'm hard pressed to think of a better usage. If you believe, we'll see where the numbers go. But if you believe the numbers they're going to be generating hundreds and hundreds and hundreds of billions of dollars of free cash flow over the next like five years. So what can he do with it? He can't do big M and A.
- [106:09]
A
Nobody will let him.
- [106:10]
H
Nothing's going to get through any trust. They have a buyback and a dividend but relative to their market cap it's going to Be de minimis. There's no choice. And so is there anything better than you do except invest and help to grow The AI Also have a lot.
- [106:22]
A
Of different business in terms of M and A. Like with Microsoft, they buy LinkedIn. That makes a lot of sense in the Microsoft ecosystem. You know, teams, they grow Azure. Like there's all these different places where they can plug other businesses in. Like If Nvidia bought LinkedIn, we'd just put be like what is.
- [106:39]
H
But he tried to buy other stuff, right? Yeah, he tried to buy arm.
- [106:41]
A
Certainly, certainly other pieces of this.
- [106:43]
H
Could you imagine if he'd been able to buy arm, by the way? He'd be unstoppable.
- [106:47]
A
He'd be unstoppable.
- [106:48]
H
Nobody's going to let him do anything.
- [106:49]
A
No one's going to let him. Yeah.
- [106:50]
B
How, how important do you think the Chinese market is to Nvidia? Because we, we debate this quite a.
- [106:57]
H
Lot and it isn't in the near. At least it's not important to the numbers right now because it's out of the numbers.
- [107:02]
B
Yeah.
- [107:02]
I
So.
- [107:02]
H
And Nvidia with even amd, they took it out. So that was smart because. Because it's still questionable whether or not they will be allowed to sell. And so from a numbers standpoint, it's okay. From a strategic standpoint, I think it is important.
- [107:15]
B
And Jensen hasn't hidden this second largest computing market.
- [107:18]
H
It's more than that though. So it is the second largest. And again, he's talking about $50 billion of lost opportunity. And over the long term it's probably bigger than that.
- [107:26]
E
Right.
- [107:26]
H
I mean China's big, but I think it's more strategic. You have to remember the Chinese developers want to use Nvidia. They have better products. Right? They do. However, you're not going to stop China. So China has, has like companies like, like Huawei, for example, that's basically state owned enterprise and they already have parts in China that have higher performance than what Nvidia is allowed to sell there. They do, they burn a lot more power. Chinese don't care.
- [107:51]
B
They just throw that sufficient.
- [107:52]
H
Right. But, but the thing is they're much harder to use. They don't use Nvidia's ecosystem. It's called Cuda. And the Chinese developer developers want to use Nvidia's ecosystem. If you don't allow him to sell there, what you do is you potentially encourage those local developers of whom there are a lot, to coalesce potentially around a local alternative, like a Huawei, for example, and start to build up potentially over time a more robust ecosystem in China and then you're shut out. And then the longer term word would be is once it's robust in China, does it move out of China now? Do you have a more robust global competitor? And so that's why I think strategic is where. And he said almost exactly that. He hasn't tried to sugarcoat it. Where I think he gets a little bit of a benefit is at least with Huawei, the parts, because of some of the other US sanctions, they have to make their chips at local companies like SMIC on deficient process technology. So the chips don't work as well, they're not as power efficient. I do not think those chips will really be competitive outside of China where they will be competing on a global basis with much more better products from Nvidia or AMD or whoever. So I think that helps. Yeah, but ideally you and you know, Lutnick said this. What did he say? He said we want to get them addicted to our technology. And I wouldn't have said it that way, but he's out of line. But he's right, we'd like him using it and to have some control over it. And we're letting that slip away. And so yeah, and I don't think it's great that he's not able to sell strategically, at least from a number standpoint. It's out right now.
- [109:18]
A
I want you to react to this quote from the CEO of kkr. He says, and candidly, when we read some of these headlines, it's clear that many of us have PTSD from the financial crisis and are looking for what will trigger the next one. Like where is the next boogeyman? From our standpoint, this market and economy really don't provide a simple narrative like that. What do you think?
- [109:44]
H
I think that's true. And to be fair, I started this job In April of 2008, about three weeks after bear Stearns failed. That's when I made my move to Wall Street.
- [109:52]
A
Yeah.
- [109:52]
H
And so I was forged in that fire. And I have the financial crisis like tattooed on the inside of my eyelids. I lived through it. It was a remarkable time by the way, to live through Wall Street. This is, this is not like that. Yeah, that was. People thought the world was coming, but we'll see like if there is an air pocket, maybe it will be become like that. But people really thought the world was coming to an end back then. But are people looking for the. Yeah, maybe.
- [110:15]
A
Right.
- [110:15]
H
There's people always looking for pattern recognition.
- [110:18]
A
One thing that I go Back to on the pattern recognition side is just I feel like for it to feel like there is a world where there's some massive correction and like a lot of the top players see big haircuts. But it's just hard to imagine a big widespread. Like with the dot com boom, every random person on the street was trading dot com stocks. With the NFT boom, everyone had bitcoin. They were telling you, oh you gotta buy Cardano, you gotta buy this, you gotta buy that, you gotta buy this nft. And then in the housing bubble everyone was like, yeah, I just got a second house. Morton didn't zero down mortgage, they didn't check my income. And so there was a lot of like places where just hundreds of millions of Americans could participate in the bubble on the way out, up and on the way down and here it just feels harder.
- [111:07]
H
There's some, I mean there's a lot more retail participation I think.
- [111:10]
A
You think so?
- [111:10]
H
Yeah. And Robinhood and all this, they get a little nasty sometimes. They trade these zero day to expiration options. So there's some of that. I don't think it's maybe it's as widespread as what we've seen in the past, but some of that.
- [111:24]
A
Yeah. I just wonder if like if you go to the median American, do they have significant exposure to the data center build out right now?
- [111:34]
H
No. I mean maybe they're electric bills.
- [111:36]
A
Yeah, yeah, yeah, yeah. So maybe that's it. But if there's a collapse that's going to go down, that's going to be.
- [111:41]
H
Cheaper, there's two ways that it could quote unquote collapse.
- [111:44]
A
Right.
- [111:45]
H
And they have different implications. So one is just, you know, there's a digestion cycle, there's an air pocket. And you look at the hyperscalers, how they spend money even before AI, they would tend to build and digest and build and digest. It happens.
- [111:55]
E
Yeah.
- [111:55]
H
And I always say like what's the chance of a digestion cycle? It's, it'll happen.
- [111:59]
A
Sure. I like this digestion cycle, but that wouldn't be structural.
- [112:02]
B
Sounds better than a correction.
- [112:04]
H
Digestion.
- [112:04]
A
Yeah.
- [112:05]
H
It would be good for the circle you get hold it through that.
- [112:08]
A
Totally.
- [112:08]
H
Like the way it quote unquote collapses. And it gets back to some of your earlier questions on, on the return.
- [112:13]
A
Yeah.
- [112:14]
H
If it turns out we're spending all this money, there's no return, then the.
- [112:16]
A
Whole thing comes crumbling down totally for everybody. Yeah, right. Yeah. But it's like, it's like I just really struggle to imagine a world where it's like, oh yeah, like it was so bad that like Apple's at a 10 pe, Google is no longer valuable. It's like these companies have been valuable for decades. Like where are they going? They can go down a little bit, but like they're just not that exposed at this point.
- [112:40]
B
Anyway, someone in the chat asked, please ask Stacy about CDS spreads affecting debt financing costs. So that's Oracle and. Or we've ie. How does the domino start?
- [112:51]
H
Yeah, it's a little out of my wheelhouse, the cds, but people are looking at a few specific areas. Like Oracle.
- [112:57]
B
Yeah, and Oracle Core weave too.
- [112:58]
H
I mean Oracle's not exactly a hyperscale, but they're trying to be a hyperscaler sort of. And they don't have the balance sheet and the income statement to fund it. So they are going more to the debt market and core weave. I mean, clearly, you know, they have to fund with debt. Most of the other ones, you look at it, look at a Google or a, or an Amazon. Like you're not in that kind of stage. Right. They're still, they're raising a little bit of debt, but they're still primarily funding this out of, out of free cash flow, basically. Out of cash flow from operations.
- [113:23]
B
Yeah, but it is.
- [113:25]
H
People, people are starting to look at it like for sure, for sure.
- [113:28]
B
Yeah. Have you, have you tracked situational awareness? The hedge fund, Leopold Ochsenbrenner?
- [113:33]
H
I have not. What is that?
- [113:36]
B
Former OpenAI researcher.
- [113:39]
H
Oh, this is that kid that, that started the.
- [113:41]
B
Yeah.
- [113:41]
A
He's got 4 billion under management.
- [113:42]
E
Yeah, yeah, that's right.
- [113:44]
B
And I was curious. So, so his strategy, like he, he raised the fund like at the beginning of this year or closed it.
- [113:51]
A
Closed it. He was raising it.
- [113:52]
B
He'd been raising it for a while but, but started deploying aggressively this year and it's. And has performed really well.
- [113:59]
A
At least it's the best performing hedge fund in the world.
- [114:02]
H
He's in the right place at the. Right.
- [114:04]
A
For sure.
- [114:04]
B
Yeah.
- [114:05]
A
The whole thesis was just AI is real.
- [114:08]
B
Do you think some more traditional hedge funds have sort of just overthought the AI trade over the last year?
- [114:14]
H
I don't know about you, because it's.
- [114:15]
B
Not like he, it's not like he was buying like somewhat, somewhat.
- [114:19]
A
He had a somewhat unique take.
- [114:22]
B
He had a very informed take, but it wasn't.
- [114:24]
A
But he also didn't go Nvidia, he went intel. He went other places on the map.
- [114:29]
H
So by the way, Intel's gone up not because of a. Intel's gone up because you know, Donald Trump wants the stock to go, which is a bull case. It's fine. You know, it's not an AI story. My view in general. And yes, I don't know if specific hedge funds have been, you know, overthinking or not. I haven't wanted to overthink it. My general call this year has mostly been own the high quality AI names, ignore most of the rest. That's been fine.
- [114:51]
A
Right.
- [114:51]
H
It hasn't had to be complicated. Number go up.
- [114:53]
B
Right?
- [114:56]
H
It hasn't had to be. It hasn't had to be complicated as well.
- [114:59]
B
Number, number go up.
- [115:00]
A
Well, I mean, what do you, what do you put in that high quality name bucket? Do you put the iPhone?
- [115:08]
H
We've covered like Nvidia and Broadcom, for example.
- [115:10]
A
Just mostly those.
- [115:11]
H
Yeah. And you know, we've been more lukewarm on AMD and that's what I've missed. Right. Because you know, it's also like ripped. You haven't necessarily had to be quote unquote, high quality to work. Right. Because, because right now we've been again, if you think about an S curve, if we're on that exponential growth part, it takes everybody up.
- [115:30]
A
It's been fine.
- [115:31]
H
But again, you haven't had to overcomplicate anything. Not yet.
- [115:34]
E
Yeah.
- [115:35]
A
What about, I mean, talking about not overcomplicating it. Like if you just bought Google and Microsoft, that's a pretty broad index on AI between DeepMind, OpenAI.
- [115:44]
H
But you gotta remember like it wasn't that long ago that people were looking at Google as an AI loser.
- [115:47]
A
Totally.
- [115:48]
G
Yeah.
- [115:48]
A
It changed. It completely changed. Completely changed. They position themselves very, very well. Citrini has another post here on the 2022. I saw a lot of very smart people in 2022 fail to recognize the reality of reflexivity. That is stupid headlines that with a few hours of research could reasonably be dismissed as nothing of consequence would add fuel to the fire and result in further downside. Understanding that this dynamic works to the upside as well. Well, as the downside means that. Yes, even if you know that the CDS on Oracle and Core Weave are blowing out because the CDS market is easily pushed around by a few parties trying to get cute and hedge their exposure to AI lending. You also recognize that if enough people view that CDS widening as indicative of a problem, it will become a problem. What do you think about these stupid headlines?
- [116:36]
H
I mean, headlines have certainly been more of the bane of my existence probably over last couple years. Yeah. Yes. So there have been lots of movements from headlines that if you, if you were had any depth of subject matter, expertise, you'd know the headline itself didn't mean anything.
- [116:53]
A
Yeah. What do you, what do you fall back on? Do you go to earnings reports or.
- [116:57]
B
Yeah, I'll give you an example.
- [116:58]
H
Stuff tends to correct itself over time. I mean, you get pops and I mean, I'll give you an example from my own coverage. And this is stock that I've, I Qualcomm announced. This was, this is their headline. It wasn't like some stupid headline, but they announced like an AI like server and there was. Wasn't a whole lot of information. The pressure. The Stock went up 20% the moment of the. And then. And then it kind of gave it back.
- [117:20]
B
Yeah.
- [117:20]
H
As it probably. But it had. You had a pretty big pop on the headline just because it was AI. Whereas if, if you know, you kind of know. And by the. Don't get me wrong, I mean there, there's, you know, we'll see what happens with that, with that problem product with them. And you can argue it's option value and everything, but to go up 20% on one day was probably overdone and it gave it back.
- [117:39]
A
It's fine. Yeah, makes sense. Jordy.
- [117:42]
B
Giving it back, I feel like has been the story of Oracle. They got this tremendous pop off the OpenAI deal. Now they've retraced to below.
- [117:50]
H
Is it lower than it was.
- [117:51]
B
It's lower than it was before, which to me says either the market doesn't believe that it's real and. Or the other reading would be it's real, but it's not going to be ROI positive. Right.
- [118:06]
H
Yeah.
- [118:07]
B
At least for Oracle.
- [118:07]
H
Without making any comments on Oracle, I mean, those are the, for any of these investments, those are the worries that you're going to have, right?
- [118:13]
A
Yeah.
- [118:14]
H
Either because again, especially if it's OpenAI, so Altman's committed to a tremendous amount of capacity. I mean it's just between, on the chip side, between Nvidia, Broadcom and AMD, it's 26 gigawatts. And just for some context, I think the whole global data center installed electrical capacity, I want to say something like 70 gigawatts. Something in that ballpark.
- [118:34]
B
Right.
- [118:35]
H
So it's a lot. I mean, he was even making comments he wanted to deliver, what was it, a gigawatt a week at one point. I mean, it's a lot. Right. And so you already are wondering can he actually deliver on that or not? And those are valid questions. And by the way, I wouldn't count Altman out like he's aggressive, but he's got large aspirations. But then with someone like an oracle who doesn't have the balance sheet or the income statement that some of the others do, and you look at the size of the. I can't even remember what it was. $400 billion or something. It was a massive step up. 300 in one quarter. You wonder where it's coming from. It's always around, like, can they deliver it and can they pay for it? And those are the general worries, I think for anybody that's announcing something.
