Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

Compound Growth

Compound Growth artwork

Rethinking Value: How Intangible Assets Are Reshaping Investment Strategies with Kai Wu

Jan 19, 20261 hr 8 minAudio
ShareXLinkedInFacebookWhatsAppEmail

In a world where the most valuable assets are no longer physical factories but "invisible" ideas, how do we measure what truly matters? In this episode, we sit down with Kai Wu, founder of Sparkline Capital, to explore the shift from a tangible economy to an intangible one. We dive into how data science and natural language processing are being used to quantify human capital, brand value, and the "moats" that traditional accounting often misses. Kai shares his journey from the world of quantitative finance to building a firm focused on the "information economy." We discuss the "Moneyball" of human talent - using data to identify superstars - and why understanding the culture and intellectual property of a company is now the ultimate edge for investors. Whether you are an investor or a lifelong learner, this conversation reframes how to view value in the 21-bit century. We also touch on the human side of innovation, the importance of curiosity, and why the most successful organizations are those that foster a "compounding" culture. Join us as we bridge the gap between complex data and human-centric growth, offering a grounded perspective on where the markets - and our world - are headed next. #IntangibleAssets #HumanCapital #FinancialPodcast #SparklineCapital #TechnologyAndInnovation #QuantitativeAnalysis Follow Us: Instagram: https://www.instagram.com/compoundgrowthpod YouTube: https://www.youtube.com/@CompoundGrowthPodcast TikTok: http://www.tiktok.com/@compoundgrowthpod Wheeler’s LinkedIn: https://www.linkedin.com/in/wheeler-crowley-0a63933b/ Colin’s LinkedIn: https://www.linkedin.com/in/colin-walker-mba-6099a038/ Credits: Created By: Wheeler Crowley and Colin Walker Production, Editing and Post-Production: Tori Rothwell

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

SPEAKER_01A great place to start today would be to talk about how Colin and I met you, which was a pretty embarrassing experience for me. I feel like Kai, you did okay, but we were out we were out at Hindon Beach at the future proof conference, and one of the uh sponsors there set up surf lessons, which, you know, you're in Hindon Beach, and why not? Right. This is the the kind of conference where you get outside and it's not for a geocache, it's just to see see some people and hear some cool stuff.

SPEAKER_01So Colin and I signed up for this surf lesson, and I'd never surfed before. Colin, had you surfed before?

SPEAKER_00No, no, I hadn't.

SPEAKER_01I I would have thought that that was obvious, but I still have not surfed. It still hasn't happened for me. But Kai, you had surfed, like in Hawaii or something, right?

SPEAKER_02That's right. Yeah, I had done a bit of surfing and I'm a pretty like avid snowboarder. So, you know, the ext the skills kind of transfer. I mean, I I don't think I covered myself in glory. Um, but I got up. I got up, I think. Um, I just remember like um Meb Faber had a picture of me, like, you know, I didn't I didn't get like the wetsuit memo. So I was just, you know, shirtless on the thing. And he's like, Well, Kai, like you're really showing off here. And I was like, Well, I I didn't, I didn't, I didn't know. I I wish I had, you know, brought the appropriate attire.

SPEAKER_00Don't also gloss over the fact that we were trying to surf literally right after a hurricane. And the rip current carried us like a half a mile down the beach. It was impossible. Even even the surf like instructor, when we were trying to get up, she was just like, Yeah, I would never take anybody out today, but you guys just happen to have booked today, and I'm like, of course.

SPEAKER_01Yeah. I am embarrassed because I live in Hampton and it's a like surf community. Like the seacoast of New Hampshire, Hampton, Rye, like all that. It's like there are a ton of year-round surfers, and I feel like surfers of my people. Like I feel like I've that's a good vibe. I can align with that. It's but I can't surf. Like I and I also hate freezing cold water. So it does not work for me.

SPEAKER_02Yeah, that's the problem with East Coast surfing. It's just hard to get out there if you're not open to uh freezing temperatures.

SPEAKER_00Yeah, I drove down the beach the other day. It was like 20 degrees outside, and there were people out there surfing. I was like, what are you doing?

SPEAKER_01But my daughter does not care that we live in a beach town because like half the year, most of the year, like three quarters of the year, she doesn't want to go to the beach. And then like she's developed this weird I hate sand thing, which is not playing very well at the beach. Someday she's gonna grow up and she's gonna be like, I can't believe I grew up in this town. That's great. But right now she'd rather be like in a pool, literally anywhere.

SPEAKER_02Yeah, it's funny. My my wife was born in Hawaii, and so we have a one-year-old daughter now, and we took her to the beach in Florida over Thanksgiving. And you kind of like try to put her feet down on the sand, and she'll like scrunch up her feet to avoid touching it. So that's kind of where we are right now.

SPEAKER_01Welcome to the Compound Growth Podcast with Colin and Wheeler, where we talk all things growth from financial growth to career growth, personal development to societal progress. We explore how each layer builds on the next, compounding over time to shape who we become. Each week, we break down complex ideas and emerging trends into clear, actionable insights. Because growth isn't just about numbers, it's about understanding the world and our place in it. The information in this material is for general information only, and is not intended to provide specific advice or recommendations for any individual.

SPEAKER_01Investment advice offered through Integrated Partners, doing business as Co-Fi Advisors LLC, a registered investment advisor. Integrated Partners does not provide legal, tax, mortgage advice or services. Please consult your legal tax advisor regarding your specific situation. Past performance is no guarantee of future results. All investing involves risk, including loss of principle. No strategy assures success or protects against loss. The economic forecasts set forth in this material may not develop as predicted, and there can be no guarantee that the strategies promoted will be successful.

SPEAKER_01So obviously, we have a guest here with us today. Kai Wu is our first-time guest and also our inaugural riverside guest. So, Kai, thank you for coming here and being our trial horse today. Thanks for having me. We're gonna figure out all the bugs and all the things to do better next time. But Kai Wu is the founder and chief investment officer of Sparkline Capital, where he applies a systematic research-driven approach to asset management. With a background that blends finance, data science, and behavioral insights, he is known for challenging conventional market narratives through rigorous quantitative analysis.

SPEAKER_01Kai frequently writes and speaks about the market's risk, innovation, aiming to make complex investment ideas clearer, more actionable. He's also a wicked smot guy.

SPEAKER_02So don't make us thanks for that flattering intro. Was that AI generated? Is that from a chat GPT?

SPEAKER_01It's always chat GPT.

SPEAKER_02Well, really over the top, but I appreciate it.

SPEAKER_01You know, I you gotta love how colorful the the AI like it's very supportive. AI is very, very supportive. And I downloaded this new AI called purpose, and it's constantly challenging. It's like, that's not a great idea. Why are you trying to do this? I'm like, wait, who are you? And who invited you to critique me? I thought everybody just says I'm glorious every day.

SPEAKER_02Right. You're you just want the the old chat GPT that tells you you're the best all the time.

SPEAKER_00Exactly. Never corrects you.

SPEAKER_01Colin never says it anymore, so I have to get it somewhere.

SPEAKER_00That's true.

SPEAKER_01So, Kai, we're excited to have you here today. Obviously, one of the things we really appreciate you is about you is all the research and understanding you bring to all the topics that are really like top of mind for everybody today. We've been talking a lot about AI on this podcast. We spent, I feel like 30% of our first season doing this. And so we're kicking off season two and doing the same thing. We're right back here. But it's not going away. And we want to just take some time with you today to explore some of the themes that we think played out in 2025 that will play out again in 2026.

SPEAKER_01Kind of think about where we're going economically, as a market, as a culture, et cetera. And just get your take on things because people might be sick of hearing Colin and I talk about it.

SPEAKER_02Fair enough.

SPEAKER_00So I guess with that said, Kai, take us through Sparkline Capital. What makes you guys different? What are you guys excited about? And um talk to us a little bit about 2025 too.

SPEAKER_02Yeah, yeah. So Sparkline Capital is an investment management firm. You know, we're kind of in the tradition of your quantitative value investors, so factor folks. I started my career working for GMO for Germany Grantham Shop, which was a pioneer, of course, in in quant value in the 70s and 80s. I think what makes us different is our focus is on trying to adapt the kind of traditional value factor um to include intangible assets. So kind of the the basic premise is that, you know, 100 years ago when value investing was kind of codified by Ben Graham, the economy was industrial.

