← Excess Returns 
Excess Returns
The Inevitability No One Sees | $11 Billion Tech Manager on What Investors Miss About AI
01:02:35
Loading summary
Transcript94 lines
- [00:02]
A
Spring fest is happening now at Lowe's. Keep the spotlight on your yard with stay green premium 2 cubic foot mulch. 5 bags for $10. Plus when you want more help indoors, get up to 40% off. Select major appliances that help you supercharge your chores. Our best lineup is here at Lowe's. Valid to 422 while supplies last selection varies by location. See lowe's.com for details. Mulch offer excludes Alaska and Hawaii.
- [00:30]
B
You think about in world where economic productivity is not capped by labor, I think that's, that's kind of inevitability. When you're seeing massive inflection, there's usually like a, a lag between the numbers believing it. Fiber laying down like you know, telecom, that was like a declining cost curve. Like as in the more you did, the cheaper it got. Here is like the opposite. You inflate cost curve like, you know, chips are becoming harder to make. The best time to own a stock is when they're missing and the stock's going up. You know, the second best time is when the stock is beating and the stock is going up.
- [01:10]
C
Tony, welcome to Excess returns. It's very nice to have you on.
- [01:14]
B
Thanks for having me. Appreciate it.
- [01:15]
C
You run the T. Rowe Price Science and Technology fund. It's a $12 billion strategy focused on identifying long term winners across the technology and global innovation landscape. So you spend your days and your time thinking about how to analyze companies and build a portfolio of companies that are benefiting from sort of long term trends and innovation. And today's conversation is going to be sort of right in this sweet spot, talking tech, investing, AI and how you think about the inevitabilities in investing. And at the end we'll sort of walk through your investment process that you go through when building portfolio, when building portfolios focused on technology and innovation and science, you know, tech is all over and all around us in our lives and obviously plays a very important role in the major market indices. So this conversation I think is one that our audience is really going to find valuable and appreciate. So appreciate you taking the time out of your day to join us in our audience. Definitely one of the, in doing research for this, you know, we always like to read what are our guests are writing about. And one of the things that you've said is I like to invest in inevitabilities and I thought that was very interesting. I thought we could start there. So when you think about finding and identifying something in investing that's inevitable, like how do you think about that and why Is that important to you?
- [02:40]
B
Yeah, well, I think that in the public markets there's always so much news flow and noise. And so I think the art is separating the noise from the signal. And one way I thought about it is just what I think the world will look like in three, five, 10 years. And what is the inevitable outcome? That is very likely. And so I grew up in semiconductors. I've been at T row for eight years covering semis. And when I started Nvidia amd, they were kind of emerging companies. And at this time, with like Moore's Law dying essentially, so we need a new compute platform to thrive for computing growth. And it's also at a time when AI was starting to be nascent, but really reflecting. So that's an example of like an inevitability of like, okay, we need more compute, better compute at a time where compute is becoming more scarce. And so to me, I thought that's inevitability. And one thing that you could bet on through all the noise and the other major cycles, obviously, you know, today I think the notability is that we're going to have agents essentially doing work that humans have been doing and it frees up our time to do more interesting work and drive productivity forward. So those are a few thoughts and. Yeah, yeah.
- [04:04]
C
So how do you. Do you have any? When you think back to those examples, like, I guess it's thinking about what you think is going to play out in the future and then trying to differentiate between what you think is inevitable between or from maybe what is just like the consensus view or in some cases, you know, these things that are in the narrative out there in the media, they're actually overhyped. So when you think about trying to differentiate those along that spectrum, along that line, like, how do you kind of work through that process in your own mind?
- [04:39]
B
Yeah, I mean, there's definitely a lot of in the market, in tech cycles. And I think that part of it is like, what is the big secular trend that's happening? Right. And then also then you have to invest behind companies with durable competitive advantages that are well positioned to capture that. So, you know, if I take the example, Nvidia, you know, I think the controversy back then was, you know, how is this durable? And the insight, I think, was that they have a lot of domain libraries in Cuda. There's not just a chip but a software, and there's a lot of software layers and an ecosystem really to build on. And so to me, I think they, you know, comparing to old fans going in various forms. But I think part of it is like just having a deep understanding of what really matters and then just thinking about, like, can we be positioned in a way that you capture that upside if that, you know, kind of more optimistic bull case plays out?
- [05:42]
C
And talk about, so how does this inevitability in your mind play into sort of AI and what maybe the, you know, in your mind the future of AI looks like? I think you mentioned agents, so you sort of shake that out.
- [05:55]
B
Yeah. So I think that's, you know, if I take a step back, like what, what, what is this kind of breakthrough that we're kind of seeing is that, you know, you have the ability to drop the cost of intelligence to pretty much zero. I think if you think about it, right, like, cognition is becoming really plentiful. I think that it brings up the average to be very high. And as a result, I think you take a look at, like, different parts of tech, right? Like software, for example. If you think about it, software has essentially been designed for humans to use, right? Like driving, clicking, typing in operating software. And I think we're going to have a shift in terms of how we're seeing, like, agents, software be designed for agents. And I think that's a pretty big unlock. And you think about what companies and enterprises have been doing is that, David, spending on software and enabling humans to be more productive. And that cloud mission is like you paying probably somebody to input data into a piece of software. And that's the cognition piece. But now that's compressing the software token now. So I think that now to get a similar intellectual cognition, you pretty much ping Azure for anthropic model or OpenAI model. And I think that's going to be probably what's going to happen. And just think about your everyday life. How many times you call a restaurant, they can't pick up because they don't have enough people. So an agent now has been, I've been calling restaurants and increasingly they have agents picking up the phone. So I think that's an interesting, just a small little, little sliver of what it could be. But I think I'm more optimistic that, you know, it unlocks the economic opportunity of like, labor, you know, plus, plus productivity equals economic output.
- [08:01]
C
Now, the one thing you said that I might have to disagree with you on is you said that it brings the cost of intelligence down to zero. But I'm paying 60 bucks a month because I have chat GPT premium, I have Claude Premium, and I have Gemini. So. But to your point, I mean, the level of intelligence that even one of those models, like offers for effectively 20 bucks a month is basically ridiculous. You know, so I'm kind of being facetious here, but you know what I'm saying.