- [119:19]
A
How do analysts try and dig into there's quality of earnings, quality of revenue, quality of backlog. Feels like an important thing to get into. But I don't know that the contracts are available.
- [119:32]
H
They're not. I mean, you talk to whoever you can and you've been doing this while you look for pattern recognition. But I, I mean, look, nothing in semiconductors is ever really baked.
- [119:43]
A
Sure.
- [119:44]
H
Right. I mean, in fact, we just discovered like during COVID a lot of these companies were experimenting with quote unquote, non cancelable orders.
- [119:50]
A
Sure.
- [119:51]
H
And we saw what happens. It's like, yes, maybe I could contractually force my customer to take a year's worth of parts they don't need.
- [119:56]
A
Yeah.
- [119:57]
H
Am I really going to do that? Historically, the answer was no. During COVID for some companies, the answer was yes. And those particular companies are still paying for because, okay, great, I just took all my parts. I don't need to order anything from you for a year.
- [120:10]
A
Right, yeah, that makes a lot of sense.
- [120:12]
H
And so I'm always a little hesitant of backlog and commits and even some of these. By the way, Even with these OpenAI deals, they put out big numbers, but it's not like the whole thing is committed, like maybe the first gigawatt at this point.
- [120:23]
A
Totally, totally.
- [120:24]
H
Which is how it should be.
- [120:25]
A
Yeah, yeah, yeah. No, that makes a lot of sense. It's great.
- [120:28]
B
Are you AGI pilled?
- [120:30]
H
No, probably not. I'm not exactly sure what AGI means yet. You know, there's a lot of different definitions and sometimes it tends to change.
- [120:38]
A
Well, do you find AI useful in your day?
- [120:41]
H
No, I do. I have to be a little careful because from a regulatory and compliance standpoint, it's not like I can just start throwing stuff into very limited. We have some internal stuff that we.
- [120:51]
A
Can use, which is enterprise plan business.
- [120:53]
H
Yeah. And it's ring fenced and everything. But I mean, simple stuff.
- [120:57]
E
Right.
- [120:58]
H
I mean, what is this notebook lm? I can literally take. I could take a Utah of this, what is it? Three hour podcast and toss it in there and it'll give me a summary. Not just a summary.
- [121:07]
B
Then we can take that summary, put it into another model and say, make me a three hour.
- [121:10]
A
Sure, yeah.
- [121:11]
H
Pretty soon it's just AIs making and watching podcasts, Right? It's a slurry of podcasts, but that's just one example. So there were clearly ways where it's influenced my experience and my workload and I'm probably on the low end, frankly, of what I could be using. I am constrained on what I can do with this. I'd love to use it more if I could.
- [121:32]
A
That makes a lot of sense. Well, thank you so much for coming by Studio.
- [121:36]
H
This is a lot of fun.
- [121:38]
A
Have a great rest of your day. We'll talk to you soon.
- [121:40]
B
Cheers.
- [121:40]
A
While he's hopping off, 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. Speaking of turning things into podcasts, Kristof has some fake news here on the timeline. I love Kristoff. Coastal futurist Christoph says Still crazy to me that Steve Jobs slash Apple invented the word podcast and that it's a mix of ipod and broadcast. And apparently this isn't true. Apparently a BBC journalist, Ben Hammersley, coined it in a 2004 Guardian article, but then it got ported back to Apple and the term podcast was adopted. But it is funny that podcast it.
- [122:28]
B
Did come from ipod. It's always felt like a very boomer term. Yeah, and it makes sense.
- [122:34]
A
Radio show.
- [122:35]
B
I never knew that it was a combination of.
- [122:37]
A
Oh, you didn't know it was iPod plus broadcast. Yeah, yeah. What else? The cast casting is very, very popular.
- [122:46]
B
Anyway, what's this article from Evan Armstrong.
- [122:50]
A
Over at the Leverage published an article on who actually makes money when robots work. Looking into some of the venture funding that's flowing into humanoid robotics and he is trying to create a field guide for separating the real companies from the grifters. Talks about 1x the home robot costs $20,000 upfront or 499amonth. The website looks like every other VC funded DTC brand from 2015. Millennial beige, sans serif typography, the works buried in the fine print and tech coverage is the actual product. For most chores, a human in a call center will drive the robot around your house via teleoperation while the system records training data for autonomy that doesn't exist yet. And so there's a big question about how like what will the actual margins be and he was trying to dig into that that so you can go read Evan's piece over on the leverage if you want get a seven day trial.
- [123:53]
B
Who actually makes money though when they work? Is it the call center operators?
- [123:58]
A
I believe. I mean I would not bet against Elon on this. I believe that the hardware manufacturer will ultimately be the one that makes money in the long term. I think that everything else is more commodity in the stack but there is going to be be a compounding advantage to actually having the manufacturing capability to build the robots at scale. Now this is years away but I'm certainly AGI pilled in the sense that the tele operation will become less and less important and the Tesla model of having a economically producible product will be very, very important. So that would be my take.
- [124:42]
B
In other humanoid news, Brett Adcock was putting a company in the truth zone. Shenzhen based company ubtech claims to have completed the world's first mass delivery of humanoid robots. And the Shenzhen based company has secured apparently over 112 million in Walker S2 orders this year. Brett Adcock says look at the reflections on this bottle and then compare them to the ones behind it. The bot in front is real. Everything behind it is fake. If you see a head unit reflecting a bunch of ceiling lights, that's a give. That's a giveaway. It's cgi. So he's putting, he's putting ubtech in the truth zone and then Christopher comes over the top and puts him in the. And then says such an embarrassing post for a CEO at this level to make make. So people don't necessarily believe that the CGI claimed.
- [125:41]
A
Did yubytech actually deliver it? There's no community note on the original video for something like this. I mean we've seen iRobot like you can actually just do flawless CGI that is indistinguishable. There's no way to tell. So I would, I would be relying on some sort of, you know, on the ground reporting. Do you have an idea Tyler, of whether or not this is real? We've seen demos of this robot before where it pulls the battery out of the back.
- [126:11]
C
Yeah, I don't know for sure if it's real, but it is a public company.
- [126:15]
B
Yeah, down 2%. Brett kind of needs this company to be totally fake because ubtech is a public company. It's valued at around seven and a half billion dollars and they have real revenue.
- [126:32]
A
Okay. Ubitech, man. The human market over in China is extremely competitive. You have UVtech and then you have. What's the other company? I keep forgetting? Unitree. Unitree. They're two different companies. Right. But they're both publicly traded, both have revenue, both make humanoids. I wonder if there's like, like I wonder what the CEOs of these two companies would actually say. Like, what is the differentiation between the two. The two products, because they look pretty identical. Maybe we need to do a side by side, but I don't know what else here. Martin Shkreli said, best part of ubtech is it's public with real revenue. So its valuation is naturally far below pre revenue robotics.
- [127:18]
B
Naturally.
- [127:19]
A
That's wild. Yeah, yeah. No, I do think it's real, but I don't know, like, it's really hard to prove it one way or another. It's hard to prove any of these videos that come out with the humanoids because humanoids are just like, it's the textbook like CGI product. Like you can just, if you're at all good at cgi, you can make a humanoid robot look great because it's all just polished steel, basically perfect reflection. It's much harder to make a human face look CGI real. And so we've had the ability to make like cgi stormtroopers, CGI C3PO. That's been at perfect indistinguishable level for years. But getting to the actual human level.
- [128:04]
B
Yeah. To be honest, I care a lot less about did they use cgi to put 100 of these robots, robots in one room versus like what are the actual capabilities of the individual robot?
- [128:16]
A
Yeah, right. Yeah.
- [128:18]
B
So that's why I think 1x, I think they're pretty straightforward about, about it being tele operated or that being kind of the value. As you're buying a robot that can be tele operated in your home, it.
- [128:29]
A
Really feels like the question is like the United States, like why is.
- [128:32]
B
Why is Adcock like obsessing over a competitor? And then what, what he. What can he do in the next three to six months to justify his $40 billion valuation?
- [128:44]
A
It probably needs to ship a bunch of these things just for any reason. Just manufacture a lot of them, I would imagine, because with Unitree and this other company and 1x is starting to ship and Elon's clearly getting serious, the race is on to actually start, start manufacturing and shipping them. And it'll be interesting with the flying car thing. A lot of it's like regulatory, so it's harder to assess what the progress is like because you can always just kind of get hung up in regulatory. I wonder in the humanoid space. Well, speaking of flying cars, what's up, Tyler?
- [129:23]
C
I was just gonna say I think it's interesting we still haven't really seen any humanoid CEOs like talking about sanctions against China.
- [129:32]
B
Yeah, I know that's the thing to call for because it's like a very easy argument.
- [129:35]
C
I would not fake. I mean, you can do that, but you should just say like, even if they are real, like we want these made in the US like in the way that you see kind of Dario saying in AI and it's just kind of weird that you don't see like Elon saying that we need to make these in the US or the 1x.
- [129:51]
B
Guy is saying, yeah, bring out the ban hammer, the banhammer.
- [129:55]
A
Let's.
- [129:56]
B
Yeah, I just. That's going to be an easy one. One for. All you have to do is like, do you want a million robots in American homes that could have the sci fi scenario where you have a backdoor?
- [130:11]
A
And the argument against DJI was always like, well, if some tiny drone is in your bottom sock floor, what's it gonna do? It's not just gonna bust out of there, but a humanoid robot will just bust out of the whatever you put them in. Unless you store them in like you're gun safe or something. Walk in gun safe. Let me tell you about numeral.com. sales tax and autopilot. Spend less than five minutes per month on sales tax compliance. And then we have an update from Keller from Zipline. We're going to watch this two minute update from him.
- [130:43]
B
Pull it up.
- [130:44]
D
Hey everyone.
- [130:45]
F
It has been an insane two months, but I thought it'd be cool to give a two minute update. Whenever I post something on X people, people are always asking me, like, when is it coming to my Metro? Why aren't you scaling faster? We are definitely hearing you. We are scaling as fast as we possibly can. In fact, I am standing in our expansion space for the manufacturing facility as we speak. We're getting ready to build 20,000 autonomous aircraft a year, all here in South San Francisco in the United States. Mid December, we will actually start producing the first aircraft here in this space. So this past week I was in Dallas visiting a lot of our different customers, going and visiting a lot of the different stores that Zipline is delivering from. And my mind was basically blown. Right now we're growing the number of deliveries we do per day at around 15% week over week. And we've been growing that fast. For about 30 weeks straight. A lot of our customers out there are placing orders three to four times per week. In fact, some customers are ordering three times a day. People actually just fundamentally change their order. Some people are grocery shopping once every one to two weeks and then ordering from Zipline three to four times a week just to do fill ins. We've also been able to launch a new Walmart Supercenter every week across Dallas over the last couple months. And by the way, I've been talking about really exciting hyperscale in the US But a lot of people are often like, well, wait a minute, isn't that Zipline operating a huge logistics network in Africa? Yes. In fact, that network is growing faster than ever. Today we serve 5,000 hospitals and health facilities across Rwanda, Ghana, Nigeria, Cote d'. Ivoire, In Kenya, that has become the largest commercial autonomous system on earth. Together with what we do in the US right now, Zipline is doing an autonomous delivery about every 30 seconds. We're also adding a lot of new products. For example, in Ghana, we just added HPV vaccine to the overall network. And in the first week, we delivered 150,000 doses of vaccine.
- [132:28]
B
This is a crazy scale. Congrats. John's back.
- [132:33]
A
I'm back. He. He's going on the show next week. We love Keller. Congrats to all the folks over at.
- [132:39]
B
Zipline on, yeah, really wild building in South San Francisco, scaling deliveries like crazy. Yep. I think this is one of those things that's hard for it to be hard to be that excited about until you.
- [132:51]
A
Ryan, there's a spicy question in the chat. How long until they pivot to a weapons company? I don't think it will happen. I think that drone defense is like the most. It's such a, such a crowded industry. It would be really, really hard to break through there. I don't know, maybe there's something in logistics. I've talked to a few folks in defense that will do drone based transportation or delivery for, let's say, how do you get supplies from land onto an aircraft carrier that's stationed off of the coast? Well, you could fly a whole helicopter there. But what if you just want to deliver one smaller package back and forth? You could use a smaller drone for that. There's companies that are building in that category. So, I mean, I could see it happening, but I wouldn't expect Zipline to be, you know, on the front lines of Ukraine competing with Neros to, you know, arm the resistance there anytime soon. The drone delivery, like just delivering a burrito is going to be a mess. Massive market like Doordash does it that way it's a hundred billion dollar business. Like I don't think that they need to deal with all the craziness of DoD procurement.
- [134:06]
B
Anyway, so someone in the chat mentioned that Ubtech fired back and released a behind the scenes video. No.
- [134:14]
A
Okay, okay, we got to pull that up. Do we have it? Thank you to the chat for letting us know. Let's add it. In the meantime, more on robotics. Blake Robbins has a great take that I was curious about. If anyone knows anything about this, please fill me in. He says, as robotics continues to get more attention, I always wonder what Boston Dynamics is cooking. It feels like Boston Dynamics should have been the OpenAI of robotics. And yet I can barely mention anyone that worked there. And there's a whole bunch of different debate on the, you know what happened. Boston Dynamics was sold a few times to be studied. Yeah, Boston Dynamics.
- [134:55]
B
What's the dynamic over there?
- [134:56]
A
Wait, wait, Boston Dynamics, it's the original browser company of New York. They just took their city and what they do and put it together. This is basically the Dynamics company of Boston Rebrand come out with a humanoid. Let's play this video. This behind the scenes video.
- [135:11]
B
Los Angeles Dynamics would just be like an influence, like Lincoln bios.
- [135:16]
A
At one point Boston Dynamics sold the Hyundai Heavy Industries. So I believe they were across the Pacific Ocean. Let's react to the second video posted by the humanoid robotics company. Okay. Ubitech industrial walkers, they've been pumping these out. It feels like the question has never been, can China make a lot of of these things? Like obviously they can.
- [135:41]
B
In the bio they said, they said it looked too perfect to be real.
- [135:45]
E
Yeah.
- [135:45]
B
And then they used some ChatGPT slot. But perfection isn't fabricated. It's delicately.
- [135:50]
A
No.
- [135:53]
B
It'S so good era, you know.