SPEAKER_02The the most important assets that drove business value were factories and and railroads and and you know machinery that was tangible in nature. You fast forward to today, and the biggest companies which you mentioned, like the Googles and Videos of the world, are more intangible. It's the assets such as brand equity, human capital, intellectual property, and network effects, those big four pillars, which we can get back into later on, that kind of drive business value today. So a lot of the focus of the firm has been on trying to expand the definition of intrinsic value to include these intangible assets, which are becoming increasingly important in the modern economy.

SPEAKER_00Got it. So going back, Ben Graham obviously wrote Intelligent Investor, focused very much on value investing. So basically looking at things like PE values, everything like that. But what you're basically saying is that there's a lot more that plays into a company's valuation these days and their success in the market versus just the historical way of going back and valuing a company. Things have changed.

SPEAKER_01That's right. Yes.

SPEAKER_00Okay, got it.

SPEAKER_01What's interesting about your approach, Kai, is that value was kind of like dead, right? People have been talking about value as almost like your grandpa's approach to investing in the market. And what you've done is take the concepts behind value and modernize them, essentially, apply them to our markets today, which I feel like are almost different than our markets just six years ago, which were also different than the markets 15 years before that. It's like we we've been going through these segments in the market where how we value things has to shift by whatever economic policy, interest rate environment, et cetera, we're navigating.

SPEAKER_01What I think is really exciting for us to talk about, because we are kind of finance meets culture, right? Like it's it's the financial aspect of this with like what does it matter to our everyday investor, everyday client. I think human capital is something that I'm really excited to explore today because everybody feels or is being treated as if human capital isn't valuable in the modern AI world. And I feel like maybe what you've identified in some of these companies, and we want to talk a little bit about how you assess this, is that it's there's significant value outside of just, you know, AI learning.

SPEAKER_02Yeah. I mean, I would say that if anything, human capital has become more and more important as technology has scaled. And so just from an abstract standpoint, the reason why is that as the economy has become more intangible, intangible assets are very unique because they provide a lot of leverage. So a top employee armed with intangible assets can achieve a lot more leverage, like a force multiplier on what they do than perhaps back in the day. There's a lot more differentiation between dispersion between the best and median employees today. So it's almost like an analogy would be like a sports team, right?

SPEAKER_02We're talking about the Patriots. Like Tom Brady is like, you know, significantly better than the replacement level quarterback. And I think companies, while not quite that extreme, are heading in a direction where the rise of the knowledge economy is leading to the idea of like a 10X engineer or a hundred X employee. You see this with AI, you know, to your question, with Meta, you know, giving out billion dollar deals to lure employees, AI, you know, elite AI talent to their firm. And like at first, you're like, that's kind of crazy. How can you pay a guy a guy a billion dollars? It seems kind of wild.

SPEAKER_02But the more you think about it, that actually makes a ton of sense because Meta is spending$100 billion on building out AI data centers. And so if that employee can make their research process even a few percentage points more efficient, right, that's actually a huge impact on their productivity. Right. So again, like the rise of technology and you know, this information makes the economy because of the scale economics of intangible assets, will allow individual employees, if they're a top employee and elite talent, to really drive results in a much more powerful way than even several decades ago, where yeah, you're you're better than the next guy on the assembly line, but like, you know, you can't beat next better.

SPEAKER_02So yeah, I think and I think I think human capital is becoming increasingly important. Technology is actually, rather than being like a replacement, a compliment, right? You need talent to run this technology. Another example would be like a technology like early adopter. So imagine you're like a big enterprise company, you're a Fortune 500 company, and you're trying to do a digital transformation. You're trying to modernize your software practices. Well, you can't just go out and like buy like a hundred million dollar like enterprise software platform and just plop it in there and hope that you get good results.

SPEAKER_02That's not going to work, obviously. You know, you need to surround that that transformation along with like managerial expertise, employee training, and perhaps even re-organizing reorganizing the structure of the company around these new practices. Right. So human human capital is, you know, both on the RD and on the implementation side, you know, tremendously important for the kind of success of any technical innovation at the corporate level.

SPEAKER_00So, Kai, is it that Sparkline is purely trading off of these intangibles, or are you still taking into account some sort of, I'll say, historical ways of valuing a company?

SPEAKER_02The the latter. So, like our starting point is to say, let's look at how things have been done, because we don't necessarily want to throw out, you know, that's the thing. The baby with the bathwater. Yeah. And like it's like, look, like obviously all else equal, you have two companies, and one company has a huge real estate portfolio. Yeah, that should be worth something. But, you know, my contention would be that that something, this tangible capital, well, at one point it was like 99% of the value of companies, is on average 40 to 50%. And, you know, in certain sectors like technology or biotech is like, you know, 5% or close to zero of the value of these companies, right?

SPEAKER_02Or even consumer brands. So even we've been talking a lot about technology, but think about consumer brands, right? If you're Coca-Cola or any of these companies, you're Disney, right? Your IP, you're you know, the the Mickey Mouses of the world, that that that's your asset, not you know, not the tangible capital you might employ. And on the services side, if you're like a McKinsey or Accenture, the same, same thing.

SPEAKER_01Yeah. That's it's really interesting to think about IP and how that's gonna be shifting, right? Because you saw the Disney deal, right, with OpenAI. I mean, to basically take all your great classic IP, I guess they're not what they're doing is they're not using like people's faces, right? So you're not gonna a lot of the Marvel characters, you're not gonna have like, you know, Thor on there because Liam. Or Robert Danny Jr. Yeah. I mean once he's p he can put his mask on, and then you can have him. Yeah, yeah. Uh but it's, you know, the value of that IP it didn't seem like it was as valuable as you might think, right?

SPEAKER_01I think they've got, you know, based on open AIs, they got some equity in open AI, but based on their current valuation, it was like 1% or less, I think. So, you know, what's the value of that IP? And are we valuing things correctly? Because one thing you said I want to go back to, you know, if human capital, what you the way you described, like how we value people, it's almost a more it's like a more empowering like vision, right? It's it's like you can step up, like anybody can step up to become, they're not necessarily going to become Tom Brady, right?

SPEAKER_01But um, they could. And I think the idea that we can, you know, it reminds me of the Mr. Beast memo, which was, I don't know, like a year ago or something like that, right, Colin? And he talks about A players and B players and C players and how like, you know, some people can graduate to one level and others are never going to do that. So I think about the value of a person obviously is more than what we get as an output at their job, but they can increase because they are people and we're all capable of this growth and innovation in ourselves.

SPEAKER_01I told my family today that I'm renovating right now. My daughter is growing and still building the person she's going to become. And I'm in the process of renovating. So please excuse my mess as I'm as I'm figuring things out. But I do think it's a really cool idea that we can develop ourselves into the A players and create our own value within these larger or even smaller companies.

unknownYeah.

SPEAKER_02I mean, look, human capital is not an immutable thing, right? We we spend our lives at birth, we basically have no skills. Then over time we learn and we develop, and you know, primarily through formal education, for better or worse, right? We spend, you know, the first couple of decades of our lives. Um, and if you get an advanced degree, even more, right? Like my brother, who's a brain surgeon, he, you know, was training for for like 20 years. And so you develop a lot of human capital in the beginning. But I think what's really important, you know, and I think what you're saying is it resonates with me, is that you know, learning is like a lifelong uh quest.

SPEAKER_02If you just basically train and build human capital and then just stop when you're you know 21 or whatever when you finish college, that's you're not going to be a competitive in the labor market. You need to be adapting to the changes in technology and society over time. Upskilling, they call it. Right. A lot of people are, you know, very concerned. It sounds like you've been talking about this theme in your prior episodes around like, you know, job displacement from AI. Right. And and my retort has always been like, look, like technology has always been disruptive. There's always been this kind of creative destruction.