- [08:28]
B
Yeah, it costs. The intelligence go to zero. Yeah, that's probably a little hypercolic as well. I mean, I think if you listen to some. Some folks, you know, cost of intelligence going to the cost of, like, electricity or, you know, compute. But I think it's. You're going to see deflationary forces, and I think you can do more with fewer. With fewer people, and the people will be more productive. Yeah, I guess that's where I was going to think about. In a world where economic productivity is not capped by labor, I think that's kind of inevitability.
- [09:03]
D
It's funny, too, on your point about the restaurant. I called the auto dealer the other day to schedule service, and I was 100% talking to an AI the entire time. It didn't say anything about it, but, like, it was 90% of the way there. Like, he was responding to me like a person. It was. It was really good. I mean, these things have come a long way.
- [09:18]
B
Yeah. I think that customer service, for example, I mean, you're seeing various companies, like, adopt this and embrace it. Right. It's a different way to think about it. But I think a lot of times, like, people kind of prefer instant responses that are 90% there, and it can be elevated, you know, to. To a real person, and that real person can do more interesting work, you know, on the customer service front.
- [09:38]
C
Now, here's the question with the agent and this inevitability, do you think that the market has fully recognized that, or do you think that that is maybe still underappreciated by investors? Or do you think that. That the range of, you know, outcomes, I guess, for different companies is.
- [10:01]
B
Is.
- [10:02]
C
Is. Is so variable because it's so wide open. Like, I'm just wondering, like, how do you think about when you're thinking about investing in technology names that way, you know, how do you know if that is being actually underappreciated in the market or if the markets may be realizing it or recognizing it? How do you think about that?
- [10:20]
B
Yeah, I mean, we. I mean, that's the whole. That's the whole art of the. The craft. Right. And so we as a team do a lot of quantitative analysis, also sentiment analysis. We look at valuation, we look at technicals, we look at fundamentals. And so I think it's a multifaceted approach, but I think there are signals, you know, like here, like Nvidia trades at, you know, sub 20 times earnings, you know, for the growth, and then Micron's trading out four or five times earnings and you're, you're seeing like tremendous demand and number revisions up and people aren't really buying it because the market has never seen anything like it. You know, semi investors and market in general, when you're seeing massive inflection, there's usually like a, a lag between the numbers, believing it to some extent. And so here I, I do think there's a lot of skepticism. These Playbub in Sandy is essentially like when things are, sentiment's high, you know, the backlog is strong, gross margins off, you want to sell, you know. But I do think this, like, I think, you know, that's one lens to look at is like, yeah, we want to manage risk, but also like, what is the, what is the first principles? Like kind of driver of. The driver is what I like to talk about to the team is like, okay, like what is driving the actual driver? Right. And I think agentic does change things because you have this like autonomous agent that we've never seen before. You know, it's like it can be always on. It's always like thinking for you preemptively. And I've been using Complexity Computer, I've been looking at using Claudic, what's going on with a clawbot. And I mean it's pretty amazing what's going on. And I think that just looking at the valuations, the market clearly believes here that these numbers are not sustainable and it is cyclical. But at the same time, I think there's a lot of doubt just giving what we're seeing here.
- [12:24]
C
One of the things that you've talked about is describing the current AI build out as like a space race, but one with, with multiple moons. And so. And I'm kind of hoping in your answer you're going to reference like Star wars or something here and like the loops on Tatooine, whether they have the multiple galaxies. But what did you mean by sort of that and what do you think it means for investors?
- [12:46]
B
Yeah, I think that, you know, I, I was, you know, there's a, I was thinking about like when I was covering Sax and you know, there was a view of like, okay, which one's going to win like zero sum game. The TAM is like X, you know, but the TAM was actually like 10x bigger than people thought. And you look at like Nvidia amd broadcom larvell, they all did pretty well, you know, because the TAM was growing and the all having like different ways to think about, you know, delivering the demand. And so here I think like you know there's, there's you know obviously space race for large language models and intelligence but I don't think it's going to be a one player, you know, rules of all like kind of, kind of thing. And, and if you look about like their businesses, they all have like a little bit of a different flavor to what they're trying to do and they also have existing businesses that they're trying to roll out, you know, the, the intelligence too. So I think about like Google, right? I mean they have a massive search business, digital properties and then Meta has you know, Instagram and digital advertising as well. And then you have like Tesla that's doing physical AI and space and so and then you, I think you have anthropic, there's been both side enterprise and OpenAI that's got a little bit more of like a consumer end to it but also you know, looking at like, you know, their, their, their partnership with SoftBank delivering agentic intelligence in Japan for example. So I think the market's really big. There's a lot of different demand, demand TAMs I guess. And I think it's not just like a zero sum game and I think the market probably is bigger than people, people expect because it is like anytime there's a new market people always like equate it to like an existing one generally and it's generally bigger because of the productivity enhancements.
- [14:40]
D
Back to the inevitability is if you look at inevitability is right now, I mean are most of them all related to AI? Are you seeing things that you think are inevitabilities that you want to position your portfolio for that are outside of the AI space?
- [14:51]
B
I think that AI is generally the dominant driver just because I think you have like multiple like tech waves, right? We had like you know, the PC, the handset, the Internet and now I think do you think AI is kind of like the primary driver? But there's definitely like different ways. I mean here like you know we have like agentic AI and then also like you know, physical AI coming up. And I think it's, I do think it's probably the primary driver. I mean you look at the market, right? Like it has been like a K shaped distribution. Are you an AI winner and AI loser? So I do think it continues to probably be like that. And so yeah, I Do think that's probably the biggest one. Of course there's other ones that there's stocks that have idiosyncratic outcomes, but AI definitely is like what every company has to think about positioning for.
- [15:54]
D
I know it's not your job to be an economist, but I did want to ask at a high level since you know the technology really well. Like, we've had a lot of debate recently about what this means for the economy. Like on one side you've got the tech guys who are saying like, we're, we're in for a world of abundance, which sounds really great if we can get to that. And on the other side, we had this Citrini piece. I don't know if you read it or not, but talking about this idea that AI could create a lot of job loss, at least initially. I'm just wondering, do you have any thoughts on that, that balance between those two things?