- [135:55]
A
You know, for a long time people were saying that there was no way to watermark AI created content with because like, oh, how would you do it? You change one. It's like, like actually all AI content is watermarked. It turns out it's just perfectly watermarked. Yeah, this looks real to me. I don't know. It also could be cgi.
- [136:15]
B
One of the comments is it's pretty funny to see Brett say this was all faked.
- [136:19]
A
Yeah, it looks real. I mean it would be way they could have done so much better to make this look real. Put a bunch of humans there, touch them. A lot of the hard part in CGI is the handoff between the CGI character and the Human character. And so what you should be doing is you should have a human who takes like a smoothie and pours it on top of the robot and then like wipes the robot off clean with a towel. And you're seeing how the fluids interact with the robot, interact with the person. And like that was. That was not actually that satisfactory. I don't know. I think, I don't know. I'm not calling it fake, but it could be fake because CGI is really, really good. Like CGI just is at that level where that's possible. Anyway.
- [137:09]
B
Yeah, we know that. We know the caption or the description of the video was AI generated.
- [137:14]
A
That is hilarious that they had to use AI generated for the caption. It's like AI on it and it's like, after all it is an AI company. Like it would be on brand anyway. Fin AI, the number one AI agent for customer service. Number one in performance benchmarks, number one in competitive bake offs, number one ranking on G2. You can get started for free. Our next guest is Luca from Bending Spoons in the Restream waiting room. Welcome to the show.
- [137:40]
B
Welcome to the show.
- [137:41]
A
Good to meet you. How you doing?
- [137:42]
E
Hello. Hi.
- [137:43]
A
Thank you so much. I imagine it's late there. Thank you for staying up late and, and coming and chatting with us. For those who don't know you, would you mind introducing yourself?
- [137:54]
E
Of course, one of the co founders, the CEO at Benningspoons. And what we do is we look for digital technology businesses with unexpressed potential and then we acquire them if they'll sell them to us and transform them, sometimes quite radically by rewriting big chunks of the software, re architecting the cloud infrastructure, redesigning the ui, launching lots of features, optimizing monetization and marketing, rebuilding big parts of the organization. So lots of hands on work. And then if we do it right, we generate a lot of value. We plug back into bigger acquisitions and strengthening our platform. So basically our proprietary technologies, our expertise.
- [138:35]
A
Access to talent, the company's huge now $11 billion valuation. How did you get started with all this?
- [138:45]
E
So I postponed a startup in 2010. Interestingly talking about AI, we were trying to create a self rising diary or Journal with AI in 2010, which was pretty early. I mean it worked fine, but it wasn't good enough. So that was like your typical startup, you work from a garage. It was really our living room in our apartment with more or less the concept of the same. We worked on it for three years, couldn't make it work commercially, and then we Kind of shut it down. And through that experience, we came up with the strategy for Benning Spoons. Basically, why don't we try to outsource looking for product market fit to the market. We try to be the best in the world at the call it functional expertise that's necessary to run a digital technology business. So software engineering, product design, growth and all these things. And then we buy businesses where the owner basically doesn't want to work on it any longer or where maybe we can do better so we can offer an exciting price for all involved.
- [139:49]
A
Is the name a reference to the Matrix?
- [139:53]
E
Actually, yes. Yeah, it is.
- [139:54]
A
Cool. Yeah, it feels like, at least to me, it's kind of an odd name for any business. It's cool. I like the reference. But why did you pick that name in particular?
- [140:08]
E
So we knew we weren't going to work on just one product, so we couldn't call it, say, Facebook if you do.
- [140:16]
A
Yeah, you need like a name for a holding company on day one, sort of.
- [140:20]
E
Yeah, something like that. So we chose to find a name that would somehow convey a couple of principles or values that we thought were important to us. And benningspoons reminds us of two things. One is the power of the mind. For obvious reasons, if you are going to bend spoons with your. With your mind, it means you believe it's powerful and you can do great things with it. And the other one is, call it perseverance, hard work, dedication. In my imagination, to get to the point where you can bend spoons with your mind, you probably have to work pretty hard at it. And plus, it was kind of memorable. So we. We liked it.
- [140:56]
B
Yeah, I like it. I reached out, or one of us reached out when the AOL acquisition got announced. Give us a backstory on that deal, how it came to be, when you started thinking about the business, what value you saw in the business, and maybe why. It was a target that was overlooked by maybe Americans and private equity players and what you saw in it.
- [141:29]
E
There were multiple private equities that were looking at it too, so we weren't the only ones. It was owned by a private equity, or it's still owned because the acquisition was only signed. It hasn't closed yet. And so you could imagine how they would run a proper competitive process. The typically we follow a business for a long time before an opportunity to acquire it presents itself. I don't have precise statistics, but off the top of my head, I'd say it's quite rare that we end up acquiring a business we haven't Followed for at least a year, sometimes more. So we have been following AOL for a while and when it became available we made an offer that we thought was quite competitive and here we are. What do we see in it? So I think it's, it's actually a great business. People are a bit stuck with the idea of AOL from the 90s or maybe early 2000s of Internet connectivity. Doesn't do that. Haven't done that for a long time. Today it's two different products. There's a web portal for people to consume news and entertainment and an email client like Gmail, pretty much that sort of thing. They have well in excess of 30 million monthly users. Users, 8 million daily active users. So it's huge, give or take. 1 in 10Americans uses it and it's just not necessarily the new hot thing that you would read about online. But it's still massively used and very good retention because people are self selected if they still use it for really liking the brand and the offering. And nevertheless we think we could help make the product more modern, more effective. We think that through AI we could create better recommendations for the web experience, essentially better content to consume. We'll see. But it's exciting. We look forward to working with the team on it. Plenty to refine and expand. And it's exciting to work on a storied brand. I find it kind of cool actually to be able to get our hands dirty with a brand of the caliber of aol.
- [143:43]
B
Yeah, it truly is one of the great America online, one of the great names in business history. How do you think about. I feel like Silicon Valley has a sense that businesses are either in hyper growth or they're dying. Right. And there's nothing in between. And that's just because like the industry venture capital dollars are deployed into companies that are growing really quickly. And as soon as you stop growing like that sort of source of capital is turned off and you kind of fall out of the headlines. But how much like I assume the kind of part of the thesis for bending spoons is that a lot of these businesses are just like way more durable than the tech industry maybe gives them credit for. Because even after they're not getting headlines in TechCrunch, they can still generate a lot of cash flow for very long, long time. But how do you think about durability of digital businesses and when they can become sort of like Lindy and survive across decades versus when they evaporate?
- [144:52]
E
I would say I kind of agree. And I will say that if anything, a Business that has a lot of history, it's a lot easier to project its future and make accurate forecasts. Maybe you know, it's not 10x but you also have a ton of data, historical records for all the cohorts at scale to know roughly where it's going. So yes, maybe the upside is not transformative as it would be in a seed or VC investment when all the stars align, you know, maybe 10x or even 100 extra money, but, but you also know you're very unlikely to, to see it melt in your hands. Yeah, many of these businesses have excellent metrics, they're just not, not cool and hot, so to say. And so the entire capital markets, particularly on the private side, not so much the public side, but on the private side, is geared towards growth, very aggressive growth and there's merit in that. But it also means there's perhaps an opportunity to be less opinionated about growth at all costs. We, we know on our part we try to, we're basically quite mathematical about it. We make our projections have our own return thresholds and then we are happy to buy fast growing businesses. We have many times stagnant businesses, decline businesses as long as the math checks out. And this has been quite good for us not being too thesis limited by a very narrow thesis, but actually staying opportunistic.
- [146:21]
A
How are you structuring bending spoons? Do you have a deal team and an operational team that goes in and actually works runs the companies? Or is it more of a hybrid model where a partner who finds someone who finds the deal might be immersed in that company throughout the life cycle of owning the deal?
- [146:42]
E
Yeah, so essentially say that we're kind of 25% private equity, 75% a company. So we do have an M and A team. It's actually pretty small, probably eight, nine people. And they do what you would imagine a private equity firm would do. They create a pipeline, scout the deals to negotiate the deals. And then really probably 95% of the, of the team here are software engineers, product designers, growth managers, researchers. And once we close a transaction we add a, call it a task force of experts to get, you know, get in the trenches with, acquire team and study things in fine detail and help with those kind of radical transformations I was describing earlier. So that's the tech technology company part. We almost all we do is writing software, developing and refining technologies, user experiences and we buy to hold and operate forever. We're not a fund, to be clear, more like a Berkshire Hat where we buy off a balance sheet. If you've never sold a business. Don't do intent to.
- [147:42]
A
Is it fair to think? Yeah. You imagine AOL and Vimeo existing as companies in 50 years or brands as 50 years, or both. Like what is the critical business units? I will call it business units.
- [147:56]
E
Okay, so if a company is more a kind of a legal entity that may be dissolved in the future, maybe yes, maybe no, depending on considerations. But there will be a business unit with its dedicated management team and engineers and designers and. Sorry, I don't know why that's on our side.
- [148:17]
A
Zoomed in. We just zoomed you in a little bit for the viewers at home. Sorry.
- [148:22]
E
You can see my baby's cradle just in the background.
- [148:25]
A
Oh, cool.
- [148:25]
B
Nice.
- [148:26]
A
Congrats. We have five kids on our side. It's amazing.
- [148:32]
E
Yeah, mine is two and a half months old, so. Wow.
- [148:36]
A
Well, thank you so much for taking the time to come talk to us. We really appreciate it.
- [148:40]
E
My pleasure. Thank you for having me.
- [148:42]
A
I'd love to know like, what is it actually like buying a 5,000 person company? Because I imagine if you buy a company and you can actually go interview everyone, but with 5,000 people you're like layers and layers and layers deep in everything. How do you actually go and start right sizing. We've seen a little bit of what happens through the stories of Twitter, what Elon did there. Obviously that was a company that it seemed like post Elon was massively overstacked. But even if the staffing is correct, I mean Apollo's owned AOL for a while, so I imagine that it's not wildly overstaffed. How do you actually go in and just get your handle on everything that's happening within the business because it's such a huge entity.
- [149:36]
E
So we have never acquired a 5,000 person company. The largest is about 1,000 people. I think your point stance, I mean, but just for accuracy's sake. Well, I think it's very difficult, perhaps impossible to do if you are again a private equity and you have a small investment team. In our case, we will add a task force that's sized in relation to the size of the organization we're trying to study. And so if we acquire a 1000 person company, we'll probably have say 50 people. If it's a 100% company, maybe as few as 10. And then over the course of, of couple of months we will split the work. So everyone is in charge of understanding a piece of the organization that's not too big for like the right size for them and specific to their capabilities and expertise. So if there is a more technical part, maybe an engineer, if it's something has to do with design or marketing and so on and so forth. And then those people will be talking to each individual in that part of the organization multiple times. They will be looking at the code base, they'll be contributing on ongoing projects. So we take our time to learn. I mean, I was about to say everything there is to know. Of course, that's not entirely true. Like you don't learn everything in two months, but truly learn the vast majority of what matters and only then do we come up with, say, a new vision for the company, a new roadmap, the, A new design for the updated organization. It is extremely time consuming and only feasible you have if you have a large, call it corporate staff of experts, like in our case. Only, only then can you do it at scale. Otherwise, yeah, you couldn't. I agree.
- [151:17]
A
Yeah.
- [151:18]
B
What, how, how much is like, I'm assuming, when you're thinking evaluating a potential acquisition, you're thinking, how much can this company benefit from AI and what is kind of the AI disruption risk? Oftentimes it could be both. I can imagine with aol, you're saying better content recommendation that you could roll out maybe with less resources invested, but at the same time there's risk of new companies entering the market. But what's any sort of framework that you're using to evaluate what companies are going to do? Well, because nightmare scenario is you buy a great business today and that becomes less relevant. But there's also, I think, Silicon Valley also, maybe again, going back to my earlier point around durability, there's companies that have been disrupted multiple times that can still continue to produce cash flow. But I'm curious what your framework is.
- [152:19]
E
Yeah, so I think there's, I believe you can come up with plausible visions for the future. I mentioned earlier we, we were working with AI in 2010 where I didn't know a single other person who was interested in AI back then. So, I mean, I'm sure there were many in the world, but it was not mainstream. So, you know, we did have a big vision for AI in the long, in the long run. It turned out to be way too early, maybe a decade or more. So I'm a big believer in, yes, you want to have a vision for the future, but at the same time, if you think you know what's going to happen and especially over what timeline, you're very likely to be disappointed. So I would never want to bet our future on those sort of the assessments. What we try to do is evaluate whether a company is more or less likely to be disrupted and of course also enhanced through AI. And that, that's basically a qualitative assessment. We try to embed that in our acquisition thesis. But also we always need a contingency plan in case there is a very aggressive, rapid disruption. So we need to make sure that we have ways to get back our investment in case we end up in that or most of our investment if we end up in that scenario. So you, some businesses are structured in a way that if things don't go well, you may be able to trade long term health for short term returns and at least salvage your investment overall. I mean it's not going to be a stellar investment, but still the break even. So we try to have an escape route just in case things don't turn out the way we think it will. And of course the fact that we're so diversified, the biggest business we own contributes about 15% to our revenue and they are across many different segments. We are massively less exposed to this sort of disruption than any company running just one product, no matter how successful that one product may be today.
- [154:10]
A
Yeah.
- [154:13]
B
Any plans to create a Neo cloud or anything of the sort or you guys like staying at the product level?
- [154:22]
E
No such plans for now.
- [154:25]
A
What's the Italian early stage?
- [154:30]
B
I want to dig into that a bit more just out of curiosity because I'm assuming there's a lot of different players that would love to give you guys capital to help deploy cloud, given that you're running a lot of companies that are going to be buyers of, of compute and you have a track record of being able to deploy large amounts of capital. Is the decision to like the strategy is you're invested in a bunch of different companies but do you like them to be the same kinds of companies and that like again, is it just a mandate around digital products or would you one day branch out? Because again, if you use the Berkshire Hathaway comp, it's like you've seen them buy everything from Coca Cola to Google. Right. And operate a bunch of different businesses.
- [155:17]
E
Yeah, look, I think there's a trade off between, let's say there is a, you know, going back to Warren Buffett, a circle of competence as he calls it, where you, you have proven you're good or very good, better than most. And so on the one hand, if you stay within that circle of competence, you're more likely to do well but at the same time if you venture outside of it, you extend your Toolkit your capabilities and the TAM grows with it too. So there is a trade off between expanding that circle over time while seizing the opportunity at hand with what you already have proven, that works in our case, next year will probably be at about 2.5 billion in revenue and the.
