AI and the Great Transformation: A Global Investor’s View

“The Angle” Music

Cold OPEN “I think the most fundamental thing that I find when I discuss AI with clients, or analysts, or portfolio managers, or even companies, is that AI is not software, right? So, so software is incredibly scalable. AI is a capital intensive thing. So the more AI you use, the more you need to build.”

Jennifer Martin

Welcome to “The Angle from T. Rowe Price”, a podcast for curious investors. Just a reminder that outside of the U.S. and Australia, this podcast is for investment professionals only. I’m your host, Jennifer Martin, a global equity portfolio specialist at T. Rowe Price Associates here in Baltimore, Maryland. In this new season, we look at artificial intelligence and how it is moving from promise to practical impact, reshaping companies, industries and economies. We look beyond the hype to examine where AI may create lasting value, where expectations may be too high, and what investors should be watching as the technology moves into the real economy.

My guest today is Dave Eiswert, who is the ideal person to open up the season. As a global equity portfolio manager, he assesses disruption, competitive advantage and where the next sources of relative return may emerge across the market. I would also describe Dave as half man, half AI, and he is one of T. Rowe Price’s AI wizards, helping us shepherd our firm into the world of agentic AI and beyond.

Welcome to the podcast, Dave.

Dave Eiswert

Thank you Jennifer.

Jennifer Martin

All right. So let's start with the first question. When you are investing across global equities, what is your first question you're asking about who wins and loses from AI?

Dave Eiswert

You know, I think we have to be, we have to be very careful when we think about what we do as investors and sort of what our core principles are. And, you know, I think apart from AI, the way we invest is really looking for places around the world where we see the best accelerating relative returns in businesses, and that sort of philosophy, and of course, not not paying too much. Right. Valuation is an important part as an investor. So, we're you know, that's in our DNA as a firm. And in the strategy that I run is, we're constantly looking for those insights about improving relative returns. And when you find yourself in a cycle like AI, you see that the best returns improvement are in this particular area, where investment is happening at a, at a really epic kind of scale in proportion. And so when we look around the world, we really look for those places that are levered to this AI cycle, where pricing is a very important part of the, of the algorithm, the quantity, the price, the incremental returns that you get, the amount of CapEx that's required to sustain revenue growth.

And I think in AI, one of the very interesting parts about it is revenue growth for people selling AI, selling LLMs are very tied to the CapEx they put in their business. And so there's a little bit of a double-edged sword there. You grow faster, but you have to spend more. And so, you compare that with other areas of the of the supply chain around the world where they're not having to invest as much and they're benefiting from those incremental returns. And that's really an AI infrastructure and the supply chain. So I would say, in general, look for the area. We're always looking for areas where we see that improvement in relative returns.

Jennifer Martin

Yes, so the harder question is relative return, and which companies are getting better faster versus expectations. Okay, so is AI becoming the market's favorite alibi for companies that were already losing momentum?

Dave Eiswert

Yeah. Well, I mean, I think, you know, I think between the global financial crisis and, and, and the Covid crisis, we really went through a period of what a lot of people called secular stagnation. And if you look around the world, you had Chinese growth slowing, which had been a big driver in decades before that period. You had negative interest rates in many parts of the world. You had a really massive technological cycle going on in that period, and it really was about software and cloud. And software is a, is, is, a capital light business, right? You write software and then you copy it and you give it to a lot of people. So that period of time between the GFC and Covid, you can almost think about it as a relatively capital light, very software eats the world period of time. And so, in that period of time, you know, investors always go look for the best relative returns, the best relative earnings growth. And that was really happening in companies that you would consider kind of durable growers, right. They were able to do M&A. They were able to buy back their own stock. They were able to take advantage of software and cloud to grow their businesses faster than the rest of the market.

And so, the market really crowded into those areas, and they ended up at the end of the Covid, as we came out of the Covid cycle, those particular areas ended up being extremely expensive, extremely crowded. And it really was sort of the, the investment strategy de jour; was I, buy durable growers, I buy high quality companies. Don't get me wrong. There's nothing wrong with buying high-quality companies. And I think there's nothing wrong with durable growers. The issue is, is that, you know, as we came out of Covid and as AI was invented, we, you know, we suddenly were faced with a world where capital intensity was going up dramatically. Revenue growth was going up dramatically. Earnings growth has been going up dramatically. So this period as the regime changed from really, call it the secular stagnation of the global financial crisis, to, to, Covid, this regime change had led to a much more capital intensive world. And so, the principles of investing haven't changed. But where the relative returns growth is the, is the greatest has changed. And so you really have seen investors go after the cyclical parts of the global economy, which is levered to AI. So yes, AI is where everyone's focused. And it's not because it's a fad. It's really because that's where the relative returns are the greatest.

Jennifer Martin

So, let's try it back to the alibi, because AI disruptor can be really a lazy explanation for several different problems. Whether it is slower growth, the higher rates that you mentioned, the weaker pricing power, all those things that are happening. So, you know, I think one of the things that you've been kind of nudging people to really think about is do you not label AI disruption as a substitute for actual analysis?