- [16:18]
B
Yeah, I don't think it's going to be quite, quite as extreme or, you know, one way or the other entirely or centrist on this one. I, I do think there's going to be a lot of productivity improvements. I think companies can do more and there's going to be a real kind of retraining of the workforce. Luckily, I, I do think that technology is becoming easier to use instead of harder. And if you think about agents, like, right. Like a few years ago you had to be like pretty deep in like programming to get what needs to work and now they're off the she. Like, you know, you could kind of just deploy one. And so in some ways, like it's all natural language, right? So it's all English. English is the new programming language. And so, and it's becoming easier every six months. And so I think that, you know, previously, you know, if we're to start a tech company, we had to get a bunch of money, hire a bunch of engineers. Like, you know, I don't think the barriers are that high anymore. So I think if results in like the best ideal wins and you can be more nimble. So I think it's going to create a ton of innovation and I think we're seeing that in the private market. No, I, I think that it's not going to be exactly even thing because you're seeing a lot of like, you know, electricians getting paid $2 million right now, right, to build data centers. And so there's going to be kind of refactoring probably of the job market. But then it's like, you know, you look at every Tech transition, it's been pretty positive for humanity. And so, you know, just because ATMs happen, you know, all the cash, all the people at the ATM teller, right. They become wealth managers. And so I think this is going to be more similar to that than different.
- [18:09]
D
So, yeah, to your point, I'm, I'm probably an okay programmer at best, but like, what I could do with Cursor and Claude code is insane. Like, it's just taking, it's taking anybody. If you can't code at all, it's teaching you how to code or it's not teaching you, it's letting you code. Like, if you're someone like me, it's elevating you a lot. It's pretty amazing to think about what this could be.
- [18:27]
B
Yeah, I totally agree. And I was talking to my friend who's a software senior software programmer, and I think we were having dinner with some other folks and we're like, okay, what do you do in this agentic world if like you have so much free time? And like, some people said travel, some people said, you know, art and my software program friends, like, I would write more code and do more agentic stuff because you see such crazy productivity. He's like managing all these agents already. He's like, I will do more work because I could do. So I think companies that decide to kind of maybe wholesale cut programmers or cut headcount, maybe that's the right move in the short term, but long term, it's really like embracing the technology and doing more with it. And I've seen my workload actually increase. Like, I, you know, it's amazing. I use Claude and I use perplexity computer using OpenAI and it's just like the level of insight you have, like, and, and you know, the depth you can go and the sophistication you can have is incredible. Right? Like, so I actually work more with these tools than less, you know.
- [19:36]
D
Yeah. And then one of the challenges you, when you look forward, like if you look at past innovations like the Internet, like, you couldn't see when the disruption was happening, you couldn't see what the jobs were going to be on the other side because they don't, they didn't really exist yet. And it's the same thing here. Like, it's hard to think about, like, what the new jobs that are going to be created are, because they're not there yet. But, but the worry is that this is such, this is intelligence, this is so advanced, it might be harder to create those jobs because this can do Way more than past technologies could do.
- [20:02]
B
Yeah, I mean, I think that's fair for sure. Like, it is probably a bigger, you know, much bigger breakthrough, but I think it's just going to be more productive. Like, people are going to be able to do more interesting work. And yeah, I do think it's going to create a lot of jobs. Like, a lot of different jobs, though. A lot of better jobs. Like, you know, I think a lot of jobs, you know, quite frankly, people probably don't want to. You know, we've all seen office space, right? Like, you know, I think, like, there's probably better jobs out there that don't require you to show up in office and type things into a computer. I mean, there used to be people like that made a good living from being just being able to type really quickly. Numbers that were handed to them at the bank teller, they go back and type really fast and make probably equivalent below $150,000 now. But I think this is an evolution of what humans can do and more interesting work and creativity and judgment. Creativity and judgment will always be very important, Right. Like, these models. Like, you can ask the same question two different ways and get, like, very different responses. Right? Like, and so there's a. There's a ton of that. I think that's going to be more important is like, judgment and creativity. USAA knows dynamic duos can save the day. Like superheroes and sidekicks or auto and home insurance.
- [21:28]
A
With usaa, you can bundle your.
- [21:30]
B
Your auto and home and save up to 10%.
- [21:32]
A
Tap the banner to learn more and
- [21:33]
B
get a'@usaa.com bundle restrictions apply.
- [21:37]
E
It's tax season. And at Lifelock, we know you're tired of numbers, but here's a big one you need to hear. Billions. That's the amount of money and refunds the IRS has flagged for possible identity fraud. Now here's another big number. 100 million. That's how many data points Lifelock monitors every second. If your identity is stolen, we'll fix it, guaranteed. One last big number. Save up to 40% your first year. Visit lifelock.com podcast for the threats you can't control. Terms apply.
- [22:09]
D
Yeah, one of our, one of our past guests pointed out to us that both computer and calculator were human jobs at one point, which kind of shows like you would have never seen what ended up coming. Coming. I want to ask about the physical, physical part of this, because I know you've looked at this a lot, and this is probably the most exciting thing is the humanoid robots and all the stuff that's eventually coming down the road. So like, can you just talk about what you see coming there and maybe talk about like, is if you see ways to invest in that now or maybe it's too far in the future.
- [22:39]
B
Yeah, it's definitely an area that we're watching. I mean I've been to kind of robotic warehouses where they're trained the robots. Super interesting. Like, you know, you can see the future in that. And I think it's like one of those things where it's kind of like an S curve. You know, it's not that great initially and then like it starts getting, then it gets really good and you're like, oh wow. And so like, I think, you know, there's, there's a, it's a very competitive space. So there's a lot of companies in robotics, but I do think the compute layer is going to be the backbone that a lot of this stuff runs on. And you know, these companies, like, in terms of like Catherine being able to like, you know, be scarce, in terms of how many companies can do them, like the scale that's needed, like the, I actually think like, you know, computer chips are like the new oil to AI, right? And specifically you want to get the bottlenecks of like, you know, it's increasingly more memory bound. And so the hbm, the memory, the objects, the networking that is required. You know, to me that is, these are like robotics, they're going to exchange a new cam of that and like at the same time, like, you know, you can pick out the right winner in the application layer, the infrastructure, I'm sorry, the, the robotics layer. Like, you know, there's going to be a lot of very innovative companies and, and probably where you get like, you know, 10 to 100x returns.