- [156:02]
I
Overall.
- [156:05]
E
The overall digital technology market market is about $2 trillion. So it's not exactly we're saturating the opportunity here. So before we get into something we know very little about other than on a high level, I think we should make sure that we are saturating the core opportunity. But yeah, maybe in five or 10 years if we ever feel that we're getting too big for, for the tam and maybe we'll look beyond that. Who knows what will be appealing then.
- [156:34]
B
Do you think it's funny that the US venture capital ecosystem is somewhat off balance sheet R and D for you guys? It's like they can deploy a bunch of capital into these categories, create good products like Evernote and then you guys can come in and own them for the long term.
- [156:53]
E
Look, when it comes to acquisitions, we're opportunistic. Whatever the market offers, if you think we can deliver great returns, we'll be happy to buy it. Regardless of, regardless of the particular history.
- [157:05]
A
I don't know.
- [157:06]
E
I don't have other gratitude or any particular opinion on that. VCs that many have done really well. I guess it's normal that at least in some cases their investment didn't turn out to be the exceptional success that it could have been.
- [157:22]
A
Well, in many cases some of Those firms actually IPO'd and delivered a bunch of returns for the LPs in the venture fund I am interested in. Now you've acquired Evernote's, probably a good example. But any example of acquiring a company that's already publicly traded, how does that work? How do you work through that process? Are you going directly to the board? Are there best practices around when a company or a group of shareholders might be more receptive to a takeover? Or do you see yourself ever getting into more of like a hostile takeover scenario? Does that matter to your strategy?
- [158:12]
E
We've done two take privates so far. One is brightcove early this year and the second one is Vimeo. So I'm now I wouldn't consider myself a world expert in, but I've seen a couple of them so I can probably provide at least some, some input here. We have done it quite collaboratively reaching out to the board and, and just saying look, we'd be interested and we make an Offer the. I would say the negotiation is not massively different from what you would do in a private deal, but there is a lot more pressure on a board of a public company to, to act in the best interests of the broader shareholder base because the liability is a lot more. Legally speaking, it's probably not very different. But of course, the transparency of that process makes it really difficult to say, I don't like it, fuck you. If it's a good offer, you have to entertain it. On the bright side, if you have a great offer relative to the stock price, you have, I think, better certainty that it will be at least entertained than with private companies, where sometimes you find you're absolutely confident you're making an incredible offer, but say the founder is just not going to sell, or a particular investor maybe invested at a really high valuation and they're now maybe even delusional in thinking, oh, we'll get back to 3 billion if we wait for long enough. And so there you don't really have any leverage. It's like, okay, end of the conversation. With a public company, that can't really happen, or at least it's quite rare. So that's the path. Positive. On the negative, the process is a little bit more uncertain because with the private deal, typically if you have a handshake agreement with the two, three procedure makers, then it's quite unlikely that the deal falls through. The public company basically, at any point in time, if someone comes with a better offer and it can be quite public. So everything is back to square one. You need to shareholder vote. So you only know after multiple months. Months. And so you're kind of keeping your fingers crossed. But overall, I love public deals. I think they're in many ways more straightforward. There's no hiding behind illusions. I mean, your stock price speaks volumes. And if it's been at a certain level for a long time, that is it? For the most part, yeah.
- [160:32]
A
Is it? Also, do you have more confidence in audited financials or any other sort of process power that comes from actually being a public company? Do you feel like when you go into a company that's been taken private, you can just feel the difference? Oh, okay. This company has been operating like a public company and that has maybe pros and cons, but you can definitely tell. Or is it purely in the deal stage that you feel that there's a difference?
- [161:05]
E
We have seen private companies that are run really well in terms of FPA and we've seen public companies certainly have to reach a pretty high bar. So, yeah, I Guess you're more certain that on the one hand you're more certain that the public company will be better geared to provide you with the, say, the due diligence materials you need and whatnot. On the other hand, there's generally, and understandably so, and rightfully so, more risk aversion in a public company. Because if there is a leak that can truly, you know, create a pretty difficult situation for, for everybody involved with private companies, it's leaks are also less likely because the incentives to leak something are lower. There are all sorts of reverse incentives with a public company for obvious reasons because the stock is so liquid. So sometimes deals are leaked that if the company was private wouldn't be leaked. So I would say yes, the data tends to be a little bit more ready, cleaner but on the other hand the company tends to be more careful in bringing people under the tent. And so at the end of the day it's not always necessarily faster or easier. I think, frankly, I don't see that's a major point of difference, to be honest. I think it's not so important. When you look for an acquisition.
- [162:27]
B
Do you have any type of internal forecast around when some of the current, the new generation of like or just the new crop of AI native startups will start becoming for sale? Because like right now, if you're growing quickly and you have a great product, you can probably raise a bunch of capital. But I could imagine in 2ish years there's companies that maybe don't fully break out, that have great products, but aren't necessarily going to be public companies one day themselves.
- [163:02]
E
No, we haven't discussed that. We are because we try to focus our time and effort on things we control. So our focus right now is to just meet as many great entrepreneurs and investors and private equities and management teams and, and bankers as possible to make sure we are involved promptly every time there is a, an intention to sell a business. We have historically been a great acquirer, very fast bid every single time we actually competed and essentially no requirements on management teams to stay if they don't want to stay. So we just need to make sure we are called upon if something is happening. But we don't necessarily need to know or have an opinion whether in three years time or two years time a particular company becomes available. There are so many variables that we wouldn't be planning for, for being ready for that anyway. So who cares in a way?
- [163:55]
B
Yeah, makes sense.
- [163:56]
A
Last question from my side. How are you thinking about synergies across the portfolio? You're not doing rollups, but there are some similarities between the video products developed by Brightco Vimeo. Are you thinking about how these businesses fit together over the long term or do you see them all as individual products and companies that you want to kind of grow in their own way?
- [164:23]
E
Great question. So the short answer is the latter. The reason is that the. I think those synergies are actually much more limited than people would imagine. And, and also, and perhaps more importantly, we believe that our business units tend to perform a lot better if they are allowed to operate with extreme levels of autonom and flexibility. We want our small teams to feel like they're almost like a startup. Like we basically, we trust them to make decisions and move quickly. And if you, if you start asking many business units to coordinate on branding and cross selling and yada yada yada, then all of a sudden you're creating a ton of glue and that kills sense of ownership, agility, excitement. So it's a trade off. You want to have teams that feel very entrepreneurial and empowered or maybe extract an additional 5% of the revenue through say, cross selling or some sort of bundling. We find the former in the long run yields much better returns, so we go for it. In a way, I think you can think about Ben influence a little bit like the technological version of the PNG Procter & Gamble where there is a ton of shared infrastructure and capability. In their case, I suppose I don't know them well, logistics and distribution and marketing and product development, but they, when you look at the brands they sell, most people wouldn't even know that they are by the same broader corporation. And in a way we would like Benny Spoons to be similar. We don't really care necessarily that say, an Evernote customer knows that Benny Spoons is behind it and we're fine out of way, but we don't try to push that message on them as long as they think Evernote is perfect for them and they stay subscribed and they love it. And in fact, we think that if we try to homogenize our brands, we would be destroying a lot of the value that we acquire in the first place with these storied brands and loyal customer bases.
- [166:21]
A
That makes a ton of sense.
- [166:22]
B
Makes a lot of sense.
- [166:23]
A
Thank you so much for taking some time so soon after having a child to join our show. We really appreciate talking to you. I learned a lot. So thank you so much for taking the time.
- [166:33]
B
Thanks a ton, Luca. Come on. Anytime.
- [166:35]
A
Yeah, we'd love to talk to you guys.
- [166:36]
B
Look forward to following the journey.
- [166:38]
A
Have a good one.
- [166:39]
B
Cheers.
- [166:39]
E
My pleasure. Take care.
- [166:41]
A
Before we bring in our next guest, let me tell you about Adeo. Customer relationship Magic. Adeo is the AI native CRM that builds scales and grows your company to the next level. Fun fact. Jordi Brightcove, acquired by Bending Spoons, was founded by Jeremy Allaire, who runs circle now. Yeah, 2004. Very fun. Well, our next guest is Healy from Boom Pop. Healy. How you doing? Welcome to the TVPN Ultra Dump. Hey, how's it going? I haven't seen you in a while. What's up, man?
- [167:09]
H
Been a while.
- [167:10]
J
How you doing, man?
- [167:11]
A
Good to see you. Good to see you. Introduce yourself. For those who don't already know. Yeah, yeah.
- [167:16]
J
My name's Healy Cypher. I'm the CEO, co founder of Boom Pop.
- [167:19]
A
Yeah.
- [167:20]
J
And we are actually, you know. So funny. Hold one second here. I'm gonna close this.
- [167:24]
A
Yeah, no worries. Okay, I got you. Yeah.
- [167:28]
J
We're an AI powered group travel company. I've been running and selling companies my whole career. I'm from the Nebraska, which is a small state in the center of the country you probably haven't been to.
- [167:36]
B
And the plots, I think we've all flown over it.
- [167:39]
A
Wait, what? Grew up in Saudi Arabia. I didn't remember that. That's crazy. Oh, yeah, yeah, yeah.
- [167:44]
J
Riyadh for 16 years.
- [167:46]
A
16 years. Wow. Yeah, very cool. Anyway, let's go back to the business.
- [167:50]
B
What is somebody. If what is the must do thing in Saudi Arabia for you?
- [167:56]
A
Beast World wasn't there when you were there, how did you survive?
- [168:00]
B
Beast Land.
- [168:03]
J
The new hotness in Saudi.
- [168:04]
A
There'S this place called.
- [168:05]
J
I think it's called Alaba. It's like Petra in Jordan, but northern Saudi. And they actually, I think they're building an Amman there right now. So there's like, there's some stuff going down.
- [168:14]
A
Yeah, it's pretty cool. So give us the latest, give us the news, give us the fundraising news. Yeah, yeah.
- [168:24]
J
Thanks, man. So we raised a $25 million round, pulled in some.
- [168:28]
A
Oh, yes, thank you.
- [168:31]
B
Appreciate that, Appreciate that.
- [168:34]
A
Thank you.
- [168:35]
J
Jordy pulled in some debt and equity and yeah, it's all about growth, you know?
- [168:39]
A
Yeah.
- [168:39]
J
I think travel has produced some really good returns for venture. You know, you think about Airbnb, Expedia booking, it's crushed. People don't think about it. Travel's like 10% of global GDP. It's $11.7 trillion dollars.
- [168:52]
A
Wow.
- [168:53]
J
And often when you think of travel, especially in the corporate setting, you got to think of like whatever your EA is booking you a flight.
- [168:58]
A
Yeah.
- [169:00]
J
Turns out 60% of corporate travel is stuff involving groups. It's like an off site or an sko. And, and especially in today's day and age, we're finding the demand for group travel is the most it's ever been. Some fun facts. One, most people are exceedingly less trustful of zoom meetups. They want people, people in person. Most AI companies now are hiring event teams. Like it's one of the first hires because it turns out for scaled ARR businesses, events are like one of the number one demand channels. And so as we looked at group travel, we were kind of like, wait, wait, wait. It's this massive part of travel and it's so anachronistic. If you want to, for example, book over 10 hotel rooms, you literally can't do that online. Did you know that?
- [169:43]
A
Yeah, it's crazy. We run into this. Yeah, we run into this a fair amount of times just with a small team traveling. Like it happens. I don't know if you saw that block party that they had where it like showed up in their earnings because they spent so much on.
- [169:56]
B
Where they spent like 68 million or something on one.
- [169:59]
A
That was one of the craziest stories on one trip.
- [170:01]
J
No, this is real. Like I was at this conference and the then president of Brex walked over to me. He was like, hey, are you hulia Boom pop? And I was like, yeah, but who cares, man? Like, you're the president of Brex. Like I'm nobody. He goes, no, no. Guess what the second biggest expense category is for all of our companies after payroll. And I was like, no way. He goes, it's group travel and events, man.
- [170:23]
A
It's like, holy shit.
- [170:24]
J
So it's on the rise. It's crazy.
- [170:25]
A
Yeah. So talk through the actual product experience. How do you. And then how do you actually make money? Are you just taking a fee on top of whatever is booked? Is that the secret?
- [170:38]
J
Yeah, yeah, yeah.
- [170:39]
B
So the product is pretty rad to success.
- [170:41]
A
Or I mean it could be like seatbelt.
- [170:43]
B
Yeah, yeah.
- [170:45]
J
Turns out you should buy low, then sell high.
- [170:47]
A
Yes, this is good.
- [170:48]
E
This is good.
- [170:49]
B
Organic.
- [170:52]
A
Yeah. Heard it here, folks.
- [170:54]
J
Yeah. So it's simple. You talk to our AI agent, you give it a simple prompt. You're like, yo, I want to do an off site within whatever, a two hour drive of LA, a budget of this 50 people with some cool fun outdoor activities. What should I do? And it looks at millions of data points. It looks at weather, seasonality or teleprompter. Pricing what you've done, what your guests like, and then it puts together a couple of really good options. Like, here's Joshua Tree by the minute of what you do. Here's Pioneer Town you haven't heard of. You should go there. Here's Montecito.
- [171:21]
A
Go there.
- [171:21]
J
And then if you like it, you just say, hey, cool, do it. The AI will then go out. It'll reach to the vendors, it'll negotiate for them, look at the contract. It'll book them for you. It makes an agenda. It even makes your website in a couple seconds, keeps it live. And when people rsvp, my favorite thing is, is everyone gets a text message. All your guests get a text, and it's from the AI on the phone. And so over text, it's like, yo.
- [171:42]
B
How can I help?
- [171:43]
J
And you can say, hey, like, how.
- [171:44]
A
Do I get to the hotel?
- [171:45]
J
It's like, take an Uber, you idiot.
- [171:47]
E
Like, oh, okay.
- [171:47]
J
Or like, who else is landing right now? And it's like, oh, well, Jordy and Coogan land around this time if you want to share an Uber.
- [171:52]
A
And you're like, oh, that's awesome.
- [171:53]
J
So it's like this corner of travel, which is so big, and it's all these random joint solutions, like point solutions. And we decided, let's just put it all into one place where it's super easy. Our ultimate vision is very simply, we want to be the default way the world gets together. I'm sure you guys have seen all the. You know, I saw Chesky on. On ebpn. I think it was last week. Like, we got a problem, guys. We have a real problem. I mean, people are lonely. They're not getting together. I don't know if you saw this study. Back in the 70s, it was eight out of 10 high school seniors said they would party twice a week. Now it's 1 in 10.