SPEAKER_02Over half the jobs today didn't even exist, you know, a few decades ago, like airline pilot or social media manager. And so, you know, the key is that, you know, we as humans need to constantly be, you know, upskilling in in areas that are complementary to technology and not competing with technology. So like when the calculator came around, right, like there are humans whose job was primarily to add numbers, like, yeah, maybe you should maybe spend less time on that and focus instead on the things that are uh complementary or augmented by that sort of thing. So yeah, I mean, I think like the idea of constant improvement is kind of like really important.

SPEAKER_02And you know, to the extent a company can, you know, again, again, we're investors, so we care a lot about, you know, not just the human capital, you know, stock of a country country, but also about the individuals at companies. So to the extent that they can encourage and help their employees with, you know, on-the-job training and you know other types of things, those are generally good assets, right? So like, you know, our whole framework, what we were saying is we're looking for undervalued intangible assets. We're looking for companies that have strong IP, strong brand, but that the market is not recognizing that in their stock price.

SPEAKER_02You know, one of the many things that we have found is that companies that are investing heavily in the uh upskilling of their employees, that those companies tend to outperform. And the hypothesis was as to why that may have been the case is that at the investments these companies are making in their human capital of their employees is kind of an undervalued intangible asset. That the market, you know, is really quarterly, quarterly focused, very short-term focused, very focused on, you know, what's your earnings per share, things like that, kind of missing the big picture of the, hey, these investments, they may not pay out tomorrow.

SPEAKER_02But in the long run, you know, who who want who does who doesn't want to work for a company that will invest in their future and help them advance their career over time? If not, then I'm gonna leave and go work somewhere else.

SPEAKER_00Right. So you want to know what this reminds me of is in our own world, is the mutual fund space just in general. Whenever you have a mutual fund manager or a team that's like an absolute rock star team, people tend to invest in it. They get a ton of flows and it does really, really well. But the moment that management team leaves, it's like all these alarm bells go off and get notifications about it. We revisit the mutual fund because now it's like completely different because the management team ultimately runs the decisions and makes the selections. So it's basically the exact same thing as what we're saying with every other company, is what you're looking at, which brings me to a question, which is are there specific people that you track?

SPEAKER_00And how are you evaluating these people?

SPEAKER_02Yeah. So there are different ways to quantify human capital. Obviously, on one extreme, you can focus on like the CEOs, right? The top management. It's a little challenging to do that just because you're N of N equals one in this case. So you have to have a deep understanding of Elon Musk's psychology to know whether or not to invest in Tesla, for example. Good luck with that. Yeah, good luck with that. So we like to focus instead on like the rank and file employees. Um, and then that and that that could include like VP level or like, you know, the engineers, the ops people, the finance people at a given company. And there are a bunch of different ways to do this.

SPEAKER_02Um, you know, one area we found pretty interesting is to look at the LinkedIn profiles. So what we can do is we can say, let's look through the LinkedIn profiles of all the employees who work at a given company. So, hey, who works at Apple today, right? You can kind of look at all the profiles, pull them all up. What's the composition like? How many people work in you know Cupertino versus in London? How many people have PhDs versus master's degrees versus bachelor's? How many people graduated from MIT, you know, or Stanford in specific? How many people um, you know, used to work at very prestigious or prominent employers, right?

SPEAKER_02So, for example, like you mentioned, NVIDIA, great company. Everyone wants to work there because the stock's only gone up. If you're a company that has managed to poach talent or acquire talent from NVIDIA, you're doing something right. These are the most sought-after employees in the world. And so I think there's an element of that too. Um, so LinkedIn's a really interesting data source. You can also look at like the underlying text of descriptions of what these employees say they're doing in their profiles. Are they doing machine learning AI? Are they doing kind of more traditional statistics and you know, build a build a representative view as to the, you know, the uh talent pool for a given company.

SPEAKER_02And you can look at how that evolves through time. Like, you know, turnover is a big variable we look at too. If you're, you know, on one hand, if you're acquiring talent from you know your top most prestigious competitors, that's good. But if you're losing talent, as you mentioned, right? If you your top guys, you know, if you're an asset management, publicly traded asset manager, and all your PMs are leaving to go, you know, into the ETF business, let's say, then that's not a good sign, obviously. Right. And so, you know, you the best firms are those who can kind of keep their employees happy, right? One of the big unique elements about um human capital is that like your employees like can walk out the door.

SPEAKER_02They work from nine to five, they walk out the door five each day. You can't force them to stay. Yeah, there's some non-compete you can do, but you know, that's not that rigorous. Like there's no slavery. And so, unlike other assets where you may have like a 10-year patent or something, you're not really able to force your employees to stay. And hence the ability to kind of attract and retain talent is almost like the primary goal of most companies. Like that's kind of like the if you had to boil it down to one thing, like what drives corporate success today, it's it's that. It's the human capital management.

SPEAKER_01Right. Well, it's expensive to train people, right? Like it's it's not just like some cost, obviously. But yeah, they it's a long time.

SPEAKER_02They say that about like half the value when you hire is not just like their ability to contribute value, but it's like the kind of embedded know how they bring from their foreign former employees. Like The best practices, the technical knowledge. Like if I hired someone from OpenAI to work for me, I get a glimpse and kind of into how that organization is run for better or worse. And That's that itself is is gonna half the value of the hire, is you know, is what a lot of these studies have shown.

SPEAKER_00Yeah. And turnover is super expensive too. So you want to avoid it.

unknownYeah.

SPEAKER_01It's also interesting, like you can have these. Sometimes what you're hiring, Colin and I talk about this as business owners, right? It's, you know, are we hiring somebody who's gonna come in day one and start doing the things that we want them to do, but maybe they don't have some of the well-rounded version, like aspects of talent that we're looking for, versus like we get somebody who's just a great in a player who's bringing a lot of, I guess, a variety of experience, but can be easily trained up on certain parts of the business. And I think I would probably value that variety of experience more.

SPEAKER_01Like it's I think that's, and maybe that's the way to look at college too, sometimes. Like, you know, there's the trade route, there's the college route, and you get a very specific education or like training for to be a surgeon. You get a very specific path to go down. You don't have that broad experience. But I think that broad experience, you know, personally, my background in finance was zero before I got into finance, right? Like it was, I was a creative. I got my master's in screenwriting from Boston University, and here I am as a financial advisor.

SPEAKER_01That's like a journey that had a lot of life experience, a lot of different employment experience that helps me out today. Um, I think sometimes about, and we can kind of talk about the job market right now, but there's been a lot of talk about, you know, grads, college grads having a hard time finding work, what entry-level positions might even look like in the future versus hiring for that talent, right? You know, or just like focusing on growing people in the business and maybe you don't have as many people coming into the business.

SPEAKER_01I think finding that diversification, that well-rounded experience is going to be maybe harder. Um, have you thought about that? Is that something you think about at all?

SPEAKER_02Yeah. I mean, I think the the general framework I would use is there's kind of whether you whether you want a generalist or a specialist, um, depends on two things. So one is like, what does the organization look like? Like for a smaller, kind of more entrepreneurial organization like ours, where you have a few employees, like, you know, you need to wear a lot of hats. The employees need to wear a lot of hats. Um, whereas for like a very large, kind of structured, siloed or is, you know, kind of more enterprise Fortune 500 company, yeah, maybe if someone gets hired to just do one thing over and over again, that's fine. And that actually is the optimal scenario.

SPEAKER_02And the other thing is a bit of like the time horizon question, which is like, you know, in in general, you can hire somebody to kind of plug and play and they hit the ground running right away because they have all the skills required. Or you can hire someone who may not have the specific experience you need, but has like a good foundation and a lot of upside with regards to once they eventually learn the planning stuff, they actually have a lot of life experiences that can help them relate to clients, to give an example. Um, and so it's a bit of a J curve question, which is like, do you need to monetize these employees in in the next quarter? Or is it like more like, hey, I'm making an investment in an employee that, you know, maybe not, you know, it'll cost me managerial time and and resources to keep them around for the next year.

SPEAKER_02But in like five, 10 years, like they're gonna be like a superstar and you know, may take my job one day, which would be in in a good way, right? So, like in a good way. So I think there's like, yeah, I think there's those those two two elements. Like obviously, I'm much more like you. Like, I studied liberal arts in college. I took one like finance class, it was like an accounting class, pass fail. Um, and you know, didn't never saw myself really going into finance, but you know, here here we are now. So, like, you know, and I do think that, you know, as you said, like my life experience is, you know, my mom's an artist, right? I do a lot of research, as you mentioned, a lot of data visualization, trying to understand how to kind of you know conceptualize information in a visual way.