- [24:08]
D
Ask you about the whole AI CapEx spend right now because that's something we've been debating a lot on the podcast. You know, we have some optimists who think, you know, there's so much demand for this stuff and this is such an innovative technology that all of the spend is completely justified. And then you have people to look back at past innovation cycles and say, you know, the companies that built out the infrastructure didn't do very well. Like the return, the return for society was great, but the return for the builders was not. Like, how are you thinking about this Capex buildout and putting it in context?
- [24:33]
B
Yeah, I think that's a great question. I mean you look at like the 2000 tech boom and bust there. Like, I think it Was a very different setup. Like one is that when they're laying the fiber down like utilization was never more than 20%. It was really a build it and hope they come and then. So that's number one. Like GPUs right now are at max capacity. Like you have a GPU, you have a business essentially you got some GPUs you can string together, you can sell them out like for immediately Even the old GPUs, the pricing, you know, has actually held up really well and you don't, you're not seeing declines there which is like pretty, pretty wild, right? Like a 10 year old GPU is still like renting out for similar amounts. And so one is utilization. Two is like a cost curve, you know, fiber laying down like you know, telecom. That was like a declining cost curve. Like as in the more you did, the cheaper it got. Here is like the opposite, you know, inflating cost curve. Like you know, chips are becoming harder to make. HVM is three times the wafers as a normal dram. And so the fact that we have like you know, very rational supply in terms of like it's really expensive and really, really expensive and these tech jumps are not easy. You know, that's where I think we have a different setup. It's like, it's like oil. You know, if oil is super easy to drill, everyone can do it. That isn't going to be too much. But if oil's super far to drill, that is going to be like pretty, pretty rational and there's not that many players, it's going to be scarce. So I think of it as like the abundance of compute is becoming more scarce in some ways versus what it was like you know, kind of the fiber build out in 2000. Does that make sense?
- [26:21]
D
Yes. Yeah it does. Yes. You would leans on the side of the interview if you talk about, listen to someone like Jensen, he's telling you like the demand is, is really really strong for a really really long time. You know what, what people on the opposite side will say was like fiverr. Builders would have told you that back in the day too but it does seem like it's more real this time. Do you, do you agree?
- [26:38]
B
I mean you are seeing anytime data centers go up they immediately get utilized. So I do think the utilization is a big point. These are companies that I think a lot of the demand is funded by free cash flow versus you know, kind of what it was in tech bubble where you know, company like essentially have a dot com behind it and then like you know, no customers like this is, this feels a lot more real. And I think that you are seeing the, we're all using ChatGPT and all these like large language models, right? And, and it is, you could see it, you can use it versus like the Internet was like very much so tbd I think in terms of what they've had at the time, how do
- [27:25]
D
you think about like working your way down the stack as this gets further? So for instance like early in these cycles you typically want to be involved with the builders. So like the semiconductors which I know you have been involved with, do you think like we're still in that early stage where those are the types of companies you want to be investing in, the play in or do you think we're starting to move down to maybe the beneficiaries a little bit more?
- [27:43]
B
Yeah, I think that, you know, there's been this whole debate of like semis versus software, right. And when does it go to the application layer? The tricky thing is that like you have new income, new players come into the application layer in the form of large language models and they're, what they're doing is like very different than what the existing software companies are doing. And so there's a real innovator as a lapa essentially. Software know, for example is like you essentially use, it's designed for humans versus this is like agentic. And so can you make that transition? And so I think it's more important that you're the platform and not just a feature, you're a feature. Like you know, the cost of intelligence, the cost of coding, the cost of software creation is like plummeting, right? And so you, you have to have like data, data gravity, you have to be entrenched. There has to be regular regulatory probably. So it's, it just makes it so that I think where it shows up in the application layer can be look very different and then like the unit economics, right, is just super different now in terms of what existing application software companies are used to charging, right. They're used to charging per person, but now you have agents that it's really hard to charge and you have to charge outcomes based perhaps and that's like probably inflationary that the customers expect that to be passed on to themselves. And you think about like, you know, Adobe, you know, it's an interesting debate because like you know they're, they essentially, you know, have like nearly like they're the dominant platform for people creating content, right? And, but like you know, if you use it up, chatgpt you can essentially create, generate images for unlimited images for 20 bucks a month. Right. Whereas like there's innovators dilemma with I think Adobe where they have to essentially like charge button token, you know, so, so you're so it's just like a really hard, I think and it's not that they can't make this transition but it just like then the value creation could come from a different set of companies that have a different set of unit economics that they, that they need to adhere to.
- [29:59]
D
I'm curious just since you mentioned software, do you think the whole idea that AI is going to disrupt software, do you think that's being overblown right now? I mean obviously the software companies have been, many of them have been killed by this. Like do you think that's overblown or do you think it's a case by case basis? Like how do you look at software in a world?
- [30:14]
B
I think there's definitely, it can be overblown in the short term but it does feel like a lot of there will be companies that make the transition. But they're also, I think the disruption risk is very real and probably most software companies because essentially I think there's a few different things. I think the first kind of step down evaluation is probably just like being a little bit too overvalued from 210 interest rates. And then these COP companies are also a lot more mature up the gas curve. So when I think about investing I also think about where we invest from adoption. Right. And so software at the world for a good 20 years and so clearly it's just like they're more mature and now we have this terminal value question of what do agents do, what does AI do? I think now you also have companies that across the enterprise that were used to just paying more for software every year. Maybe it's like oh this is another year we're going to pay 10% more probably because we're hiring more people. We're just used to it cost inflation and, and that was just like spot for it. And so but now you have this new disruptive technology that you have to invest behind. So now it's like oh like we need to really save here. So and we're also you know, probably over hire in many instances. So we've got to like rationalize. So, so I think we're seeing some of that like is it a macro? But I do think that you know the software companies, they're not going to go away but it's just like are they going to be relative Winners within and lead technology, you know, the technology index over the next 10 years. And I think most of that is just like they are more mature as a, as a segment and then I think you've got just new way of doing things, of delivering intelligence when you think about software, right, it's really a conduit to deliver intelligence in action, right? So now we have agents that are, you know, driven by large language models that are doing that and so, and at a much lower cost, I think.