- [172:28]
A
Like, that's a problem. Party. Red alert. Red alert, Alert.
- [172:34]
B
We don't even know how to party in this country anymore.
- [172:37]
A
That's right.
- [172:38]
H
That's right.
- [172:38]
B
That's right.
- [172:39]
J
So that's how it works.
- [172:40]
A
Yeah.
- [172:41]
J
And then, yeah, It's a simple SaaS model. You pay a very low amount, and then we make money from the hotels.
- [172:46]
A
Okay, cool, cool.
- [172:47]
B
Yeah.
- [172:48]
J
All the money in travel, it turns out, is in hotels. It's not in flights.
- [172:51]
A
Yeah, that makes sense. Are you. So we were thinking about taking. Taking the team to F1, and I ran something through ChatGPT. Hey. Build a whole thing. Couldn't get to a place where we could Just build it didn't really hit the goal. But are you worried about some of the agentic commerce being a headwind or can you turn AI and chatgpt into a tailwind and actually more of a top of funnel for you?
- [173:18]
J
Yeah, yeah, totally. No question it'll be a tailwind, top of fun funnel. I think about this a lot. Like AI applications are under a lot of scrutiny. Like aren't you screwed? Like it's kind of like the 1990s. Everyone's like, why would you build an app? Microsoft is going to roll you. And that's what everyone says about AI apps. It turns out it's not going to do everything.
- [173:36]
E
Yeah.
- [173:37]
J
And so there's kind of, I think there's three things you can do in general as an AI app to like protect yourself. One is you have some sort of proprietary data. We've got that we spent two years building out a database and it's got all this stuff you can't find in ChatGPT, like private dining rooms, which, how would you do that today? You got to Google it. You got to call meeting spaces. We have a bunch of stuff you wouldn't get. The second thing I think you got to have is a learning network that they don't have access to. So every event that happens in our platform, it makes the next event faster and easier and the next event faster and easier. So if you talk to hotel and they come back and try to negotiate something, we know exactly what they did at the last hundred events. We're not going to send that to them. We make it easier.
- [174:14]
A
Sure.
- [174:14]
J
And the third thing which tends to be under indexed is just building a purpose built application for a specific use case.
- [174:22]
B
It turns out like so underrated.
- [174:24]
J
Black and white.
- [174:26]
B
I know, I totally agree. I totally agree with you. I mean I've just been testing like LLMs on finding like browsing the Internet for cars that I'm interested in and they're just, it's so yeah, you would.
- [174:41]
A
Assume that like Auto Tempest is cooked and yet you can still go to like an Auto Tempest and search across all the cars, right?
- [174:47]
B
Yeah, yeah.
- [174:47]
A
This wild.
- [174:48]
C
Totally.
- [174:48]
B
And that's not even to mention. I mean my, my experience on John. John and I are, are probably not good at at like event logistics. Like we just don't pay any attention to it. So whenever we're traveling we're like what airport are we going to? What, what hotel are we at? Like five. And so Nick on our team, bless his heart, is just like kind of locked in around the clock. When we're traveling to make sure. And it would just be nice to be able to go to a single place in order to.
- [175:17]
A
You can. And we thank you for coming on the show. Have a great day. Thanks so much for coming and hanging out. Thanks, Jordan.
- [175:25]
G
Appreciate it.
- [175:25]
J
And congrats, by the way. This is awesome.
- [175:27]
E
Thank you.
- [175:27]
B
Great to have you on and congrats to your whole team.
- [175:29]
A
Yeah. Cheers. We'll see you soon.
- [175:31]
E
Thanks.
- [175:32]
A
Let me tell you, public.com investing, for those who take it seriously. They got multi asset investing industry leading yields. They're trusted by millions. Our next guest is John Tennant from Chaos Industries, I believe. Chaos, yeah. Chaos Industries. Correct. Yeah.
- [175:45]
G
How you guys doing?
- [175:46]
B
Great.
- [175:46]
A
We're great.
- [175:47]
B
Welcome to the show.
- [175:48]
A
Thanks so much for hopping on the show.
- [175:49]
G
I love one. I love the suits.
- [175:52]
A
Bo and I were going to wear.
- [175:53]
G
The dumb and dumber tuxedos to our.
- [175:54]
A
Christmas party this year.
- [175:55]
G
We might need to go yellow.
- [175:57]
A
Yellow is a good one. I imagine your whole brand is like black and white, so you can't get too crazy with it. But if you ever.
- [176:03]
G
Yeah, you know, we got some. We got some tan in there. But that's what we figured like maybe.
- [176:06]
A
Go down with tuxedo, light it up a little bit. Maybe like high vis. High vis. Like you're on the work site. Like you know, orange or something. That might be like high vis multicam. Like I think that could work. Yeah, yeah, yeah, yeah. Bring the construction site into the. Into the suit, the formal suit this holiday season. Anyway, let's kick it off.
- [176:26]
B
Half a billion dollars. Huge number. Who'd you raise it from?
- [176:30]
A
Let's go.
- [176:31]
G
Yeah, we led from Valor equity partners. Hey, by the way, also, sorry.
- [176:38]
A
I got it.
- [176:38]
G
I'm like swaggerjacking you guys in every turn. Where did you get the gong? Because again, this is something Bo and I have been talking about. We want to get like a sales gong in the new.
- [176:46]
A
You should.
- [176:46]
B
Got to get one. We should honestly make. We should make.
- [176:50]
A
We'll figure out how to send you a gong.
- [176:52]
G
I'll buy it from you guys.
- [176:53]
E
Yeah, yeah, yeah.
- [176:54]
A
We have a couple. We actually become like somewhat of gong experts where we've tested a number of different gongs. We've learned that if you don't warm them up, you can break them.
- [177:04]
B
Sizes, sound quality.
- [177:06]
A
Also the price gets exponentially bigger, like exponentially higher. So going from like a 30 inch gong to a 80 inch gong will like 100x the price. So you go from like you're going to need 800 bucks.
- [177:19]
B
You're going to need to.
- [177:20]
G
It sounds like it's a good margin for you guys.
- [177:21]
B
I know you raised, you raised quite.
- [177:23]
A
A bit who's like you can afford it.
- [177:25]
G
My co founder is probably like the greatest technical mind of a generation. The like the in depth conversation he and I got into about the right type of gong that he wants to buy from the LA office.
- [177:34]
A
Oh yeah, yeah.
- [177:35]
G
So I think we may have to both come back and talk through this.
- [177:37]
A
With you guys pretty soon.
- [177:38]
E
Yeah, yeah.
- [177:38]
A
Anyway we're not here just to talk about gongs, we're here to talk about you. Can you introduce like yourself the shape of the business, how you're describing it these days? Yeah, yeah.
- [177:48]
G
So first of all thanks again for having us.
- [177:49]
I
Of course.
- [177:50]
G
Yeah. So founder Chaos Industries. So we sort of look at the world in three waves kind of as this sector has grown over the last 20 some years. And you know wave one you have Palantir and SpaceX. Wave two you have Anduril and you know wave three we didn't see like another big multi product product sort of new prime being built. And that's sort of where that was the sort of initial thesis seen kind of where the world was going in terms of especially you know the Russians have been taking advantage of the lack of integrated air defense systems in Ukraine and they've been attacking energy infrastructure and civilian targets sort of in an effort to break the will of the, of the Ukrainian people. You know the Iranians have been helping them, them build their drone capacity and so we sort of saw this and all the while by the way the Chinese are watching here about what could happen in Taiwan. So I think we took it from the angle of we got worried about the fact that America's losing air superiority and so we wanted, so like the first wave of our products we wanted to build get away from sort of the monolithic legacy structures that were built during the Cold War and bring new operation systems to the war fight and really protect our people downrange.
- [179:03]
A
So multi product, interesting. Anduril certainly gotten there but started with a sensor tower and then anvil like one counter US drone system. Did you follow a similar path knocking down single products to get multi product or did you do the compound startup thing where you on day one were working on multiple products to bring them all together together.
- [179:27]
G
So we, we were, so we were the chunk we we're biting off first is sort of in the radar market.
- [179:32]
A
So we've got four different products.
- [179:34]
G
Okay, that makes sense now, now there's now but there's A sort of long tail of products that we're investing in that the Department of War has been asking for as well. And we'll come back on and talk about some of that stuff in the.
- [179:46]
A
Future with you guys.
- [179:47]
G
But, but the goal here is you're going to do this the right way. You have to be multi product.
- [179:50]
A
How much of what you're doing in radar, when you say multi product within radar is, is different radar technologies or the same radar, like the same data output. But this one is waterproof, this one flies, this one goes in the back of a truck. So I would say different packages.
- [180:06]
G
Yeah, I would say like completely different form factors, different technology.
- [180:09]
A
Oh, everything.
- [180:10]
G
Types of sensing technology. So we've kind of like rewritten the book on how it's done. And I think big credit to obviously Beau, he knows more about, about, you know, weapon systems, radars then you know, most people. He's forgotten more than most people will be able to learn in a lifetime. He's probably the most brilliant technical mind I've ever seen.
- [180:31]
A
And so why is the old radar not good enough? Is that because the drones are getting smaller? These can't see drones.
- [180:36]
B
Right.
- [180:37]
G
These are legacy monolithic systems.
- [180:38]
E
Right.
- [180:39]
G
So one, manufacturing times are way too long.
- [180:41]
A
Sure.
- [180:41]
G
The form factors are enormous. So I think if you look at Patriot, right, like the things size of a tractor trailer, you'll be lucky to get those within two and a half years of and three and like, look, they do some things really, really well, don't get me wrong. But these are sort of like monolithic, legacy, Cold War era built systems. And the reality is the battlefield's changed very dramatically and you need more tradable systems as well as sort of like more expeditionary systems. I think like you guys had my buddy Scott Sanders on this week, right, for Terra. So shout out, Scotty. Love you. But you know, we partnered with Forterra, right, because like the, the ability to have these systems be mobile and move around the battlefield, not have your transmit and you receive in the same location, I think is, you know, buys our warfighters so much more time than they would already have. And I think if you talk to our chief mission officer, Chris Musselman, you know, Musk was an AP SEAL named Chris Mosselman, by the way.
- [181:42]
A
Oh, let's go.
- [181:43]
G
Half the reason I started, by the way, half the reason I started trying to get my PRs in the weight room again is because Musk comes and talks shit to me every day in the office.
- [181:49]
A
Sorry, excuse my language. I don't know if I Can curse on here.
- [181:51]
G
But I think Musk would tell you. Sorry, but I think Musk would tell you that, you know, 30 seconds is an activity in a gunfight. Like, we're buying exponentially more time.
- [182:04]
A
So that's really the goal Here, is the U.S. department of War are happy to buy systems that are maybe influenced by the war in Ukraine. Are the learnings they are translating, or do the Ukrainians want something that is maybe fundamentally different from the way the US Is set up?
- [182:26]
G
No, I mean, I think Ukraine is one of the best test beds out there. I mean, we've had our systems over there for a long time. I think is we sort of look at this as two pronged.
- [182:35]
A
Right.
- [182:35]
G
Because there's a big international component to what we do. So our international partners, all they care about is, hey, has this been tested on the front lines in Ukraine?
- [182:44]
A
Right.
- [182:44]
G
And then I think, you know, if you look at some of the work we've been able to do overseas, you know, in the Middle east and in Ukraine, it is a. It is a real key discriminator. When we go in, it's like, we don't have a PowerPoint. Like, we've got data. The thing works, you know, take a look at it. Right. And so I think it's been a big, big help.
- [183:01]
A
Yeah. Yeah. Are you optimistic about the Ukraine war winding down? It feels like it's gone on much longer than I expected it to. It's been this war of attrition. I've been very hopeful that there'll be just some sort of conclusion.
- [183:19]
G
You'd hope so. I mean, I think, obviously it's just terrible what's been happening over there. And so this is why we want to do our part to help. But it doesn't seem like. Like it's ending anytime soon. But again, you know, I'm not a policymaker, so I don't have an asymmetric view into it.
- [183:33]
A
Yeah. Putin needs to pivot to AI. He's falling behind in the race for artificial intelligence. Should be focused on building data centers.
- [183:42]
B
Instead of transitioning the, you know, building tanks into building gongs.
- [183:47]
A
Yes, yes. Do something productive. Putin, stop fighting.
- [183:51]
G
I still wouldn't buy gongs if he made them, though. I'm only buying them from UK thank you.
- [183:55]
A
Thank you.
- [183:55]
B
American gongs.
- [183:57]
A
Well, well, thank you so much for coming on the show. Do you have another question?
- [183:59]
B
Great to meet. And congrats, congrats on the massive round.
- [184:03]
A
Congrats on all the progress, and thank you for everything that you're doing to support.
- [184:07]
G
Appreciate It. Thank you guys. Hope to see you guys in person soon.
- [184:10]
A
Yeah, that'd be great.
- [184:11]
B
Come on the show for, for the next round. Whatever, whatever letter that is.
- [184:14]
G
We'll do. Absolutely.
- [184:15]
B
Cheers.
- [184:16]
A
Thanks to you soon. Bye. Let me tell you about 8 sleep.com get a pod 55 year warranty, 30 hour risk free trial free returns, free shipping. And also let me tell you about AdQuick.com out of home advertising made easy and measurable. Say goodbye to the headaches of out of home advertising. Only Adquick combines expertise data to enable efficiency technology out of home expertise data to enable dramatic seamless ad buying across the globe. I was checking my eight sleep score. I got a nine 93. Let's go. Let's hear it for me. And we got Reed. Welcome to the show. How you doing? Have a seat while he's sitting down. I'll also tell you about.
- [184:59]
B
I love that.
- [185:03]
A
Your bezel concierge is available now to source you any watch on the planet. Seriously, any watch.
- [185:07]
B
Quick wrist check start.
- [185:09]
D
Yo, I missed the yellow jacket and pants.
- [185:11]
A
Yes. Yeah, the memo. It was, it was. So our, our lead sponsor Ramp raised a big fundraising round today. We have the CEO on the show.
- [185:20]
B
Fourth round of the year. We knew they were going to be raising a lot.
- [185:23]
A
We decided to get these suits and when we initially. Yeah, when we initially thought of the bit, oh, let's get some yellow suits, I sort of assumed that we'd just go and buy like a yellow suit from Target. Jordy called our tailor and got a very nice tailored yellow suit and I was like, that's hilarious. It was very funny. But we wound up using it a lot. They look great, fellas. Anyway, please introduce yourself for those who might not.