SPEAKER_02Like that's actually quite powerful, I think. I think a picture tells you is worth a thousand words. Um, so these sorts of like things that have, you know, uh political science, economics, like my last piece, which we can get into around AI CapEx, like a lot of it drew on the idea of these like credit cycles and capital cycles, which actually dates back to my senior thesis, my economics thesis in college, where I worked with um Ken Rogoff, uh, who studied financial crises, but also this guy named Ed Chancellor, who's kind of the preeminent uh historical um historian on the topic of capital cycles.

SPEAKER_02Right. So that's something that I kind of did, and like, you know, it was a really cool and like interesting academic exercise. But like, you know, you may not have thought that it would actually be useful in in investing. And then, you know, we end up here in 2025 and AI companies are plowing five trillion dollars into building out the new AI platform, and suddenly the capital cycle is extremely relevant. So, yeah, I mean, just going to your point, like I think a breadth of experience can be a huge asset, but sometimes it takes you a little bit longer to find your footing if you're not specialized um from day one.

SPEAKER_00Well, we always talk about that too as a business. And we had a Tony Robbins thing a few months or a year ago, whatever it was. And it's basically all about who you surround yourself with. And you're only gonna be as successful as the group of people that are around you. And if you're hiring people that I'll say like haven't had prior success, then that's probably gonna bring your intangible value down versus if you're pulling people from Google and Apple and building up your company. And to Wheeler's point, it's like, you know, do we want someone to come in and do what we tell them? Or are we looking for someone to come in and revamp and take us even further?

SPEAKER_00And we think about that a lot. And I'm wondering, like, in regards to Sparkline, just you in general, if you were to hire a team and build out further, you would probably take this into account given this is your whole thesis. And what would you think about that?

SPEAKER_02Yeah, well, I I think one thing I spend a lot of time studying is culture, like corporate culture. Um, and and the reason why corporate culture is so important is that it's kind of like the unspoken norms, like the way of doing things that can either be a multiplier on raw talent or or a repressor, right? If you're a smart innovative person in a culture that doesn't reward innovation, you're you know, likely going to be unhappy and eventually leave, and vice versa. Right. So I and I think what's really interesting about culture. So I actually studied this like quantitatively, is you can look at the cultural profile of a company.

SPEAKER_02And then you can say, hey, in two years, five years, 10 years, how does that profile look relative to where it started? And it turns out like it's a very, very slow decay. It's kind of like, you know, America was founded on the principles of like liberty because of the founding fathers. And like, you know, we fast forward several hundred years and like, yeah, it's kind of the same vibe, right? We still care a lot about freedom. You know, in other countries might be different because of the culture that their founding fathers impose on their nation. So I think like to your question about building out a team, I think the most important thing is get it right the first time around. Because it's very, very difficult to go back and fix things that are broken.

SPEAKER_02And even if that takes more time, you know, to get right, like it's a worthwhile investment.

SPEAKER_00How do you how do you go about grading culture? Like, is it just about like these people are pulling these people from this company and this company had a great culture? How are you going about valuing that?

SPEAKER_02Yeah. So for the way I look at culture is you know, viewing culture as not imposed top-down. It's not like the CEO says, hey, today our culture is innovation, so we're gonna do that. Right. No, culture is like an emergent property, it kind of emerges bottoms up from like the collective, you know, processes and norms of the you know, senior management, but also the rank and file employees. And hence the way that I go about quantifying culture is to look at the glass door reviews of different companies. And so glass door is kind of like a Yelp or like a, you know, for for companies where by their former or current employees can like rate their employer on a variety of dimensions.

SPEAKER_02And you have to kind of demonstrate that you are an employee or were an employee to uh to be able to post a review. And there's l it's a pretty rich data set. Like a company like Nvidia will have thousands, if not tens of thousands, of reviews posted. And you can the the data set is both structured data, like how many stars does this company have in terms of like how good is the CEO, how you know, how good's the place to work for, is it a you know high employee satisfaction or not? As well as unstructured text, where employees like kind of fill out you know, effectively a paragraph about what they think the culture is like.

SPEAKER_02And that's I think really the gold line. And so I started looking at this in 2021. This is before ChatGPT was released. And so, you know, my contention was that the structured star ratings were fine as a starting point, but you know, really the the information content was going to be in the kind of the more nuanced data of the unstructured text data. Um and you know, this is before full-fledged large language models, which we now use, but back then it was more, you know, doing these word embeddings. And so we would parse the text of all the comp all the reviews for all the companies and kind of like put them in, like factorize them into a few dimensions that were most relevant.

SPEAKER_02And so one of the things, if you step back, one of the interesting things is you can actually go on corporate websites and like look at the values they post. And what you'd find is this kind of like power law like long tail thing, where like there's a few values that all companies say they ascribe to, whether they do or not, which is like integrity or maybe innovation, teamwork, transparency. And then as you go down to the long tail, it's like more idiosyncratic things like loyalty or whatever, right? Like risk taking or risk aversion. Like sometimes they're not even consistent. And so what we do is we focus on like the main values that most companies would ascribe to and say that this is probably a good thing.

SPEAKER_02This is what we we hope to achieve. And so we can kind of like rate all companies on basically seven different dimensions, um like process orientation, detail orientation, teamwork, innovation, customer focus. And what's really interesting is you create these kind of like Myers Briggs or big, you know, big ocean type personality test things, big five personality type ratings for each company, it kind of like maps in a similar way, where it's like, you know, Amazon cares a lot about the customer. That's been their focus, you know. Whereas, you know, Meta, they move fast and break things. They care a lot about, you know, innovation, maybe less about um, you know, integrity.

SPEAKER_02Yeah, yeah. I was gonna be a bit more cr critical. Um, so yeah, I mean, every company has a different personality, a different thing that they value. Um and so you can kind of look, you can kind of do that to kind of quantify the personality uh traits or the I guess cultural personality traits of each of each organization. Um and then you know, you do a variety of things. So for for example, we found that of the um traits that are considered you know universal goods, they tend to relate, they tend to correlate with with good stock performance. So if you basically build a portfolio through time of all the companies with the best cultures on these seven dimensions, those stocks have then gone on to subsequently outperforming the market or subsequently outperform companies with with their scoring lower on those dimensions.

SPEAKER_02You can also do things that are more kind of interesting and like look for unique cultures, right? If you're like in an industry that's very process oriented and you're like doing something totally different, is that good or bad? And historically it's actually been good because it's like, you know, you're willing to book the trend and like be think different if you're, you know, to take Apple's terminology on that. So that's you know, one that's one area, which is like these cultural profiles of companies. And then the other area is employee satisfaction. So you can kind of reread these documents, but in this case, look for dimensions that are correlated with like employee satisfaction.

SPEAKER_02Um, and we create these like different dimensions in a similar way, where it's like looking for cultures that are considered like rewarding for career growth, that are fun, that are like, you know, good if this is the work. Companies that have uh better, you know, more employee goodwill or better cultures tend to outperform those that have less good cultures. And then the final thing we did on this vein was look at the opposite, was to say, let's look for toxic cultures, right? Look for cultures where, you know, uh there's sexism, racism, like double standards, phrases that you know might appear in a review or like old boys' club or uh, you know, double standard, implicit bias, like racism.

SPEAKER_02And so this is kind of obvious. Like it obviously, if you're if people are like young to talk about how your company is racist, that's probably not a good thing. And lo and behold, unsurprisingly, the companies that um are scoring high on like toxic cultures tend to underperform. And all these results are both like across sectors and also within sectors, right? Because you might say, well, like technology tends to be more innovative than like industrials, but even within industrials, the most innovative companies tend to outperform. Even within a technology, the most team-oriented cultures tend to outperform. So these do appear to be universal goods.