- [32:28]
D
And it also seems like the cost of the software being wrong is a huge thing here. Like so for instance, I'm a registered investment advisor. Like I'm not going to vibe code by CRM or something because I've got, I need a good company with good compliance like behind that thing. So it seems like it depends on that a lot as well. You know, when, when the, the 0.1% mistake is catastrophic, then you're probably less likely to, you know, have AI disrupt that company.
- [32:50]
B
You're absolutely right. So I think that's why a lot of existing software companies, they'll still exist, they'll still be important. But I think that in the market it's more about who the relative leaders are and the winners are. And I think about rings in terms of disruption, right? You kind of want to invest in the central moats that have, that are mission critical, that are the platform that agents can actually sit on and they can essentially be the platform for everything. So I think that's, we doubly hold software companies. We believe in the ones that have those things that are important. But I think there's going to be a lot of disruptions in the space though.
- [33:39]
D
You mentioned the cyclical playbook for semis back from when you were an analyst. And I'm wondering, does this turn that on its head AI? Because you know, you might argue we're getting to a point where cyclically you'd want to sell semis, but we've got this massive, massive trend behind them. Like do you think it changes the cyclicality of them or at least for a period of time?
- [33:55]
B
I think that, you know, we've seen kind of more predictable cycles within semis up until prime this one, you know, you kind of saw like isms go up semis peak like you're buying when the isms are down. It kind of worked like for a few cycles like that where we had like more garden variety cycles and like AI was emerging but it wasn't the dominant driving kind of factor. And so I think it does and it's like a super ness, I think cycle because you previously had like everything kind of working at the same time. China, the US geographically were working at the same time. And Covid, kind of post Covid, it kind of turns on its head and then also you have like ChatGPT get launched and then you had like a AI kind of GPU shortage and then you had networking shortage, then you had memory. And so it's like, it's, it's very different. It's, it's definitely a cycle that is like unlike anything we've seen. And so I, I think it does, I think that's why it's more important to probably you get these like market sell offs that people are looking at past cycles at, right? They're, they're comparing to maybe 21 and they're comparing to like the, the tech bubble. So I think that you know, you definitely want to pay attention to that, but I know you also want to pay attention to like what the driver. The driver is and how sustainable that is.
- [35:22]
D
Another topic we've been talking about a lot is the changing nature of the mag 7. I mean if, if you look back, I mean these were close to perfect businesses. I mean they were very capital light businesses, they were high free cash flow businesses and now they're spending a ton of money. And I think all of us are having trouble like putting into context how to think about that. Like are these permanently changed businesses the greatest things about them not present anymore? Like how are you thinking through that idea and the changing nature of these companies?
- [35:44]
B
Yeah, I mean it's fascinating because the Mac sector were definitely like kind of viewed as defensive balanced names. But now that they've spent all their free cash flow on a foreign investment, I think that the market is kind of dinging them a little bit for that. But I think it was like kind of like leverage, you know, like you spend a lot and it works. It's awesome for equity prices. Right. But if, if it slips, then you get punished pretty hard. And so to me I, I think is like a dramatic change in their business. Like, and I think that, but at the same time it's probably needed to have that next chapter of growth. I think they would be all growing much slower without, without AI. Right. And so I think they're big bets for all these companies clearly. But I do think they're probably seeing the demand signals and investing behind it. And I do think like the cost of not investing is probably greater than the cost of over investing. And so yeah, I think that it does change it. But I think it is either going to work out really well or really bad. So I think, you know. Yeah, and one of, one of the
- [37:02]
D
things you mentioned is they're, they're not trading the same anymore. I mean I believe two out of the seven outperformed last year. So the market is starting to differentiate them more between each other.
- [37:10]
B
Yeah. And I think they, there's definitely a lot more dispersion to max 7 and, and so I think you've got like, you know, Google and Meta. Like, I mean Google has seen like greatly rating off the bottom right. And just seeing as the durable kind of compounder and I think Meta also, you know, they've been accelerating their business with AI so that's been great. I think Microsoft has been kind of like, you know, in the software bucket. Right. So when the software sector sells off, they also kind of go with it a little bit more. So you know, I think of that as been like kind of a differentiate and then Apple, right, like they're not doing the space brace for AI so much in terms of spending their capex. So I think that is kind of viewed as more like the balance, the one that can still participate through their ecosystem. And then Nvidia and Broadcom are probably a different bucket of like the, you know, AI infrastructure builders.
- [38:10]
D
Do you have any thoughts on if this helps or hurts them relative to everyone else? Like on one hand I could argue they've got the money, they're spending all the money in AI, this really helps them. But on the other hand, like someone can start a billion dollar company from their basement or something now. So you could argue maybe it hurts them relative to what other companies can do. Do you have any thoughts on that?
- [38:24]
F
This episode is brought to you by Indeed. Stop waiting around for the perfect candidate. Instead, use Indeed sponsored Jobs to find the right people with the right skills fast. It's a simple way to make sure your listing is the first candidate. C According to Indeed data, Sponsored jobs have four times more applicants than than non sponsored jobs. So go build your dream team today with Indeed. Get a $75 sponsored job credit at Indeed.com podcast. Terms and conditions apply. K Pop Demon Hunters, Haja Boy's Breakfast Meal and Hunt Tricks Meal have just dropped at McDonald's. They're calling this a battle for the fans. What do you say to that, Rumi? It's not a battle. So glad the Saja boys could take breakfast and give our meal the rest of the day.
- [39:06]
A
It is an honor to share.
- [39:08]
F
No, it's our honor.
- [39:10]
B
It is our larger honor.
- [39:11]
D
No, really, stop.
- [39:13]
B
You can really feel the respect in this battle.
- [39:17]
F
Pick a meal to pick a side
- [39:20]
E
and participate in McDonald's while supplies last.
- [39:24]
B
Yeah, I think that their businesses like part of AI has been like, you know, it's a big investment and so it's like two, two fold. I think what's the, what's the benefit of owning the intelligence and then what's the benefit of amortizing that across your business essentially? So when you have a big business like you can actually get tremendous leverage on it and pay that and then you also own the intelligence. I think it's definitely the right move. But does it mean that, you know, other companies won't come along and also be more competitive? But I think then that these are, I think these AI companies are kind of finding other areas and being a different use case, even perplexity. Right. Like they kind of evolved from going at Google kind of head to head right into doing more agentic and other types of searches. So to me I think it is, it's becoming more heterogeneous in terms of the competition. And if you can, it's more important to find a really good domain that is a new market than to probably be attacking a Mag 7 on their own turf.