- [185:47]
D
Oh, yeah, my name's Reed. Founded a company called Knight. You and I have known each other now going on probably like three years, but we represent the biggest creators on Twitch YouTube. It's kind of taken on a life of its own now. I'd say we're like the. The idea at the beginning was like be Internet's the management company that's kind of transitioned to like be the Internet's media company. And so we bought a podcast network from Warner Brothers called the Roost. We have a venture studio, a few things that have come out of that, like Feastables with Mr. Beast Tone with Kaisenet Outtake, which is the company that you guys probably familiar with. So that's a little bit of the company. And then, you know, I represented Mr. Beast for seven years. So it was a crazy seven years.
- [186:23]
A
Okay, maybe. Maybe we start with like, like a State of the Union in just, like, where opportunities are for creators. It felt like TikTok was the hottest place to. If you were gonna be a creator, TikTok was the place you could go. Breakthrough. Then Mark Zuckerberg copied it with reels, and YouTube answered with shorts. And maybe the plateau in TikTok world was like, a little bit, like, maybe they were slowing down and then they were going through this. Will it get banned? Well, it won't. And so there's a little bit of hesitance if you're a new creator to maybe pick that platform. Is that overstated? Is there still opportunity on TikTok?
- [187:00]
D
No, there is. I think being a creator today, it's the easiest it's ever been. Just because discoverability is so easy.
- [187:06]
A
Yeah.
- [187:06]
D
You know, I think. I think that also becomes the hard part with TikTok is the algorithm has gotten so good that, like, if you. The three of us pulled up our for you pages, like, all of them would be completely different. So I think breaking it also makes.
- [187:17]
B
It more competitive for existing careers where the. If you were living in a world that was like, wasn't algo feeds, if you just got to a critical mass of subscribers, it wasn't. It wasn't as competitive with, like, a new creator would have to grind it out for five, 10 years before they could actually be competitive. And now if you're just making better content, it will get surfaced faster.
- [187:38]
D
Yeah, yeah. If you like. During COVID everyone was seeing Charli d' Amelio dance videos. Like, she went from 0 to 100 million.
- [187:44]
E
That.
- [187:45]
D
That's very challenging to do in today's world just because your content gets fed to the people that only want to watch that and your content doesn't get seen by people who don't want it. I think the platforms have actually done the same thing across the board where they don't really want creators to break out and become a Mr. Beast anymore. They'd rather, like, widen out the mid tier. And so they'd rather have YouTube, for example, I think would rather have 10,000 creators with 5 million subscribers than have 100 creators with 100 million subscribers. Like, they don't. They don't want that. I think a lot of that is, like, they don't. They also don't want creators to have any leverage against the platforms. And so it's been interesting seeing that, like, shift over the last three years to this, like, you know, push down a little bit further and make. Make it Harder for people to really break out.
- [188:29]
A
And what is the nature of Mr. Beast's leverage over YouTube? Is it just that if he says YouTube isn't treating creators, you know, correctly, that'll be front page news? Or is it something more about the structure of his business?
- [188:43]
D
I think a lot of it is when, you know, there is like negative press. And we saw this for those people that are like OG YouTubers, like PewDiePie kind of went through this whole like adpocalypse back in the day that was very negative on the platform as a whole.
- [188:55]
A
Sure.
- [188:56]
D
And so I think when you, when you get individuals that are at the top of those platforms and they start to almost overshadow the platform because they've gotten so big that when negative press starts to come out about them, now advertisers are pulling out. We started to see this a little bit on Twitch because it is a very top heavy platform. And so I think that has a lot to do with it. You know, I think in Jimmy's case, like for the longest time you would go on trending page and he was just dominating trending page. And so it's, you know, now it's like people don't really go on trending page, but even on, on homepages now, there's a lot of variety of different content. That's ultimately what the platforms want. They're trying to widen out their fan base as well.
- [189:32]
A
Talk about the evolution of Twitch. You gave us the basics on TikTok. Twitch is interesting because when I see what Google is doing, YouTube is front and center in so much of what YouTube does. And yet I feel like with Twitch, it's not really the front and center star property for Amazon as a corporation. And I'm wondering if that reflects just it being a little bit more arm's length, a little bit newer of an acquisition maybe. I mean, it's still been a decade, but what's the vibe on Twitch these days, man?
- [190:07]
D
We could have a whole conversation about this. I think your observation is correct that Twitch inside of Amazon, they haven't really cared about it. You know, they'll do deals now that is predominantly focused on Amazon prime video and Twitch is somewhat, somewhat ignored. You know, I do think, like it's such a small, like peanut inside this Amazon ecosystem that even if Twitch's revenue quadruples, doesn't really do anything for Amazon as a whole.
- [190:33]
A
And so it also has this weird dynamic where, and correct me if I'm wrong, but there might be a lot of money being made on Twitch. But a lot of it is driven by the Twitch prime program, which is not actually new dollars for Amazon. Whereas subscribers on YouTube, YouTube, premium subscribers, that's just actual dollars that people are paying.
- [190:53]
D
Yeah, you're saying like net new or. I can explain that for a second. So a couple years ago Amazon released something where, where every Amazon Twitch or sorry, every Amazon prime person gets a free Twitch sub. So they can essentially use their Amazon prime to subscribe to some person switch which, which would be $5 a month in most cases. So what you're saying is like, no, that's no longer net new revenue new. They've never released how many actual Twitch like subscribers are like actual Amazon Prime? So we have no idea.
- [191:21]
I
Oh we don't.
- [191:22]
D
But I would guess it's a large number.
- [191:23]
A
It felt like it was like at least in the early days it was like 80% because it was like, yeah, my parents have, have somebody Amazon Primedy.
- [191:31]
B
Try to make a stream that's just like make me a millionaire and they're just streaming and they're just trying to get people to like use their one Amazon sub.
- [191:37]
A
Well, I mean, tell some stories about subathons.
- [191:39]
D
Yeah, I mean the subathons became a thing a couple of years ago. Kai's obviously had the most notorable one, the mafia thon, which led into the third version of that that we did this last year. And it was, you know, 31 straight days, 24 hours a day. He had, he broke a million subscribers over a 31 day period. But subathons are not a new thing. But people using Amazon prime to subscribe to Twitch has been around for a while and that, and that is like the hook that a lot of people use. They'll be like, use your one Amazon Amazon prime sub on me. And. But the, and so Twitch does like we know kind of what their advertising revenue is, but we have no idea what the makeup of like a Twitch prime subscriber is. But it has been disappointing I think from someone who sits on the creator side that, you know, whose company represents the majority of the Twitch streamers. You know, YouTube leans so heavily or Google leans so heavily into YouTube as a platform underneath their umbrella. Umbrella where Twitch does feel like this like kind of like band of misfit toy inside of Amazon that they don't really care about, don't really talk about. Even when they do the, you know, the, the NASCAR deal and they're negotiating to get all these rights for the NBA. It doesn't even feel like Twitch is in those conversations that, that's also been frustrating for us too, because for a lot of guys that are streaming on Twitch, that's their main distribution. Have you tried to buy Twitch love? I think that we would have to because.
- [192:59]
B
I know. No, and I. And I. I know, like, it sounds crazy, but there's a world where you have all of the talent, like a lot of the most important talent on the platform. There's a lot of investors out there that if they would happily, probably if you're like, hey, this is being under, you know, this underappreciated Amazon and we can turn it. I don't know. What, What? Wasn't it like a billion dollar acquisition? Yeah, like, amazing. That feels like worth. Well, more than that.
- [193:25]
A
The psychology of the 99 is so funny. It's like, clearly there was some weird board fight.
- [193:32]
D
I just don't know why Amazon would sell it. Like, why would they part ways with it? It would only ultimately make them look bad if someone could come in and actually operate this. So I don't know. I would be the first in line.
- [193:44]
A
We were joking about how we were praying for Andy Jassy to get on Twitch because obviously he's a very by the book. You know, he operates aws, he's a very quantitative executive. But if you look at like Mark Zuckerberg is on Instagram, he's using the platform and it's like, of course Apple's keynote will be streamed on Apple TVs. Of course Google's going to do IO on YouTube. And yet Amazon hasn't really leaned into Twitch in that way of like, hey, maybe if we bring our example executives here, it's always been like, it's a little bit too crazy. Like, that's a little party and we're like, a little bit more serious.
- [194:19]
D
You'd have to think, like, if you're going to pay a billion dollars for the NBA rights, that you would have to allow Twitch streamers to go stream courtside whenever they want. They get full locker room access. Like, you'd have to open up the aperture. So all those creators could benefit from Amazon owning the NBA rights, which ultimately, like, just keeps people in the same system that can then watch the games or watch their favorite Twitch streamer backstage or whatever that stream ends up being. Who knows?
- [194:45]
A
It does feel like it hasn't been fully integrated in any meaningful way. Whereas with YouTube is like front and center in so many of the different parts of the ecosystem. With like VO3 and the generative AI stuff, it flows right back.
- [194:58]
D
Yeah.
- [194:58]
A
And so what's the Best timing.
- [195:00]
B
What's the best platform to be a top creator on? When you think about Twitch, YouTube, Instagram.
- [195:08]
D
TikTok, it's, it's to me not even, it's not even question. It's YouTube by far. Like I think just their alt monetization, their AdSense monetization. Like if you're at the, even in the top 10,000 channels on YouTube, you're making significant income. And I've said this for a long time, like we, we kind of value a fan of just like relative time spent with that individual. And, and people are spending a lot more time with individuals on YouTube than they are on TikTok. You know, you'd have to watch I don't even know how many TikToks of a single person to get up to like a 17 minute Mr. Beast video where the average person's consuming 70 to 80% of that video.
- [195:45]
A
Yeah, yeah, yeah. So do you believe in this? I like this exchange rate concept, but maybe it all just boils down to watch time. But I feel like it's 10 times harder to get like a live viewer than a video essay viewer. It's 10 times harder to get a video viewer than a shorts viewer. Maybe there's, maybe there's like a chain of exchange rates through these things.
- [196:07]
B
Yeah.
- [196:07]
D
But you can parlay it now on YouTube, which is why I said YouTube, because you can use YouTube shorts as the discoverability mechanism which is a lot easier to get someone in the door to then figure out what your content is to then create long form videos to. Now they're watching a 20 minute video and they're watching mid rolls and an unskippable ad. And So I think YouTube has done such a masterful job of just like continuing to build the platform for an amazing place for creators.
- [196:31]
A
Yeah. What do you think about this, this idea that like YouTube just seems to be coalescing around the old TV formats. Like Mr. Beast, it feels like he landed on like oh, it's about 22 minute video which is like exactly how long like an Apple episode of the Simpsons is in a 30 minute time slot. And then though there will be around eight minutes of ads because that's just the ad load that TV discovered. And some of those will be in, in the video and some of those will be out of the video. But effectively it just feels like humans landed on like, yeah, like half an hour slots.
- [197:04]
D
Yeah, I think a lot of it comes down to yes, you can put multiple mid roll ads within a 22 minute video and so you can Kind of just understand like how many mid rolls you can fit in a 22 minute video. But TV watch time on YouTube is now, I believe 11% of watch time consumed on an actual like smart TV. And so YouTube also is feeding those, those videos into a system. And YouTubers have gotten smart. A lot of them have syndicated their videos on other platforms like an Amazon or a Tubi and in those formats, usually you have to deliver a video that's like 20 to 25 minutes in length. You can't distribute a ton of video videos that are nine minutes onto Amazon or Tubi or some of these other places. And so I think just creators have gotten smarter over the years of like playing to the how do I get high adsense and how do I syndicate my cons and other platforms?
- [197:51]
A
Yeah, there's like these one off ARB opportunities. I feel like that happened like for Snap. Snapchat was a, was one for a while where the creator monetization program was really good. I don't know if it still exists in the same like lucrative way, but there was a moment where it's like if you have a backlog on YouTube, just go put it on on Snapchat because you're just making money.
- [198:09]
D
It still exists, but you have to make native content. So the ones that do what like David Dobrik does really well on Snapchat, but he's making 100 plus pieces of content a day and then the programmatic ads are just like slated. Yeah, he's posting 100 times a day. So it was like, I mean there's a lot of creators that are doing this where it's just like spam posting your entire day.
- [198:28]
C
Okay.
- [198:29]
D
So you're, you're essentially vlogging your entire day day, but you're filming it in 10 to 15 second increments.
- [198:34]
A
Okay.
- [198:35]
D
And so people will just like continue to click and like watch the entire video and then Snapchat will insert ads and that whole thing and it's done. Well, I don't know if it's going to continue like that. I don't know if Snapchat will continue to be able to sell ads within that system. But it's worked for the biggest creators.
- [198:50]
B
At least the ones that do you have a prediction? There was sort of the era in which a podcaster could get paid 50 to $100 million to like go exclusive. That kind of happened. There was the same thing happened in live streaming. Do you ever expect like, I don't think both of those have like necessarily panned out that well, for the platforms, do you, do you expect somebody to like, try to run that playbook back again to like kickstart a new network? Or do you think it's been learned enough times that it's not necessarily.
- [199:23]
D
I don't know. It always always kind of feels like someone has to continue to try to spend $100 million to create a competitor to Twitch or YouTube or TikTok. So I, I don't think it'll be the last time. I do think someone else will come to the system and put up money to do some of these things. Mixer probably won't be the last, like competitor to Twitch that fails. I would imagine someone else is going to try and do it and they're going to be able to raise a ton of money to try. But we've just seen time and time again that like, even in the Twitch example, the community on Twitch is hard to compete against. If you're Mixer, you could spend $100 million and still not even put a dent. And like Kick is trying this right now. They're probably the newest one, which is like owned by stake and they've spent hundreds of millions of dollars on trying to figure out how to compete with Twitch and they just really haven't been able to like really crack any of the live stream.
- [200:11]
A
Nvidia has not gotten into live streaming yet.
- [200:13]
D
That may happen.
- [200:14]
A
Microsoft, Google, Amazon, all these companies have plays in live streaming, but not Nvidia. Maybe.
- [200:22]
B
What's your point of view on creator payouts on different platforms? Feels like obviously you would love for platforms like Instagram to do ad scale creator payouts. We've had a debate on the show whether X payouts are good and we ultimately got to a place where we think like, like creator payouts on X make the platform worse because the content is relatively easy to make and if you just create the sort of like profit motive on the platform, it just floods the platform with content that I wouldn't say the content is like better today than it was like 5 years ago pre creator payouts. Whereas on YouTube making great videos is really hard. And so you talked about it like there's. If you can be in one of the top channels, you can, can make a great living doing it and that has a very positive effect and that more people can spend all their time, you know, creating content. But curious what your view is.