SPEAKER_02And then like there are like that long tail I mentioned of a lot of more interesting credit cultural traits, which are contextual. Like, for example, like process orientation or whatever is like it's good in like certain contexts, but also not good in other contexts. So, in a more creative role, it's not actually that good to be very rigorous and like command and control, whereas like in the military or something, like that's something you do want. Um, you don't want like rogue agents. Like, you know, there's a couple of interesting dimensions. I guess the final thing I mentioned on culture is that one of the really important things is that it's not, it's not really the the way I measure culture through these Glauctor reviews has a pretty low correlation with what the CEOs are saying about these companies.

SPEAKER_02Because if you're a CEO of a company, of course you're gonna say I have the best culture in the world. Like Enron said they had a great country culture, and then they have they all did a deal for drive as well.

SPEAKER_01Everybody said that. Everybody loved Enron.

SPEAKER_02So everyone wants to say they have a great culture, but like do they actually and it turns out that the correlation between what companies say they have, either on their websites or like in their earnings calls versus what they actually have based on like the reviews of their you know, rank and fall employees, are pretty low correlation, right? Like because everyone wants to like dress up and it's kind of like cultural theater and talk about how amazing their cultures are. Whereas in reality, trickle is a huge amount of things.

SPEAKER_00Trickle down doesn't really work, essentially, isn't it? That's what you're saying.

SPEAKER_02Right? Yeah.

SPEAKER_00The trickle-down culture.

unknownYeah.

SPEAKER_02Yeah, well, yeah, I I guess what I mean is like what a CEO says their culture is, like you should take it with a grain of salt because like their incentive, of course, is to make it sound like they're just amazing and they should get a high market valuation. Whereas in reality, like the ground truth of it is is what you you as an investor should be focused on.

SPEAKER_01Can we I have a couple quick follow-up questions? I really want to get into the asset light research you did, which is really cool. Uh, but before we get that, sometimes I think there's perceived culture issues or like there's headline risk that like forces us to interpret a culture without doing some of the research that you've done. You know, we have clients, for example, that are very anti-Tesla or anti-Amazon or or whatever, but you might do some through your research, say Amazon seems like it has a great culture, right? You know, there there are the horror stories out there of like, you know, people don't have time for a bathroom break because they have to be so productive and regimented.

SPEAKER_01Do you ever find any conflict between kind of like the, you know, I guess the rumor mill versus the actual data? The reality. Yeah.

SPEAKER_02Yeah. I would say kind of similar to what CEOs are saying the culture is. There's also just like the general vibe as to like what a company's culture is perceived as, which, you know, may or may not reflect reality. Um, and then I guess the final point about Amazon specifically is like this do this kind of fact that like culture is multi-factor, multifactorial, multi-dimensional. There's not just good culture and bad culture. There's good on dimension X, but bad on dimension Y. Amazon's a great example. Like they're known for like their innovation, of course, and their customer focus, but they're also known for their brutality to their employees.

SPEAKER_02Like, you know, what's the net net impact? My guess, I haven't looked at this recently. My guess is that they'd be kind of in the you know, just net positive, like in terms of like if you were looking at a purely from an investor standpoint, is their culture one that should you know foster excess stock performance? I think you know, putting aside all else that they do on that one dimension, they probably are positive, but recognizing that like they're they're not scoring strong in everything. And and there are probably very few companies that are strong in everything because there are inherent trade-offs in terms of how you can allocate your cultural capital.

SPEAKER_01Um, they're large companies. Yeah, like it's really hard.

SPEAKER_02And there's and there's a lot of dispersion too, right? Yeah, exactly. Amazon is almost unique because they have a huge, you know, they have a ton of employees, many of whom work in warehouses, and then they have their software engineers, and then they have their AI employees, which they may, you know, have may have different subcultures.

SPEAKER_00Well, that's what I was gonna say is like kind of my hardest part with this is that these companies are so massive that one line of the business might not have as big of an impact as another line of a business, but you're going to have the reviews co-mingled, so to speak. Right.

SPEAKER_02Right, right. Yeah. So the uh for very big companies, that does become a challenge. I mean, there's things you can do to disentangle it. Like people will, you know, in an anonymous way, post like what they did. Uh I'm a software engineer or I'm like a warehouse worker. Um, and you can, you know, weight things how you feel appropriate for a given company.

SPEAKER_01That makes sense. Yeah, that's true. I have um a client that is I I think my like my test client. He's the he's the client that I think embodies a lot of the FOMO. And then on the other side, when the FOMO drains out, like just the fear. And what's interesting about where he's at right now is he's really struggling with the fact that he sees all this potential. He's looking at a lot of like the Bitcoin miners that are trying to pivot into energy, right? And he's seeing like these great PE valuations.

SPEAKER_01And then he's like looking at Costco and Walmart and saying, but wait a second, why are like Costco Walmart being valued like the great growth companies of the world? And is that maybe more because of like the brand value or some of the intangible assets that those two companies have? And I'm picking on those two in particular because they're the ones we just talked about. But I mean, Costco has been the poster child all year, or all 2025, anyways, you know, consistently historically for a long time. You know, it's a great culture, it's a great brand, and the stock is rewarded for that.

SPEAKER_01Do you think that's what we're seeing there versus like some of the startup enthusiasm that maybe ran too far as part of like the AI wave in 2025?

SPEAKER_02Yeah, absolutely. I think the reason why Costco has has been considered such a great company, right, is because of the things you mentioned, like really strong brand. People love Costco. They love talking about their trips with their families to the Costco. And, you know, it's almost like a viral social thing. And, you know, they're also very famous for paying their employees really well, you know, given their given their position. So like if you're working in a Costco warehouse, you get what economists call it an efficiency wage. You get paid like higher than your competitors, your peer group might pay. And they have this theory that that generates employee goodwill, which would like increase the retention rate and you know generate externalities to the business.

SPEAKER_02And that has been true. I think in the case of Costco, you know, the one thing is that they are definitely a high quality business. The thing is that, you know, from an investment standpoint, over the past decade, what's happened is that this is kind of coincident with this idea of value investing being dead, is, you know, a lot of quality businesses have become very, very like highly valued. And so I've actually been somewhat negative on Costco only because people just think it can do no wrong. Oh, this is a great company, we should just buy it at any price. And I think, you know, any stock, we'll get into this with the technology stuff too. Like we got to divorce quality from like price.

SPEAKER_02Because at some price, anything that's even a good investment is no longer a good investment. And I think that's that that's I think to some extent what Costco has suffered from more recently. Not to say it's not a good company, it's just, you know, at some price it's no longer a good investment.

SPEAKER_00This goes back to what you were saying before with vibes. Like if someone has a bad vibe about a company, it could go the other way where everybody has such a huge vibe, like positively, with a company that it could drive up the valuation to a point where it doesn't make sense, which sounds like is I mean, I'm not gonna speak to Costco, but just could happen in general.

SPEAKER_02Yeah. I mean, you see this with Tesla is another good example. People love, you know, at least there's a subset of people who love Elon Musk and think he think he's just a genius, which he definitely is. Um, and and he's gonna therefore, you know, shepherd the the company from, you know, electric vehicles to autonomous driving to robotics and kind of always shifting into the area that, you know, may be kind of the most explosive. Um, and yeah, I mean, that that is a a vibes-based investment. I think it's worked, it's worked out for cash-like investors, you know, obviously given obviously with volatility, but it's, you know, kind of a little bit outside my framework where I'm looking for more for like fundamentals as measured by like how strong is your actual brand, how high a qual quality company are you actually, and then trying to assess you relative to the valuation, the price as an investor you'd be paying to obtain that that company.

SPEAKER_02And yeah, go to your point, because I think that the Bitcoin miners or now the data center companies are, you know, a pretty good example, you know, of companies that are basically very much like their their valuations are very much like linked to this to the euphoria around AI and you know, formerly crypto. But like, are is it actually a good business model? Right? Like, is there actually a moat around what they're doing? I would argue that probably probably not. Like, yeah, they have some power contracts that might be kind of exclusive for the time being, but like at the end of the day, like, you know, these sorts of like kind of read-like things where you're just kind of like buying GPUs and leasing them out or you know, whatever.