- [40:42]
D
That just made me think, do you think in terms of the AI cycle and I don't know how valuable this is, but you always hear people saying, are we in 1995 or are we in 1999? Do you have any thoughts on that? If we're at the early part of this whole thing or if we're of in the middle, do you, do you have any thoughts on where we are?
- [40:58]
B
Yeah, I mean, I think it's. Well, there's, there's probably two questions like where are we in the stock market cycle of those and where are we in the fundamental. Right. So the market's always like ahead by a year, I think. And so, you know, it's really. We're definitely not at the very beginning anymore, but I don't think we're at the very end. So I think we're somewhere in between where we are. We've been investing right, you know, for this forward roi and I think right now we're seeing that ROI kind of come through. You know, I think token costs have been plummeting. You know, arguably the cost of token and the output it can make is probably coming to a point where it's, it's economically self sustaining. And so I think we're transitioning from the investment phase to the inferencing deployment. Prove it out. And, and I think when you see that, I do think that that means that we're probably not like at the top. Right. Because we're just seeing the inflection in terms of the inferencing deployment, which we've been talking about for a while. Right. I remember when I first started covering semis, it was all like, oh, training's done, we're going to be solid inferencing. That was like six years ago and a few trillion dollars of market fat before. Right. So I think this is probably in the alarm cycle. I don't think it's, you know, I kind of feel like, you know, we have agentic, we have physical AI. I think those are like, that's like a decade long kind of evolution, but doesn't mean we won't have digestion and cyclicality and macro shocks that can like kind of, you know, cause things to grow like below trend though.
- [42:45]
D
We're kind of portfolio management nerds here. So we always love when we're talking to a portfolios manager to dig into the process. And I want to start at the top level. Like there's so many different companies you could invest in. Like, how do you think about just generating a list of ideas for you to consider? I mean, do you use screening? Do you just use a lot of, you know, do you have a lot of analysts who are kind of bringing stuff to you? Like, how do you think about the top of that idea generation process?
- [43:06]
B
Yeah, I think the top is, you know, when I think about it is like one is like, where are we in the S curve? Adoption. Right. Like, and so generally for most companies there's the flat part and then there's like the steepening part and then there's compounding part and then there's like the mature part. And so I think I try to fill most of my ideas with the companies that are inflecting and then also compounding. And then I'm also watching the companies that are kind of on the flyer part. Those are emerging tech right off of the private space and that. And then I'm also thinking about the companies that are like mature but maybe have a, have another chapter that those are like more value companies. And I think that helps Dallas. You know, like, if you only have kind of like bleeding edge growth, you get these massive drawdowns. And so what I'm trying to do is like, you know, compound through the cycle for our clients, like, you know, catch the most of the upside while avoiding like, you know, Massive drawdowns. And so, you know, so I think that's like one framework I use. And just thinking about like you know, over five years, is this company going to be you know, essentially like a outperformer on fundamentals? Like and so for some companies, like it's very obvious that they won't and so I probably just don't spend that much time on it. And then I think about like capturing the economic bottleness. Like is this like a company that is delivering value and can they capture it because they're like a bottleneck in terms of the limiting factor for this, this new technology. And I think, I also think about the cycle like so I was a semiconductor analyst and so I think it's, I think about cyclicality because you know, depending on where we are in the product cycle, economic cycle, semicycle, like, you know, it really matters to essentially risk manage because if you don't you can suffer like massive drawdown. So I would say like I'm cycle aware also and it's a key tenant of risk management as well.
- [45:09]
C
Would you be able to give us a example of something that like recently kind of came into the portfolio and sort of like the sort of that walk us through like the process for that individual position. Is that something you are in a position to do?
- [45:26]
B
Yeah, I think, you know, with, with memory, for example, like in semicap, you know, those have been ways that we've been increasing over the last year. I think the you know, insight there was essentially like they weren't, that they weren't the economic bottleneck in the first phase of ChatGPT because they just come out of like a semi cycle. But are they extremely important companies? Yes. You know, are, do we see like them as being part of like an early part of S perv adoption? Like. Yeah. And so I think that when you did the math on HBM in terms of how much the trade ratio is or how much like supply stuff's up, it's like three times more than traditional dram. And it also has ripple effects because if you're not going to put in dram, if you have DRAM capacity that's getting like soaked up by hbm, you're going to have less NAND capacity too. So that creates like a ripple effect and then also you're going to have more, you're going to have like less NAND storage that you need more, then as a result you're going to be short HD's as well. You know, that's something that we, we were thinking about and you know, you see this kind of rolling bottleneck happening when we are standing up so much AI infrastructure. And so when we saw like the wafer utilization in memory, like increase that also then makes you want to then also creates like more numbers, inflection cap. Because it's all about the incremental supply. Right. So those companies were like second wave companies in post kind of chatgpt that I would say, you know, benefited because that incremental demand was there and the wave capacity was fully utilized. I think that's probably how we thought about right side of AI. It was typically kind of depressed and early broadcast for economic bottleneck and then kind of the cycle where, okay, it's not top of the cycle yet. And so it could be a good bet.
- [47:36]
C
What name in the portfolio do you think most people wouldn't know very much about? I'm looking at your top 10 holdings. I mean, it's kind of like, you know, a lot of them are the, you know, Nvidia's in here, Apple, but there's a few that kind of stand out that I don't, I don't know anything about this business. Like Luminum holdings, it's a hardware, electronic equipment components, I guess maker. So like what, you know, talk that, that's interesting to me. That's a name I've never heard of. You have, you have a 4% position in it. Like what do they do? What gets you excited about that company?