- [201:16]
D
Yeah, I think on X, like they reward shitposting like in all the like Twitter meme accounts or X meme accounts. Like if you start getting likes and retweets like it just continues to go and so like memes do really well or just shit posting does really well on YouTube. It now like comes down to statistics like you know, need a good click through rate and a high retention or else that YouTube is not gonna recommend your videos. So it's really hard. Like it's impossible to shit posts on YouTube. You can like have a clickbait thumbnail. But if the video asset stinks, that.
- [201:46]
A
Was the craziest thing Mr. Beast released. That video is like, here's his video to like 10 years in the future or something. And I think it flopped.
- [201:53]
D
Like, I think it was only three minutes in length.
- [201:55]
A
Yeah, yeah, it was only three minutes in length. So it just wasn't in the meta. And even though it's like I was mind blowing and you know, it's this moment, it's still just like the algorithm expects 20 minutes and a million dollars poured into the production. And so your, your face video, you're just like, hey, to the camera video, even though it's novel, just doesn't break through in the same way.
- [202:13]
D
But if X is ever going to compete with YouTube, they have to monetize, right? Like they have to sell ads. So I just, I don't know where they're kind of at in terms of like, do we need a video player? You guys obviously livestream on X, but, but it seems like they have to get to a point where they're monetizing or they're allowing creators to monetize their content or I just don't see why anyone that actually puts quality, yeah quality into their content or podcast or whatever would post it on X. Unless you're a podcast and you're already putting five ad reads in your video and AdSense is an afterthought and then you want more scale and distribution by X and you're already monetizing the video that, that I can see. But I just, I just don't think, think if X really wants to compete with TikTok and Instagram and YouTube that they cannot monetize their content.
- [203:01]
A
Yeah.
- [203:01]
B
How early, how early are you signing creators today? Because it's, I mean you guys have a lot of leverage from all the talent that you've worked with and just the track record. But at the same time I'm sure that when somebody pops up and now that they get shown to a ton of people really quickly, they can go from zero to millions of views in no time. I'm sure it's like a highly competitive dynamic where in a perfect world you're like, well, I'd like to let this person create content for another at least a few months, but if I don't, somebody else is going to jump in and sign this person and then maybe it'll be hard to kind of take over that relationship.
- [203:38]
D
Yeah, we're definitely never the first. I would say we're hopefully the last. Usually everyone that we sign already is half like a manager, an agent, or like someone in play. You know, I think we come in with a little bit different value prop of, you know, we've done it so many times and kind of have the blueprint for like, how do you get a creator to scale then how do you build businesses on top of them that have real enterprise value? So I would say we're a little later and we don't really represent that many people, so we're not like chasing, hey, what's hot and viral at the given moment. Although I pay attention to it and my screens are. Time is incredibly high and I don't think that'll ever change. But I'm just like, I keep an eye on people. I think, like, you know, we have this whole system internally where we just have thousands of creators and we'll just keep an eye on them. What are they doing? Are they. It's hard to have, like, real longevity and it's really hard to stay creative over, like, years. And so I see a lot of creators, like, come and go into the system. They'll make a good video, they'll go viral, and it's really, really hard for them to back that up month over month over month.
- [204:39]
B
Yeah, or they have like one bit that's hilarious for. And there's also, like, creators, like, we. There's this guy on, on Instagram that we've been laughing at a lot lately. He makes these videos where I don't even know is. I. I forget the account name, which goes to tell you like. Yeah, yeah. Like, it's like it's not a super valuable account, but he has this one joke that we do just think is absolutely hilarious. And like, there's probably. There's probably nowhere for it to go. Like, it's probably like he's getting like millions and millions of views with this one joke and then eventually it'll fade or something like that. But I mean, I think, yeah, not. Not all, not all views are created equally.
- [205:21]
D
Yeah. I mean, there's this creator that I really like to watch and he like, eats lemons in public places. So he'll be like, on a plane and he'll like pull up. Have you seen, seen this?
- [205:28]
B
Yeah.
- [205:28]
A
Yeah.
- [205:28]
D
I don't know how he onions. He does onions too. And now, and now eggs. Like not a hard boiled egg, like a full egg. And he'll just chew it but like that.
- [205:37]
B
Tyler, get ready to eat eggs, buddy.
- [205:38]
C
Yo.
- [205:39]
D
Have you guys, have you guys ever tried to bite a lemon? It's hard to eat a whole lemon. Yeah. Like he's like gagging so that, that like. Yes. I, I can't remember what his Instagram is called. Which goes to the fact of like I'm like doom scrolling and I'm like, oh, I found this guy. He's hilarious. He. He eats lemons. But I'm like, there's no real longevity here. Like I don't know where this goes.
- [205:58]
A
Yeah. Versus Harris who like, you know, tells his whole story.
- [206:01]
D
Beautiful storyteller.
- [206:02]
A
He's in the thumbnail and he tells you, oh, he brings you into his whole life. And so many of those folks that actually do a great job.
- [206:09]
B
Do you think the nature of some of these live, live streamers is actually forcing them to create products and startups or like businesses earlier than maybe on YouTube? Because I feel like a lot of the, the Twitch streamers, like there's so much unpredictability because it's a live stream that they're not super advertiser friendly.
- [206:29]
D
Yeah, I think a lot of them have figured out other ways to make money and that can be like apparel was the easiest one. Like I think a lot of them have Minecraft servers or Grand Theft Auto RP servers or now Roblox games. And so they've all figured out other ways to make money that's not so centered around Twitch prime subs or YouTube, YouTube AdSense. And so I think like, just in this like Internet world, these like kids are in the crevices of the Internet. They just like figure out how to make money. Like if you're a Grand Theft Auto streamer, you probably have an RP server. You're probably monetizing that RP server. Things that you guys probably don't even think about. And they're like, oh, like I know he streams on this like Grand Theft Auto game. Like you wouldn't even think that. Like, oh, people are paying to be in the server. And so things like that. Like these kids have figured out. Even the first gamer that I represented had a really large Minecraft server and it was like a PvP server in Java. It was pay to win and made.
- [207:22]
A
It was pay to win.
- [207:22]
D
It was pay to win. It Was like, not EULA compliant. I don't know if Minecraft's gonna care anymore, but, like, wasn't EULA compliant and it just printed money. And, you know, he would use his YouTube channel as the catalyst to drive people into that, and that that server was the real way that he would make money.
- [207:40]
A
Yeah. That's interesting. So do you have an internal philosophy on how you talk to creators through, like, where they should draw the line on their comfortable. Because, like, pay to win Minecraft, that feels like gambling adjacent. But at the same time, you know, like, we talk to entrepreneurs all the time and they're like, yeah, my first. First way I made money was like, doing some crazy stuff on Minecraft. And I'm like, is that really that bad? And then you look at some of the crazier stuff and it's like, obviously, like, rug pulling a coin on your audience is like, the worst thing you can do. But how do you think about, like, the gray area in between coaching creators, giving them, like, are you just hitting them with a ton of anecdotes?
- [208:20]
D
It's become so much harder. Like, there, there's. I mean, you can now just. You can buy CS skins, and now you're playing Counter Strike, and you're like, incentivizing kids to buy crates. There's just. We've kind of taken the line of, like, if we understand that gambling exists and daily fantasy football is fine, some of our creators have stake deals that will live stream on kick. And so it's just like, where do you draw the line? Like, I'm having a hard time with that right now because gambling is so widely accessible in so many different genres, including video games, which I think, like, even NCAA football. That's gambling. Like, they have packets that you open, open within their game mode, and you just like, spin packs and try and get players. So it's like every video game now has some type of gambling baked into it. And so it's hard for someone who, like, came into this industry that, like, didn't want to be in the gambling world. Every single video game, including, like, maybe a Roblox, like, there's like, things that you can buy where it's more like pay to win esque. And so it's tough. Like, I think that it's not something I wanted to get involved in, but now it's. We sit so deeply in the Internet and in gaming that it's just like, become a part of every creator's business.
- [209:31]
A
Yeah, yeah, it's fascinating. I mean, I played counter strike 1.3. 1.5. Like before it got productized at all. And it's hard to think back because I was talking to one of our buddies, Sagar and Jetty, over at Breaking Points, and he was kind of chastising me for being pro video game. And I was talking about my experience with Counter Strike, but it was a very different game then than it is now. And so, yeah, where you draw those.
- [209:56]
D
Lines, but it's still like, it's a $5 billion economy. Just. Just Counter Strike skins, weapon skins, knives. It's a $5 billion economy. And so that. That, to me, will just continue when. If GTA 6 ever comes out.
- [210:08]
A
Yeah.
- [210:08]
D
You know, there's gonna be a lot of people with RP servers and there's going to be pay to win mechanisms. There's going to be rank mechanisms that kids are paying for. So I don't know. That's like a big. That's a tough one for me.
- [210:19]
E
Yeah.
- [210:19]
A
I wonder if. Yeah, I wonder. I wonder where this will actually meet. Like, where the rubber will meet the road. Because you could imagine some level of regulation around, you know, can you target a. Something that is gambling legally to kids? Like maybe that, you know, if there's this type of advertisement in it, it goes over into this pool of the algorithm that's maybe 18 plus, I'm not sure. It's obviously something that, like, America is publicly discussing right now, but it's tricky.
- [210:52]
D
Have you. Are you guys using Sora at all?
- [210:53]
A
Sora?
- [210:54]
D
Yeah.
- [210:55]
A
No, not really. Okay. Not.
- [210:58]
D
Not on it. Like.
- [210:59]
A
No, I'm on it. I made my cameo, like, available. Anyone can do one with.
- [211:04]
D
You do. Okay, so people can make videos.
- [211:05]
A
Yeah, yeah. I was like, yeah, only.
- [211:07]
B
Only absolutely. Jack.
- [211:09]
A
Oh, yeah, yeah. I. I put in. Always depict me as a bodybuilder. So it's really funny.
- [211:13]
D
Is there a lot of videos on Sora? Like, are people actually doing it?
- [211:16]
A
My community has not. Certainly not moved over. And the retention on the. I think on both the viewing, certainly on the scrolling. We've been polling everyone in the studio. Hey, what's your screen time on SORA this week? What's your screen time on sora? Because we want to know, are people actually getting into this as like a consumption tool?
- [211:34]
B
Is there any hope for Sora?
- [211:36]
A
We haven't seen any of that. Now I do see occasionally there'll be a new, clearly a Sora video that's been integrated into vertical short form in one way or another. I found a funny guy who makes, like, tech comedy, and at the end he puts A little Sora clip just to kind of like, add a little spice on top. And it's funny. But, yeah, I haven't seen a lot of stuff that's been, like, really breaking out.
- [211:56]
D
I mean, at the current moment. I don't. If you just, like, base it on momentum, it. It feels like it's lost.
- [212:02]
A
It does feel like. It does feel like it's falling off.
- [212:04]
E
Yeah.
- [212:04]
D
But, I mean, there's hilarious videos on there.
- [212:06]
C
Pull this.
- [212:07]
B
Pull this video up so you can watch it.
- [212:08]
D
Is he fully jacked?
- [212:10]
A
Did somebody make one of me? Are you laughing?
- [212:12]
I
Yeah.
- [212:12]
B
Can you guys pull it up?
- [212:14]
D
I'm going to go home and I'm going to have, like, a Discord server just make thousands of videos of you and just flood the Sora system.
- [212:21]
A
I love it.
- [212:21]
D
And it'll just be your videos all over the place.
- [212:24]
E
Yeah.
- [212:24]
A
Yeah. I mean, I don't know. It might be, like, Canny Valley thing. We. We do use VO3 a decent amount for. For, like, previs. On, like, hey, we want to shoot this video. What if it looked like this? Let me get some ideas for, like, lighting and tone so I can send it to somebody. This is what we think we're doing, but we always shoot everything normally. And we have it. We actually have a benchmark where we have something that we shot manually, like, we shot the normal way. And then we try and recreate it in all the video. The AI video video systems. And it's remarkable how hard it is even when you have a perfect idea.
- [212:58]
B
All right, we got.
- [213:01]
A
Oh, you are. Why am I walking a pig?
- [213:04]
B
And the caption is what too many ramp ad reads does to an mf.
- [213:09]
E
They.
- [213:09]
A
They figured out how to make my legs small. They hacked me. They hacked me.
- [213:14]
D
Aren't those your real legs?
- [213:15]
A
Those are not my real legs.
- [213:17]
B
I did not scare my real arms and real legs.
- [213:20]
E
What is the.
- [213:20]
A
The name of Drew the pig? Why do I have a pig? This is hilarious. It does. It does look like me in the face.
- [213:27]
B
Gymshark.
- [213:28]
A
Gymshark.
- [213:30]
B
GyMSHARK. Stringer.
- [213:33]
A
Oh, wow.
- [213:34]
D
Yeah. I don't know. It feels like it's lost a lot of its momentum. I'm not convinced, like, people are going to go on there and watch content.
- [213:41]
A
So my core thesis for most of the AI generation apps like Midjourney Sora are that they are more like video games than consumption mechanisms. And so you go on there and you have an idea and you are trying to express it, and once you get it to generate it, you watch it, and you watch it just for you. And then you're like yeah, awesome. And maybe you said it's like one other person, but you really aren't. It really isn't just sit there and just a random person makes something and I enjoy it. It's more of the experience of like can I get it to generate the thing that I have in my mind, whether that's suno or midjourney. Somebody described midjourney as like art therapy. People like going on mid journey and just, and just whatever they dreamt of last night they'll prompt. And the images, they look just like any other mid journey images. They're no value to you, but to the person that generated them, they enjoy it and so they pay for the content that they produce themselves.
- [214:36]
B
How do you think there's a dynamic in the future? I think we're pretty far away from this point from a just model progress standpoint. But right now YouTube doesn't make any videos themselves. Why are you smiling?
- [214:50]
A
I just need to go edit my SORA prompt to say never depict me.
- [214:54]
B
With skinny small legs.
- [214:55]
D
No, I'm taking advantage of that Sora later you're gonna see all kinds of videos.
- [214:58]
B
No, but so right now any video that YouTube surfaces, let's say is, is they're paying out like a fixed basically revenue share on that. And there's a world in the future where YouTube knows exactly the kind of video that someone likes. So like at night I watch like documentary style videos, right? So it's like okay, Jordy likes World War II, like voiceover documentary style videos. We could just serve, we could serve them a video of an existing creator or we could just serve Jordy a video. So same topic, but we created it ourselves. So there's like some cost to like generate the asset, serve it, but theoretically a lot less. Like do you like? I think that that scenario creators are going to be creators that are worried about AI today are not worried about the right thing necessarily. They're worried about like I don't know but I think that's like a very real possibility.