Dave Eiswert

Right. No, no, I think that's a that's a great way to put it. So I think the reaction of a lot of portfolio managers and analysts as durable growers have underperformed is they've said, oh, this AI thing, it's just a fad or it's just a cyclical right. And don't, you know, don't pay attention. And that's been extremely costly for investors because it's sort of that, that dogma or religion around there's one way to make to make money. And you can only make money in these durable growers. The fact is you make money from identifying those relative returns.

Jennifer Martin

And I think what you've also pushed is really differentiating, you know, is the market mispricing permanent impairment, or is it just a deceleration of growth rate. And I think that has been one of your frameworks.

Dave Eiswert

Yes. There are companies that there are. So, so, we have a saying on the team that stock.

Jennifer Martin

A lot of good sayings.

Dave Eiswert

Yeah, a lot of good sayings yeah, yeah it's a good way to shortcut thinking. But you basically say prices follow returns. So, stock prices, relative prices, they follow where the relative returns are getting better. Money. Money likes to be treated well, right. And it goes to where it's treated well. And so it wants to follow those relative returns. But then narratives follow prices. And human beings talk about stocks in, in, in, narrative terms. And so, you have, what you have is, you have stocks going down because they're not growing fast enough. Multiples contracting, and then you get narratives thrown on top. So and I think what is a really interesting place to spend some time is trying to differentiate those software and business services companies where, you know, yes, there are multiple has gone down or the stocks have gone down and now they are labeled with an AI label.

But it's not necessarily true that that's why the stocks have gone down. And, and I would even go further. It's not necessarily true that those can't be good stocks in the future. So, it's just important to understand how much of a stocks performance is due to relative growth, how much is due to crowding out, right, because of that relative growth, and how much is actually due to the idea that the business is being fundamentally disrupted by AI. And the AI alibi is I just say, oh, it's AI, right. And that's not enough. Right? We got to do more. And I think especially in, in coming years, right, the ability for us to differentiate where, you know, suspected AI disrupted companies are really not disrupted. It requires a lot of hard work to find those.

Jennifer Martin

That's perfect. And it's a good transition because you mentioned software and you just came out of our recent West Coast technology trip. And one of the points that I heard loud and clear from you was the demand picture for AI infrastructure now seems obvious. There is simply far too much, you know, there's far more demand than the industry can supply. So you said the real question is duration. So maybe you could expand on that, because I think a lot of people are toggling between the words AI boom versus AI infrastructure, arm race. Like how would we provide a little clarity to that?

Dave Eiswert

Yeah. Well, just to just to, you know, build out that argument structure a little more, you know, in the period between the global financial crisis and Covid, the global supply chain became incredibly efficient. Right? And we really saw this in the Covid crisis of how, how many shortages we had. But it became really incredibly efficient, and it was really focused on building iPhones and building cloud servers. And that's basically what we were building. And so we became this, the global, the global supply chain became very concentrated in Asia and became very concentrated around making iPhones and making cloud servers. And so, and people got very used to these sort of short, shallow cycles. There's an iPhone cycle. Well, we need more DRAM and then everyone buys a new iPhone. Now we don't need DRAM anymore and the market would, would collapse.

So what's, what's happening with AI? I think if you want to sit back and think about it very simply is that supply chain for iPhones and for servers smashed into this massive cycle where revenue growth is a function of CapEx. And so what ends up the cycle that we're in now, is not just a short, sharp cycle of an iPhone cycle. It is the structural expansion of the supply chain. The supply chain needs to be multiple times bigger to support this new driver of demand, either in logic chips or CPUs or GPUs or DRAM, or even optical components. So we're actually going through a period of time where we just need to double and double and double the scale of the supply chain.

So what's really important about that is that does not mean that hardware will not in the long run be cyclical. It will be cyclical. What it means is we need to build it up to a point where we can actually meet demand, and that, you know, that the physics of building that are constrained by the reality of the physical world. You know, you know, I have four kids and each one of them it took nine months for my wife to have that baby. Right. And that is just the way that physically works and the way it physically works to build out the supply chain. So that is the part that's driving, I think, a lot of people crazy because they're like, well, I understand this. This is a cycle. I buy low and I sell high, and what they're missing is the duration of, of, building out that supply chain to meet the new demand driver, which is today. It is there's a binding constraint on how much we can produce and that binding constraint each year we're moving it, we're improving it, but we're not over supplying.

Jennifer Martin

So you, that was beautiful in addressing the really the supply piece of it. And then you have a really good framework on the demand when you're talking about scaling laws and CapEx cycle. So maybe you want to expand on that. Just, you know, what you're monitoring in terms of really if AI gets cheaper and easier to use, does that reduce demand for compute? Does it make demand explode? We have to bring up the word Jevons paradox, because I love that you named your your harness that.