- [48:06]
B
Yeah, Lumenta. I mean, lasers kind of are like kind of the hardest things to make. And so it's, they've been through a period consolidation. You know, there's, it's usually historically served telecom and now like the data center. And AI is like their new opportunity here where you essentially need to transition from, you know, copper networking to optical. And so what's happening is that when you're like putting these, get these GPUs together, they're getting really hot, but copper's really inexpensive and they keep innovating. And so, you know, here, like when we cross into Vera Rubin for Nidia, you're seeing that, you know, the adoption of optical and lasers is going to like increase dramatically. And so to me, like that's another like earlier S Curve stock where you know, you have, you know, adoption and content increases and they're becoming the economic bottleneck. And Nvidia recently just invested in the company, which is another like kind of stamp approval that this is a critical technology that is needed to enable AI. You know, to me, like, that's probably one that you are seeing a separate shift in the importance of what they do. And what they do is like super hard. And there's only a handful of companies that can do it.
- [49:31]
C
Yeah, interesting. How do you think about, I guess position sizing and concentration? I mean I think it's a pretty, a pretty concentrated fund relative to you know, many funds out there. But you know, in terms of position sizing and concentration, sort of. How do you think about that when constructing the portfolio?
- [49:50]
B
Yeah, I think about it as, you know, you kind of want to have most of your, most of your kind of like part of the portfolio like compounders that you know, can kind of outperform the close to the benchmark through the cycle and those kind of give you balance. And then you also want to have like probably 20, 30% portfolio that's like more emerging tech that could give you the real alpha, that could have dramatic kind of, you know, doubles, triples, 5ers and that earlier part. But you also don't want to only have those because then you're going to. If we do get into a 22 situation, like nobody escapes the cycle. Right. I mean that's like kind of the thing is that like no matter how great your company is, sometimes like there's just like factor macro. So I kind of want to be able to ride that out and have free cash flow ratios. I would say the complexion is generally kind of 60% compounders, maybe like 20, 30% these early emerging companies, that early investor and then verify growth. And then you also have I think a value kind of sleeve where these are, you know, kind of slower growing companies. But there's something interesting going on that is inflecting. There's a change in the business and that can really be helpful to kind of be more balanced.
- [51:20]
C
I think I sort of find that very appealing. It's like this sleeve based approach where you're taking a different set of, you know, variables and analysis and applying it to those companies that are in those different brands buckets. So it's not, it's not really like one overall, you know, technology investing strategy. It's really three different ones and probably even more than that at some point in time. And you know, that's intentionally being done so that you can, you know, give the investors in these funds those long term returns but trying to help them stay with things when things get dicey within some areas of the portfolio.
- [51:59]
B
Yeah, I think that's exactly right. And even though, you know, you might have like a bearish view on a sector, there's always winners within that sector that you want to create asyncratic kind of alpha from. And so I think about it also like being relatively balanced between kind of semis, Internet software might be services in it and try, try to also be idiosyncratic instead of just being a one way kind of bet on a singular theme or, or subsector.
- [52:31]
C
What is the sell discipline? And I guess it's probably going to be different like for all these, you know, different buckets if you will. And it's company specific obviously. But can you walk through, you know, how you kind of think about selling something what would cause you to sell it? And even if you want like, you know, not to highlight, this isn't about specific stock picks or anything like that. But, or, or stock sales. But you know, I think an example is always nice because it kind of, you know, is a just concrete example that investors can learn from.
- [53:02]
B
Yeah, I think, you know, there's definitely, I think my, my framework for selling is a few different things. So one is if the fundamentals weaken or if the valuation has gone to like extremes or really elevated where the fundamentals are not driving the stock price up as much. So I think about that aspect of it in that I think it's just like there's different leaders in the market at different times. Right. And so I think part of my job as portfolio manager is to recycle kind of that type of capital so that we're kind of optimizing for the efficient frontier of the IRR in the portfolio. And so I get that when for example, I would say like, you know, Netflix had like this great kind of run right from, from, from COVID from bottom of COVID up and they had pretty great numbers. I think that just like the valuation to me had been reweighted to previous highs and you, you see that, that the numbers when they beat but like the, the stock's not going up. I tend to like trim stock. You know, the best time to own a stock is when they're missing and the stock's going up. You know, the second best time is when the stock is beating and the stock is going up. That's great, that's the fun part. But when the stock is beating for multiple quarters and the stock's not going up, then I think about like, oh, are we more fully valued essentially? Or does the market know something that I don't know? And that kind of gives me more pause to be honest. And I've learned to be a little, to be humble in the market and that, you know, to Me, like, perhaps, like the risk doesn't mean I won't still hold the stock. So the other thing is that I think about trimming and, and buying, like, being more incremental. You know, I'm not. It's like it's okay to kind of let some go.
- [55:03]
D
You.
- [55:04]
B
You're not going to time the top perfectly, but you just want to avoid these big drawdowns. Because the other thing is that if, let's say you have in. The hardest thing about investing as a portfolio manager is like, when you're winning, it feels so good. Right. You don't want to sell. It's like being hot, you know, if it's a big boom and you're like. But like, the thing is that you, you keep your, your chip stack keeps going up, your position bets keeps going up, and the drawdown math is brutal. Like, so I, I just try to think about it from that perspective as well, in terms of kind of what the drawdown can be like and just, just kind of being a little humble, I think, as the stock's working.
- [55:41]
C
Yeah. So it sounds like there's a little bit. You didn't say it, but there's a little bit of a momentum slash behavior, like if the stock's not reacting to sort of certain news because maybe it's got this gravity, it's got this pull from a valuation standpoint or something like that. Like, that will give you the signal, like, okay, this might be something we might want to take a look at and trim or something like that.
- [56:02]
B
Yeah. And a lot of times, you know, it's just a head fake and we're positioning or short term. But, you know, if you zoom out and just think about, like, where. How valuation, I think it really does come down to. Valuation is like, you know, how extended are we? There's various metrics we love EV to sales price for earnings, free cash flow, dividend yield. And so when those are at like, extremes and the multiple starts to like, compress on higher numbers, that means, like, the market is like becoming more full, I think, in that aspect now it can do that momentarily, then like, expand again. So I think that's the art of it. But to me, it's something that I do pay attention to.
- [56:46]
F
I think this episode is brought to you by Palmolive. Family time isn't just the big moments. It's weeknight dinners, sitting around the table, everyone talking all at once. So when the plates are empty and the sink is full, use Palmolive Ultra. Palmolive's most powerful formula removes up to 99.9% of grease, leaving your dishes sparkling clean. And the new convenient pump makes cleaning even easier so you can spend less time tackling dishes and more time together. Shop now@palmolive.com tomorrow morning is knocking. Stock your fridge now. How about a creamy mocha Frappuccino drink? Or a sweet vanilla smooth caramel maybe? Or white chocolate mocha? Whichever you choose, delicious coffee awaits. Find Starbucks Frappuccino drinks wherever you buy your groceries.