- [215:56]
A
They're overestimating how much they can get destroyed in a year and underestimating how much they can get destroyed in a decade. Do you.
- [216:01]
D
But does YouTube who has been predominantly very creator first pivot to that where they're now like making content through generative AI that is ripping off let's say a documentary storyteller or true crime like did. It just feels like if they're thing.
- [216:17]
B
Is like they would, they would do that if the number if like watch time Time continue to go up like that. That like they're a massive public company, they have an incentive to like increase profits.
- [216:32]
A
Yeah.
- [216:33]
D
So I'm just saying 100% of them.
- [216:35]
B
Yeah, yeah. I'm just saying like it sounds great in theory that they're like, okay, they're very creator aligned and they've made a bunch of good decisions to date. But I mean, just.
- [216:42]
A
And I don't think it's that Sora that we just watched, like that was generated by someone. Someone who prompted it, Someone with a sense of humor, obviously. It doesn't seem that crazy to me to actually figure out how to parcel out the ad dollars if that video generates $100 of ad revenue because people watched it and then they immediately watched an ad before or after or whatever. And it's attributable. You could give me a slice because of my likeness. You could give the prompter a slice. You could give Gymshark a slice since their logo's in there. You could, you could give whoever created the music.
- [217:15]
B
Gymshark should pay for that.
- [217:16]
A
Yeah. And so if we have AI that's good enough at generating that, we should also have AI that's good enough to say, hey, this is a mashup between these two artists. Let's split the revenue share 50 50. YouTube already does this. If you put one song in, it'll route the. I actually made a video about Mr. Beast once. I used some footage and it got routed to his team probably. You got the check?
- [217:38]
D
Sorry about that.
- [217:39]
A
No, it's all good.
- [217:41]
D
Never seeing that voice.
- [217:45]
A
So there's fair use discussions, but in general, it's pretty easy for a system to understand. Okay, this is a combination of this footage from here, this music from there. That IP over here, they took Donald Duck off the shelf over there. Some Mickey Mouse, some Batman, some Spider man, and they're all from different IP owners. But let's just flow all the money through and then. Yes, if you're coming out with something completely new, then you, you don't need to pay as much and maybe you minimize that. But this all just feels like it'll happen just over decades. So I don't know.
- [218:12]
D
Yeah, I mean, music's kind of figured this out where, you know, you have writers, producers. Like there's so many different people on a track.
- [218:18]
J
Yeah.
- [218:18]
D
Oh, you say divvied out to. So I could see that happening in the future. I hope we don't live in a world where YouTube is understanding, like psychologically what type of videos I watch at 9pm and so they're feeding me generative AI video videos that, like, is taking away from a human. But, like, maybe that is a world that we live in 10 years from now. And yeah, that's, that's who you're competing against. You're ultimately competing against the system of YouTube, who is making amazing videos. And their, their incentive is to own 100%.
- [218:46]
B
Yeah, a lot of the, a lot of the AI investment thesis is around labor displacement, which is like, we have to, you know, these companies are like, we have to invest in AI because, like, future. Future spend in the economy. Instead of going to labor, we'll go to these data centers effectively.
- [219:01]
A
Yeah. I do wonder about YouTube's positioning with creators because that sort of. What do you describe 9pm serve you generative imagery.
- [219:10]
B
But that, to be clear, is like a very specific type of content that. Where I don't feel the creator matters as much.
- [219:17]
C
Right.
- [219:18]
B
If it's like.
- [219:19]
D
Yeah, but what if it's like sports news or just news in general? Like, I would imagine you could pull. They could pull that in real time and be talking about what's going on.
- [219:27]
A
What I mean is, like, that's already happening because there's going to be a company out there that says, let's use every available AI tool. Let's be the AI Native Beast Enterprises, and let's have no one in front of the camera and spend $0 on cameras and spend a lot of money on Soar credits and VO3 credits, and let's start producing as much as possible and let's use YouTube as our distribution pathway. And maybe YouTube takes a stance. I don't know that they would. I think that those creators who are, who are puppeteering all the AI slop will probably stick around enough. But I don't know, there might be a reckoning to the degree where YouTube says, hey, we're detecting AI and we're putting it in a different tab or we're not allowing it on the platform. But I would be sure shocked if they do that. I'd be very shocked because you could use. Because there are so many creative ways to use AI. Like, we just watched that. It was genuinely funny because we had context. It was not just like, you know, trying to take a dime out of someone's pocket.
- [220:29]
D
Yeah, I think, I think it'll start with thumbnails. Like, that'll be, you know, you'll get a AI generated thumbnail as soon as you upload your video. It'll. It'll watch your video and explain, like, what it thinks the thumbnail should be. And then title maybe titles is first, thumbnails is second. Where it goes from there. There. I'm a little unsure. Yeah, but you guys need Neil Mabone.
- [220:45]
A
On the show for sure.
- [220:46]
D
Ask him this directly.
- [220:47]
A
Yeah, yeah, yeah. I mean the title thing is like I've. I've titled hundreds of YouTube videos. I hated doing that every single time. And so. And I never got to the scale where I had like. Oh yes. Like I have a title person who's like amazing on my staff and like I would be displacing their job if I had a. If I had a tool for this.
- [221:05]
D
Well, it a B test it now. So there's title, a B testing thing, thumbnail a B testing. Eventually it probably just prompts you what they think you should title it and then it'll a b test it on its own and figure out where it lands.
- [221:15]
E
So.
- [221:15]
A
Yeah, makes sense. Anything else, Jordan?
- [221:19]
B
Lots more on my mind but can do it again soon. Yeah.
- [221:22]
D
Thanks for having me, guys.
- [221:23]
A
Yeah, super fun. Awesome. Thanks so much for coming.
- [221:26]
B
Yeah.
- [221:26]
A
We'll close out the show and we will talk to you in just a minute. Yeah, next time. Yellow suit. Thank you so much. We have to tell you about wander find your happy place book a wander with inspiring views. Hotel grade amenities. Dreamy beds, top tier cleaning, 247 concierge service. It's a vacation home, but better. Delian has a post here says just read this in an investor update. This is bad news for Tyler. He said older engineers who graduated from college pre GPT are actually best suited for our purposes. They have fundamental programming ability that's lost amongst most of the current generation. The AI induced thinking skills decay has begun wild. What do you have to say? Are you low background radiation steel, pre war steel or are you post war steel?
- [222:13]
C
Once you get the next model, we just need the next model.
- [222:16]
A
Then you'll be good. Then, then, then you'll be able to prompt explain to me what you did when you built the thing. No, it is funny. I don't know. I think that it certainly depends on. I mean there's obviously people that use GPT coding tools and it accelerates their learning. There's probably a lot of people who are effectively cheating at learning anything and do not accelerate. But I don't know. I wonder if this phenomenon will be sticky. I wonder with the durable. We're still so early in understanding understanding the durable trends that come out of the AI era and how it changes people. I don't know. Kevin o' Leary is neck deep in data Centers right now. I love that. What a quote. That's hilarious. What is this Citrini post? I am starting to be a real believer in AI will be a megatrend. Like no one is really imagining right now.
- [223:15]
B
I'm sure lots of people ship hosting.
- [223:17]
A
Along the way though.
- [223:19]
B
Shitposting.
- [223:20]
A
Oh well. And Bill Ackman of course said may I meet you is his preferred pickup line. What a time.
- [223:27]
B
That's why, of course, why I asked you. May I podcast with you?
- [223:30]
A
May I podcast with you? In other news, Dario from anthropic went on 60 Minutes and gave a number of interesting answers. The one thing that stuck out to me and I'd love to know if anything else stuck out to the rest of you, but previously the quote that came out of Anthropic was half of all white collar work obsolete by some timeline, five or 10 years. And the twist, the thing that stuck out in my mind was it was half of all entry level white collar work being automated by AI, which felt like a step back in terms of AAGI ness. What do you think, Tom?
- [224:11]
C
Yeah, I mean, I feel like we have to look at the exact original quote because this is one to five years.
- [224:16]
E
Okay.
- [224:17]
C
This is next year.
- [224:18]
E
Yeah.
- [224:18]
C
Half of all white color.
- [224:20]
A
He's so bullish. I love he's so bullish. But yes, I don't know. I mean, yeah, it is. It's an interesting quote to match with that Delian post about what's going on with the younger the current generation programming ability. The AI induced thinking skill decay that's begun. Maybe you don't need the skills because AGI is coming. Why build skills when you can just say, do it for me, don't make mistakes. Yes. Half of all AI could wipe out half of all entry level white collar jobs and spike unemployment. Let's play this clip. Let's play this clip from Anderson Cooper.
- [225:03]
B
Spike unemployment to 10 to 20% in.
- [225:05]
H
The next one to five years.
- [225:07]
A
Yes.
- [225:07]
H
That is, that's shocking.
- [225:08]
A
That is, that is the future we.
- [225:10]
D
Could see if we don't become aware of this problem now. Half of all entry level white collar jobs. Well, if we look at entry level consultants, lawyers, financial professionals, you know, many.
- [225:24]
I
Of kind of the white collar service.
- [225:25]
D
Industries, a lot of what they do, you know, AI models are already quite good at and without intervention, it's hard to imagine that there won't be some significant job impact there.
- [225:36]
B
And my worry is that it'll be.
- [225:38]
D
Broad and it'll be faster than what we've seen with previous technology.
- [225:44]
A
It's such a funny thing to say because it's like, you're the one doing it. I still don't get that. It's like. And that's why I I get up every day like, what is his proposal? Does he actually have a proposal? Is it UBI or something?
- [225:59]
C
I think the other here is actually even better, he says. I'm deeply uncomfortable with these decisions being made by a few companies, by few people, says Daria Mode, CEO and co founder of AI company Anthropic. It's Lee. He's talking about himself.
- [226:11]
B
It's me. It's me, Dario.
- [226:15]
C
Oh well, he has to bring peace and safety.
- [226:19]
A
Peace and safety. That is wild. Okay, let's read this post from Dwarkesh to close out, he says people with short timelines sometimes shrug off models inability to perform basic economically useful tasks, end to end by saying, oh, but we haven't trained models to specifically do those things. But this misses the point. Human workers are valuable precisely because we don't need to build bespoke schleppy training loops for every small part of their job. Every day you have to do 100 things that require judgment. Situational awareness. Situational awareness and skills and context learned on the job. These tasks differ not just across different people, but from one day to the next, even for the same person. It's not possible to automate even a single job by just just baking in some predefined set of skills, let alone all the jobs. People will sometimes debate, how much progress have we made so far between village idiot and AGI? And I'm just thinking, what the F are you guys talking about? The models are currently so much dumber than the village idiot. Village idiots generally generate trillions of dollars in wages a year. Taking shots at the village idiot. These models generate 30 billion in revenue a year. That's a good take. That's very funny. Who's the real village?
- [227:39]
B
Village idiots are underrated.
- [227:40]
A
Who's the real village idiot? It's the AI models. In fact, I think people are really underestimating how big a deal actual AGI will be because they're just imagining more of this current regime. They're not thinking about billions of human like intelligences on a server which can copy and merge all their learnings. And to be clear, I expect this AKA actual AGI in the next decade or two. That's crazy. I completely agree. It is crazy. I've been recently. What else is in here?
- [228:10]
B
Another post here. Mustafa over CEO of Microsoft AI says already our Fairwater Data center in Atlanta has taken over 15 million labor hours to build even more once it's fully finished. For comparison, the Empire state building took 7 million million and Elon comes in ratios him and said. Are you sure you're doing it right?
- [228:32]
A
Oh, it's so ridiculous.
- [228:36]
B
Okay, well, let's do a lightning round for just another pod guy who I invited on the show a while ago. He's worried about exposing himself but he says themes for 2026 in no particular order. AI trademark intensifies pannequins get steamrolled.
- [228:56]
A
Believe it.
- [228:57]
B
Humanoid robot production ramp. I guess we're seeing that with ubtech robotaxi accelerated rollout Starlink capacity ramp with coupe fast follow SMR nuclear drag and pilots federal warp speed for the US electrical grid and pharma APIs drone US manufacturing ramp enterprise software spend shifts towards AI at cost to traditional SaaS. He's a SaaS bear. AI models go true multimodal and crank engagement levers grand bargain with higher ed to restructure system in exchange for student loan forgiveness. Federal housing program that attempts to free up supply which likely fails as soon as home prices drop by 2% and boomers riot. GLP1 adoption and second order impacts accelerate with recent price drops. AI for drug discovery becomes a hot topic and naysayers will rush to downplay but will look stupid before the year is out.
- [229:55]
A
This is so crazy. There's so many predictions in here. People forget about quantum Panakins realize that autolone DQs were a non event driven by illegals defaulting. Pannikins pivot to attempting to quantify systemic risk like blah blah blah blah blah. It keeps going.
- [230:11]
B
It gets steadily wilder and wild. But we'll check back on we decouple from China.
- [230:18]
A
Very funny. Anyway, I think that's good for today. Martin Shkreli also has some data on institutional trading of private market stocks. The team's laughing because we just keep going.
- [230:36]
B
Just one more post.
- [230:40]
A
Close it. To close it out. You got to update the ramp figure in his data because ramp raised an up round. Let's go. Well, thank you for watching. Thank you for listening. What you got, Tyler?
- [230:49]
C
One last thing. Okay. Tyler Cowen on Marginal Revolution. He said there's no great Stagnation. Not anymore.
- [230:55]
A
Interesting. It's official.
- [230:56]
C
Wrote the famous book.
- [230:58]
E
Okay.
- [230:59]
A
Wait, he wrote the famous book? What book?
- [231:01]
C
His book. Great Stagnation.
- [231:02]
A
Is that what it's called? Oh, that's Tyler Cowan. Yeah. I didn't realize that he wrote a book on the Greek state stagnation. I will have to.
- [231:10]
C
Yeah. The Great stagnation from 2011.
- [231:12]
A
Very cool. Well, it's over. It's been officially declared over.
- [231:16]
B
See ya.
- [231:17]
A
I'm excited. Let's dig into that tomorrow. Let's figure out what metrics he's using, what his GDP assumptions are. I want to know more about that piece, but I want to read it before we get on air. Thank you for tuning in. Great show today. Have a good evening. We'll see you tomorrow. Goodbye.