SPEAKER_02Like, I don't know if that's a great business. Like, that's probably like a low margin, like low ROI business to be in. Yes, these companies are trading at pretty elevated ROIs, pretty elevated valuations, which is kind of like to me inconsistent with the businesses they're actually in. Like if you're Costco and you're a high quality business, you deserve a premium uh multiple, maybe not as a premium as it's been getting. But if you're like, you know, a in in a commodity industry, you probably shouldn't get such a tech multiple, even though you're just like a data center.

SPEAKER_01All right. Well, we bait we buried the lead here because one of the things we really loved was your research piece, I think from like October. Um, and you always do a great job, Kai, with your research of taking some very complex ideas and making them easily digestible for the average layman like myself. So maybe we could kind of talk a little bit today about, you know, how we're seeing a transition from in the companies, the metas of the world, for example, that are used to be very asset-light and are becoming very asset heavy, and how that is going to shift the way those stocks perform or the way that we think about those stocks and the way we think about their competitors or new stocks moving forward.

SPEAKER_00And just before you get into that, can you just talk through what you mean by an asset-like company?

SPEAKER_02Yeah. Yeah. Yeah. So an asset-like company is a company that relies primarily on intangible assets to drive results. Um, so that's a company that uses like network effects or intellectual property, brand equity, human capital, a lot of the intangible assets that we mentioned. Whereas an asset heavy company is kind of more of an industrial business that relies on physical capital, whether that's financial capital in the case of a bank or you know, physical machinery or factories or data centers in the case of most industrial businesses.

SPEAKER_00Like a McDonald's with a bunch of branches and real estate and stuff like that, asset heavy.

SPEAKER_02Unless it was a franchise model, in which case it would be white. Correct. Because they'd be basically selling their brand to their franchisees. So it's a little bit depends. But yes, if yours if you if you are a McDonald's and and you're making money because you have like a physical real estate portfolio, then that's more asset heavy. Whereas if you're just saying, hey, I have this really great brand and people will come to my McDonald's no matter where we put it, that's more more brand focused. So it's a combination, I guess, in this in this example. So that's asset light, that's asset heavy. Historically, the Magnificent 7 stocks, which are like Google, Amazon, NVIDIA, Apple, Meta, Tesla, Microsoft, are the kind of canonical examples, like the pinnacles of asset-like businesses.

SPEAKER_02Like Google is like just the world's perfect business. They have this thing where they have this like search and you know they can serve it at basically infinite scale. It costs them almost nothing. They don't require much capital and they just earn these huge returns on invested capital because they're asset light. They don't require capital. That's been an amazing business model. The Magnus 7 stocks have over the past decade massively outperformed the stock market such that they are now like one third of the SP 500. Like they've just done so well that they're becoming an increasingly big part of our portfolio just by virtue of their appreciation relative to everyone else.

SPEAKER_02You know, trillions, tens of trillions of dollars of wealth for the remisser. So it's a great thing. The problem is that this AI arms race is kind of transforming them into a kind of more commodity business, a more asset heavy business. And so the big four hyperscalers that's Microsoft, Amazon, Meta, and missing Google, yes, are spending uh$400 billion um a year on building out um AI data centers. That's their CapEx budget for this year. And they are guiding that the same number moving forward for the next four or five years.

SPEAKER_02Right. So we are set to spend, you know, over a trillion dollars just on those four companies. Analysts assume$5 trillion dollars is kind of the expectation for the next five years of how much capital expenditures are going to be put into building out the platform, these data centers, this platform for the AI revolution, which is just a huge amount, huge amount of money. And now here's a challenge which is that if you go back historically and look at the, you know, look at past technological revolutions and think about like the railroad build out in the US in the 1860s or the dot-com boom, the internet build out in the late 90s, what you find is this is this pattern where the folks who are actually making the investments in building out infrastructure, the railroads in the 1860s and the telecom companies like the global crossings of the world in the 90s actually failed.

SPEAKER_02Like most railroads went bankrupt in the 1860s like three times over, right?

SPEAKER_01Yeah.

SPEAKER_02Yeah there were a bunch of different panics. These companies like just kept going bankrupt. You know, yes, ATT is still around. They were of course a a primary player in the build out of the doc of the telecom, the fiber for the internet, but their stock went down like you know 80% or something, right? And this took them decades, decades to recover. And then as I mentioned Global Crossing and some of the other firms just went bankrupt to zero basically. So that's the kind of like paradox here, which is yes, these technologies are transformative. Yes, the internet changed the way we live our lives. Yes, the railroad connected the US, but the folks who actually built it out to actually put the capital at risk to do it all failed.

SPEAKER_02Why is that? Right. And that gets this idea of capital cycle which I discussed earlier. So the capital cycle is this idea that you know it's a kind of supply side um concept where you want to focus on the the supply of um capacity going into a sector. So the way that the kind of example would be an investment opportunity appears in this case, let's say the internet. And all the investors are super excited about this opportunity. And so these companies will start investing a lot of money into building out fiberoptic cables to underprint the internet.

SPEAKER_02This is a very capital intensive activity. And investors are like this is amazing. Every time you announce a new deal, I'll bump your valuation up because I just want more and more exposure to this amazing technology. And the problem is that it may be rational for any individual company, but collectively there's just a massive influx of of capital into this sector, fiber optic cables. And then when when demand materializes but materializes too slowly, people realize oh wait a second we have stranded assets. We have this glut of capacity now where you built out all these miles and miles of hyperactive cables and only 5% or 10% of it's actually being used.

SPEAKER_02This is the dark fiber problem. And so now you're kind of stuck with a situation where these companies can't the prices collapse these companies can't charge any money for a good that where where supply is so far in excess of demand. And it takes them years and years and years to kind of work out that excess capacity. Even if they no longer build more fiber, right? They um it takes them a while for demand to kind of get to the point where it's you know matched with supply and you know to the extent you're a levered company, you go bankrupt and even if you're not, you you struggle immensely.

SPEAKER_02Right. And so that that's the pattern of the railroads as well and the shell boom in the 2010s you go to the what's happening in AI today and you see the exact same dynamic. You see the this kind of game theoretical problem I call it a prisoner's dilemma where these these big tech companies are all in this kind of race to win. They view AI as an existential threat to their businesses. And as a result they're all trying to you know win what they believe to be a winner take all market. You have guys like you know um like Larry Page or Google saying I'd rather go bankrupt than lose this race. You know the the CEOs of Meta and other companies have made similar comments of course Altmond is probably the most aggressive at investing in this.

SPEAKER_02And so maybe the rational thing would be for all these firms to kind of like maintain their oligopoly status by investing gradually in this and kind of seeing how things go. But individually their incentive is to just kind of go all in with the idea that if they can then beat out their competitors they'll win the entire pie for themselves. And that's exactly what's happening. And I just don't see I don't see what puts an end to that right like that's just basic game.

SPEAKER_01Somebody wins right like somebody actually I think that someone will win. Yeah.

SPEAKER_02See I mean if you what you don't think you don't agree Kaya? I actually don't agree. I I actually think that the I think that these are commodity assets. These are these are commodity assets.

SPEAKER_00I mean I was gonna say by winning the race doesn't mean you've actually been successful. Like someone may come out on top but they probably got destroyed along the way and on top of that everybody else that didn't win so to speak probably is decimated. Right, exactly it's a pyric victory yeah I don't know that's a very interesting theory.

SPEAKER_02Yeah so like my argument would be this so think about the model layer. Um so in the model layer the first uh I think the GPT three came out it was a huge breakthrough right because they basically open AI realized that the scaling laws um you just throw more compute and data at the problem and you can kind of get a better model. And so if you look at the charts of like of performance of models at that time the GPT three model was like so far ahead of all the other competitors at Google, Microsoft, and all the open source models. What's happened is that over time there's been a convergence where actually now the best model by most people's assessment is is Gemini, the Google model, which recently came out.

SPEAKER_02And the Chinese models of Deep Seeks are actually not that far behind. So I think what you're seeing is yes, initially there's this view that there's a winner take all element whereas if you're open AI and you invent, you know, the best model, that'll just lead to a run around a runaway advantage. But I think what's happening is it's actually the opposite where there's kind of a convergence where maintaining leadership's hard you know GPT, what are we on now? The the last two five point five point something. Yeah 5.2 and 5.1 are basically the same five it's been kind of like a diminishing like asymptoting of the performance for the open AI models while the other models have have converged right which to me says all right these are all kind of commoditized.