- [57:34]
C
How are you and your team thinking about intangibles in this world? I mean more and more companies are, you know, a lot of their value is in these intangible assets and so are you. How do you kind of think about that when you're thinking about, you mentioned valuation ratios. Those may be for the more established companies but even, even those tech companies, there's a lot of intangible value there in terms of what's embedded in the company. And, and so how do you kind of think about intangibles when you're I guess looking at these companies?
- [58:07]
B
Yeah, no, it's a great question. You know I think there's a few things so I think I look for a five minute level strength first of all and evidence that and things that I could hang my hat on that the, the numbers are better than people think. So I think it knows kind of intangibles or like these narrative stocks. I, I, I do think it's super hard and my allocation of that is probably pretty low. But like once it becomes more real, like I do think like it's okay to buy it up because of nd risk. There's real numbers that are coming through. So if it's just on intangibles or like some of the parts I tend to find like don't gravitate to that because the market is, it's really hard. The market doesn't care if you hold up your sum of the parts and it's higher than stock price. You know, like it needs to have a catalyst and some that and some investors really good at that. Like they can you know, kind of understand that there will catalyst that comes and, or they just are very patient and it's inevitable. But I like to see like a real catalyst come through before I think I take a position or I could see it come through.
- [59:18]
D
I'm just wondering like I've gotten into doing these pre mortems. I don't know if you ever do them but I think about like putting myself in the shoes of myself 10 years from now and Thinking about, like, if this were true, what would have happened to make it true? And I'm thinking about both sides of AI from that perspective. If AI greatly exceeds expectations, like, what does the world look like 10 years from now? And if AI, you know, disappoints, like relative expectations, what does the world look like 10 years from now? And I'm wondering if you have any thoughts on both sides of that ledger.
- [59:46]
B
Yeah, I think if AI succeeds more than we think, you know, I do think you're going to have like a massive unlock and economic prosperity because you're not going to. The bottleneck of labor is going to be uncapped. And I do think, like you probably get into like very physical world AI. I think there's going to be a lot of intelligence embedded with every person. They're probably going to be having a team of agents, agents running in the background. And I think we're probably going to do more. Or you can either do more or you can just do less possibly. I do believe that like, you know, technology has deflated a lot of costs. And so in some ways I do view this like abundance bull thesis as like pretty possible. I think about like, you know, Uber, for example, right, Is kind of deflated the cost of transportation, right, versus like you just had taxis and you know, they expanded the market at the same time. So I think, I do think I subscribe to that more bullish, optimistic view that AI is really what we hope it is, that that can happen. And I think we're going to see, you know, a ton of creative ideas. Like you could look at like film. Previously you had to have like a hundred million dollars to make a great film. And now you can, you know, do it generated. And so perhaps like you would be like really good storytellers that don't have like the backing. And so to me, I think the creativity part of it is going to be really enhanced and there's going to be more ideas that are executed on as a result. And then the downside case. Yeah, what's that? Well, I think if the downside case happens, then you probably do have an overbuild in AI infrastructure for sure. And then, you know, you're going to have like probably this cycle of companies that have to absorb the computer in non AI ways. And so that can happen in other domains. I think it's just gonna be slower to eat through that. And so I think that you're gonna probably see a lot of valuations compress from these levels. And so I definitely think that you're probably gonna have software companies got rebound evaluation and so it's gonna be more like what we saw. It'll be probably more the same thing what we saw over the last five, 10 years as a result. Yeah, I think that's, but I think there's like, there's probably you know, kind of a more likely, you know, if it's probably just more delayed, like AI isn't, like doesn't. Because I think we've seen that it does work pretty well. It's just more like is there a gap between the spending and the roi? Do we need to have digestion kind of catch up a little bit? And so I think it's more of this gap of like, okay, the hyperscalers are implied to spend, you know, over a trillion dollars in 2028. Like, you know, do we go into a macro kind of downturn? Is there exagger, shock? Does the, you know, and as a result do we just need to like have a little catch up and digestion before we see the next cycle? I think that's kind of more the likely know pre mortem on, on why the stocks don't work from here. To me, I think there is like enough promise that there will be continued investment maybe just at various paces.
- [63:30]
C
Tony, thank you very much for spending the time with us today. We like to ask all of our guests one standard closing question and that is based on your experience in the markets. If you could teach one lesson to your average investor, what would that be?
- [63:43]
B
Yeah, I think that when I first started working in a job, I think I took a very static view of like, oh, I'm going to build a VCF or like, you know, I know what the company's worth and I'm analyzing what's, you know, latest filing. I think more of it is like a, you got to take a view on like what the long term is going to be. And then I think you have to think about the rate of change. And so it's all about whether you know, things are getting better or worse. In terms of perception, is it the market is a voting machine, right? Like it's thinking about a room full of a hundred people. It's like, you know, today 50 people believe in the company. The real bet is like if 80 people are going to believe the company in a year.
- [64:27]
D
Right.
- [64:28]
B
Or is it going to be 20? So it's not so much of a static thing. It is very much so. I think a view. There's as much like, you know, pursuit of like the truth as, as there is in the, in the kind of, I would say the narratives and the, in the, in the sentiment and the valuation and just realize those can have like, can swing in big directions. So that's just something that I've had to do to hone my craft here. Good stuff.
- [64:58]
C
Thank you very much, Tony.
- [64:59]
B
We appreciate it. Thank you. Appreciate you having. Thanks for the very thoughtful questions and you know, always enjoy listening to your podcasts.
- [65:07]
C
Thank you for tuning in to this episode. If you found this discussion interesting and valuable, please subscribe on your favorite audio platform or on YouTube. You can also follow all the podcasts in the Excess returns network@excessreturnspod.com if you have any feedback or questions, you can contact us@xsreturnspodmail.com no information on this podcast
- [65:27]
D
should be construed as investment advice.
- [65:30]
F
Securities discussed in the podcast may be
- [65:32]
D
holdings of the firms of the hosts or their clients.