SPEAKER_02I have all four models or I have those those two plus like grok and like Claude installed on my phone and I will just go to them kind of randomly it doesn't even matter like whatever is like whatever my finger happens to touch.

SPEAKER_01I do have I just put an AI folder on my phone yesterday. Like I'm like well I go to this AI for this thing and I go to this AI for this other thing. And yeah you're right.

SPEAKER_00Yeah. And I mean to your point it's true. I mean the difference between four and five to me I hardly notice the difference and the amount of infrastructure that was required to get there was absorbent.

SPEAKER_01Yeah it's a good point. That's right. And yeah like how you know you could be Apple and just let all these other guys play this game and then just invite the winner to or maybe not even make a deal.

unknownRight.

SPEAKER_00Yeah exactly right like with your with your huge amount of cash.

SPEAKER_01Yeah I mean you're taking a back seat you're watching this all unfold you're not spent you spending the same exorbitant massive quantity of money on that and it's maybe that's the right strategy. But I also think that some some of these like take Nvidia for example you know they're recognizing that you know the manufacturing business is this and their SaaS business is this. And if they can build this out then it creates some some more sustainability. And I think some of this is like understanding the various layers or silos even in some cases of the business.

SPEAKER_01If you think about ATT like you brought up earlier, you know, maybe their mistake wasn't in spending all this money on the fiber. I'm not I think it probably was but whatever like we built down all this fiber. We're all benefiting from it now. That's great. A company's still around that I'm happy for them. But you know what they didn't do is I didn't I don't think they thought about cell phones. Right. And like the next the the second derivative like two steps away from laying out this fiber created the you know the roadmap for the Wi-Fi that then they didn't really benefit.

SPEAKER_01They still have they they put that out there but they lost I guess the opportunity to be maybe an Apple in that situation.

SPEAKER_00Yeah good. I was just gonna say so so to that point like if you're ATT and all these other companies that spent these trillions of dollars building out the system, like who benefited and I guess what are the thoughts there?

SPEAKER_02Well so the real winners of the of the internet were like Netflix and Meta and Google who not only did they came out come came around a little later and they didn't have to make the massive investments in infrastructure, right? As I mentioned Google was a capital light business. They had raised a few hundred few tens of millions millions of dollars or whatever before IPOing these companies not only benefited from that, but they also benefited from the ability to access subsidized fiber, right? Because as I mentioned, you know, 85% of fiber was dark and the price of bandwidth collapsed 90% after the AI bust.

SPEAKER_02So they came in afterwards that great like we're going to scoop up this internet bandwidth at like a bargain price and you know that was a huge asset that they had um you know in going into the this thing. So I think like to your points like the we should really think about like the AI stack, right? And and and recognize that Nvidia is on one tier they're making GPUs their competitors are maybe like AMD and like the TPUs that Google makes and Amazon's making one too from the chip and then there's other kind of parts of the AI supply chain on the hardware side then there's like the data center companies there's like the pure plays like Core Weave, Nebius, Iron, some of these like crypto former crypto miners, as well as like the hyperscalers like the traditional cloud businesses like AWS and you have Meta as well even though they don't have a public cloud.

SPEAKER_02And then you have the model layer so you have you know anthropic and open AI in the mix right and then you have the customers as well the folks who are like like you and me like I have this like look at this this thing here is like I use 10 billion tokens so now I sent me a gift because I'm like a customer and it's cool because like I can do that and they subsidize my costs in the same way that when you when Uber first came about like you got like rides for so cheap because like the VCs were subsidizing it. And and to the extent and maybe the plan is to eventually jack it up once it becomes an indispensable utility.

SPEAKER_02But the thing is that what happens if like they overbuild? If they overbuild then you know the prices collapse and we'll be able to get this for forever basically for the foreseeable future. So so yeah I mean I think it's very important to think about market structure and recognize that like not every business is a good business to be in. I you know maybe I'm biased because of my focus on intangible assets, but I I do think that like the data center business is a commodity business with low ROI, you know, very high capital expenditures even just to maintain these GPUs which depreciate at a pretty fast rate, anywhere between you know two and five years is the estimate for how long these things last because with the innovation cycles as once the next entity of GPU comes out, the current one becomes obsolete.

SPEAKER_02Right. And so that pace of technological progress means that you these assets useful lives are shorter and shorter over time. So I think that's just a really really tough business to be in I think the the model layer like that's interesting but you know I to my point around the kind of convergence of the frontier it's seemingly more and more competitive is what it looks like. But you know Nvidia has a great position right now you know as long as there's end demand. That's the big question. Like Nvidia's P ratio is not crazy and you know they are basically a monopoly a monopolist in their thing. They have some competitors with TPUs which you were used to train Gemini which is a pretty good model obviously but but yeah generally with the the problem with NVIDIA is that their earnings have historically been highly cyclical right like semiconductors are one of the most cyclical industries in the world and so to the extent that maybe the end demand doesn't materialize they're basically Cisco in the 90s right when you're don't don't make that analogy don't I think they scare everybody I mean I think they are I think they are um you know and I'm not I'm not saying that it won't work out.

SPEAKER_02I mean maybe maybe things may maybe not but um but yeah so I mean I what I'm most interested in, you know, personally um to kind of be more positive is the the final layer right that the users or or adopters of AI. I think that these are the companies that are like random industrial business. I think you know like like C.H. Robinson or like yeah any of these businesses that are like are operating in traditional industries but for them it's like hey if I can use AI to be 20% more efficient, that's actually a game changer.

SPEAKER_02And these companies are generally not trading at the elevated multiples that the kind of more infrastructure AI names are. And so I kind of feel like that's the interesting thing today. And if you go back to the dot com boom, you know, with the benefit of hindsight if you had a crystal ball and said how do I want to play the dot com boom what you would have done is you would have bought Cisco and you would have bought um the telecoms in the in the 90s when the euphoria was at its max and then at some point you would have said you know what like the price to sales ratio of 30 is a little bit high. Why don't I take some profits off the table and then rotate into other names and so at that point you have two options.

SPEAKER_02You can either rotate into companies that are just like anti-internet, like just the most like boring kind of like businesses that are just not related to the internet, but then you miss out on the internet revolution. Then you're not exposed to the tailwind of innovation. So instead there's a middle category which is the one I'm interested in like the internet adopters of the nine of the late 90s 2000s that weren't trading at the Cisco level multiples, but you know were actually implementing these these um these things, right? These, these, these tools to their benefit um to kind of edge out their competitors in these old economy industries.

SPEAKER_02So I think you know the way I think about where we are now in the AI cycle is kind of similar where you know it's been three years now since ChatGBT was released. Right. So we're we're already kind of like a few years into the game here. And yeah, you know, there's only been whatever like hundreds of billions of dollars of spending with which much much more set to be to come. But I think at some point you need to start thinking about how do you take ships off the table and kind of find the next leg of the trade companies that um you know are maintaining exposure to the AI and stand to benefit from AI, AI adoption, but maybe aren't again trading at those premium multiples.

SPEAKER_00Yeah. One thing that um always fascinates me is the theory that we're gonna see the one man billion dollar company soon. And I think that's probably going to come sooner than we think with AI. And even just as a business wheeler and I were talking about this the other day, the amount of efficiencies that have been improved in our own business just with AI has been outrageous. So to your point it's almost like who's gonna benefit the most from it but isn't heavily invested in it.

SPEAKER_01Well that's also like some good human capital there. Like if you can be that$1 billion guy, like you know what you're doing. Like you are special. It's not just AI that's that's exactly right.

SPEAKER_02You know in our industry I used to work for GMO as I mentioned we had 500 employees of which a significant number in the hundreds were IT people you know they were managing operations. A lot of this stuff is now outsourced right or goes to the cloud. A lot technology in general provides more leverage for talent. And so to your point about the the the the$1 billion the unicorn company with one employee, like that's the idea, which is as technology becomes as we go from like the cloud to AI and you get better and better tools, it it opens up the possibility for a single talent to achieve like great things without having to you know build out a whole team of IT people or you know build out their own servers and things like that.

SPEAKER_02You know, their own own analyst bench because in theory you could argue that AI could has the potential to become like an a financial analyst and so you can kind of leverage that um rather than hire a bunch of um you know 25 year olds. So I mean that's just the these college kids man. No, I mean I was I was once 25 too.

SPEAKER_01I wish I still was I know right and we made it we made it. Well you know Kai you're not alone I was listening to Invest like the best the other day and they had uh Henry Ellenbogan on and he was talking about the same thing about how traditional media companies in the 90s you know gave way to the Netflix of the world right and that's one of our favorite topics anyway is like the blockbuster to Netflix situation. You know, moving into 2026, moving into the next wave, I'm thinking about AI in a couple of different ways because I feel like we spent a lot of time talking about how AI can be disruptive in 2025.

SPEAKER_01I think 2026 thematically is going to be how AI is incorporated, right? How AI is actually improving our lives on a daily basis. And maybe that's creating the opportunity in the layer that you're most interested in. Can you share some thoughts on like where we're going next yeah I mean I I think that is true.

SPEAKER_02I think that's the ultimate test right and and remember the GPU useful life is only two to five years. So we have a five year window let's call it whereby we need to prove out that AI can truly lead to gains for traditional businesses for these adopters. And I and I think like I can speak from personal experience AI is tremendously useful. Again I'm a small business I'm not an enterprise and so like can the enterprise users truly adopt this in a meaningful way that is the test. I think you're right the 26, 27, 28 even, that's going to be the big thing.

SPEAKER_02If the answer is no, then the the AI infrastructure players, the magnificent seven of the world stands to lose a lot. If the answer is yes, then they'll probably do well but perhaps in that state the other companies the adopters will even better right so like I mean that that's kind of how we're we're thinking about things like if you step back like as an investor the big theme the big you know force in markets has just been the dominance at the Mag 7 right that like as an investor if you are overweight mag 7 over the past decade you beat all your peers. If you're underweight you underperform right that's just the one thing that has mattered right and it's like you can go international you can go small cap, but ultimately none of that matters.

SPEAKER_02All that matters is how what was your mag 7 position over the past decades. This year was interesting because if you go back the year before and the year before mag 7 had they just you know well well outperformed the market right so especially 2024 was was crazy. So if you didn't have enough then you look bad this year I mean it's pretty much in line like maybe slight outperformance but you know especially later in the year where they haven't done so well. And I think that is a bit of a like a a bellbeather for where we should be positioned right like again like I I like these businesses a few years ago when we launched our first fund like I was overweight these companies which would be a kind of weird thing to do as a uh a value investor right like Nvidia with a P ratio of 100 like why would you hold that?

SPEAKER_02But like you know our models looked at the company and said hey great culture super innovative like strong IP strong network effects around CUDA and you know we we owned it as one of our big holdings which is great. And then what's happened is that over time we've started to gradually sell down our exposure to the Mag 7 such that today, you know, we're pretty underweight and we still have Google as our largest holding which has been good for us this year. But but yeah I mean that's that's kind of the thing which is you know this this asset heavy transition which I think we led this section of the conversation about that a lot of investors still are conditioned, they have been conditioned to think of the Mag 7 stocks as the kind of ultimate monopoly businesses.

SPEAKER_02But I think what's you know become clear is that they're they're very actively turning away from that business model into a more asset heavy competitive business. You got to think about the market structure up until today has been more like all right, take every key digital service and divide it in two and you know these companies will kind of divide and conquer the world right so there'll be like you know Azure and AWS, GCP and cloud, so on and so forth. But now they're all kind of fighting over the same pie. Oracle is coming into play like the modes are kind of coming down and with higher competition, they're you know, they're less less likely to maintain that kind of historically brilliant business models.

SPEAKER_02And so the the problem is that most investors still invest in them, thinking of them as like these amazing businesses, even though they're telling us, hey, we're actually not going to do this anymore. We're doing this other thing instead which is a less good business. And so to the extent that investors have been conditioned to think of these as amazing companies that should deserve premium multiples, yet they are very actively transitioning into what is a more commoditized business, I think that does then then may end up being and I who can predict the future the kind of turning point for these names and what will ultimately lead to a broadening of the the returns of the market, which I think is important because you know as an investor, an a everyday investor when money's in an SP 500 index fund, okay, great, but like one third of your money is now in these not these companies that are spending the most aggressively on building out AI, which is like which is fine if they want to do that, but that's your money.

SPEAKER_02You're the shareholder and you know they're basically going to the casino saying we're going to bet it all on uh on black or whatever. And so maybe it works out and they you make a ton of money. What if it doesn't? Right. So I think that that's why you know diversify diversifying into other parts of the market, whether it's international or you know smaller companies or the AI early adopters intangible value companies in my case is a reasonable approach, you know, as we enter 2026 when it feels like a lot of the tailwinds that have underpinned the the the boom in these stocks um you know starting to abate.

SPEAKER_01Yeah that makes sense. I think that's a great place to leave it, Kai. Yeah. Great conversation today. Thank you for joining us. I think we are in alignment with a lot of your thinking. And I think it's it's really important. Colin and I talk about this a lot. You know the the largest companies historically in the past are not the largest companies today. And you have to assume that that is going to stay true. And it doesn't mean that you know we're we're telling everybody the sky is falling but we can't just expect that what got us here is what's going to get us there eternally.

SPEAKER_00No that's a really good point. And Kai I have a business idea for you.

unknownOkay.

SPEAKER_00Have you ever considered trading cards for CEOs like a holographic Tim Cook card. Yeah like baseball you can play them against each other. Yeah yeah exactly it's like a Madden type thing right yeah yeah exactly I don't know I feel like I feel like there's sports betting out there. Why isn't there well I guess there is it's called the stock market. But with that being said, though there has to be a way to trade that I don't know it's kind of cool. I love this idea because I mean intangible value and building the right team in my mind is everything.

SPEAKER_00Yeah.

SPEAKER_02Yeah I mean there's a paper I wrote called uh what was it like um finding superstars like finding superstar talent and I made the analogy to like money ball right it's like Billy Bean what he would do to scout uh baseball players is like look at these quantitative metrics around their talent and that's kind of what we're trying to do here too which is come up with quantitative ways to assess uh you know the skills of of various human capital I love that sweet thank you so much guy we really appreciate it yeah thanks guy this is great compound growth with Wheeler and Collin sponsored by Coby Advisors reach out today yay

Open the transcript file

Hosts and Guests

Nobody is credited on this episode yet.

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. With Nvidia Down 30%, What Should Investors Actually Worry About Before Year End?

    Nov 16, 202650 minAudio

    Nvidia is down roughly 30% from its highs after the Jensen Huang headlines, the market has swung hard since the midterms, and year end is bringing a tax bill some investors won't see coming. It's hard to tell which of...

    Credit the people on this episode

  2. The K-Shaped Economy, Bitcoin's Rough 18th Birthday, and the Slow-Go Years That Never Come

    Nov 9, 202645 minAudio

    The headlines say the economy is strong and the market is near its highs, but for a lot of people it doesn’t feel that way. Wheeler and Colin are joined by Dan to look at who this economy is actually working for....

    Credit the people on this episode

  3. Property Taxes on the Ballot, Visa and Mastercard Under Pressure, and a Bubble on Watch

    Nov 2, 202646 minAudio

    Property tax bills keep climbing, a bill in Congress promises lower card fees while card companies warn that rewards could shrink, and everyone seems to be waiting for the AI bubble to burst. It’s hard to tell where the...

    Credit the people on this episode

  4. A Call From Inside the House, the Fall of FICO, and a Retirement Plan AI Couldn't Finish

    Oct 26, 202647 minAudio

    Homes cost about five times the typical household income, the average car payment has climbed by about a third in five years, and letters from the Treasury keep arriving with help that rarely adds up to much. It can...

    Credit the people on this episode

  5. Mailbag: Is AI Truly Being Adopted or Just Distributed?

    Oct 23, 202619 minAudio

    In the race of technology, is AI truly the fastest adopted innovation in history, or is it just a case of mass distribution? If you’ve been questioning the impact of AI on the economy, your portfolio, or even just your...

    Credit the people on this episode