Dave Eiswert

So lots to unpack there. So just on scaling laws, just for everyone, scaling laws basically mean the more capital I put towards the problem of AI towards training models, the better the results I get. And we are finding that both in training, which is like building the model, and in inference, which is the serving of, of, these questions we asked the model, we're finding that scaling laws hold. So what that means is everyone's sort of to stay on the leading edge, you've got to keep investing capital. And that is a critical part to drive this cycle. So as long as the models are getting better and we're accruing more value, right, in terms of the users of the model, then there's a lot of pressure to keep investing in capital. And, you know, just to take a step back, another concept, I think from economics that is incredibly valuable when you're, when you're trying to understand why, or why are people behaving the way they do, the way they are, really, despite the fact that, you know, a lot of investors would wish that hyperscalers would stop spending so much CapEx. So please stop. Right. If you're, if you're a Google or Microsoft shareholder, you're like, would you please stop spending all this CapEx, your free cash flow margins are going negative. And then so you sort of step back and go, why are these companies doing this? And it really comes down to this idea of Nash equilibrium, right? Which I know is used by a lot of people, but it's an incredibly powerful concept.

Jennifer Martin

It's a great tool.

Dave Eiswert

Yeah. It basically says that you are in equilibrium when no player can make a move themselves that makes them better off. And and so if you think about that, if, if a hyper scaler, or even a better example is thinking about it geopolitically, China versus the U.S., you know, which country can stop investing in AI and they will be better off?

Jennifer Martin

Spoiler alert: none.

Dave Eiswert

None. Right. So, if, if, a, if a, if a government lacks AI, they are now fundamentally vulnerable to an outside government or entity attacking their IT system. So, and if you're, if you're Google or Microsoft, if you get off the AI path, especially if scaling laws are improving, so the models are getting better and better. If you get off that path and let your competitor stay on the path, they are going to drink your milkshake, right? And so you know that. And so in a Nash equilibrium you are stuck. So although your investors may wish you would stop spending on CapEx, you are actually making the rational decision. And so scaling laws, plus that Nash equilibrium, plus downward sloping demand curves, you know, that keeps us in this cycle. And that is not a, you know, I would just say it's not a irrational thing. It's not a bad thing. It is, in fact, and I don't know if you're going to, if we'll talk about this later in the discussion, but in fact, it is likely the force that's going to drive the acceleration of real GDP growth, which will lead to rising incomes, which will make human beings better off. So those things are all very intertwined in in how we can explain the behavior. And I just if I add one more thing here, there is a very, it's human beings hate change.

Jennifer Martin

We want everything.

Dave Eiswert

Another one of our sayings. Human beings hate change. They want things to stay the same and get better. And so everyone, I think a lot of investors long for the secular stagnation. They, you know, they they don't want to invest in European banks. They don't want to invest in supply chain. They want to invest in software and they want to invest.

Jennifer Martin

It’s nostalgia.

Dave Eiswert

Yeah. Well, I mean and so there but, but I think the way we look at the world, and I think is quite true, why this is one of the most exciting times to be an investor is that we're in this period of deep change. And so, you know, you really want to think about those changes and you want to think about, just to my point on, I'm going to make a pitch for liberal arts education. You know, understanding why people do the things they do and why businesses are doing the things they do and why those things are not irrational. Right. It may be that, you know, it all depends on how you define a bubble. I mean, we're certainly having a massive amount of capital allocated to a particular area, but but making a claim that it's irrational, I think if you deconstruct that, you will find, no, it's actually quite rational. Right. And I think that's why as an investor, you, that's why we feel comfortable being positioned the way we are, because we actually see the rationality of it.

Jennifer Martin

It's a great framework, and I know a lot of people really appreciate kind of that checklist. When the music stops, you have a playbook that's really important. And so I think maybe we should go a little bit deeper into the AI infrastructure story, because it is going deeper in terms of the rack. You know, as rack power moves from kilowatts to megawatts, I never thought I'd be so bright on utility metrics. The bottlenecks are no longer just GPUs or memory, but power semis, connectors, MLCCs, optical copper connectivity, you name it. How do you think about where value accrues inside the data center? And you've been really looking at this in a much broader way.

Dave Eiswert

Yeah, I again, I will take everyone back to this idea that, you know, if you think about a long arc of investing and where value comes from investing, it really comes from identifying where relative returns are the best. Assuming you don't have to pay too much for them, right. So inside this you're talking about all the physical parts. So that, and it even goes to construction, right? Yeah. Construction equipment, outside generators. So I think what you're really hitting on is this physical part of of investing. So we're, you know, we're using our teams. We're looking for the places in the data center where companies will have pricing power and better returns.

Now, you know, I want to I want to touch on something here because I, I have news for you. Breaking news. I am not the only investor who's looking at bottlenecks. Like everyone's talking about bottlenecks and you know, and it's not. And so yeah, I mean they're vibing right. There's a lot of vibing going on. There's a lot of price momentum going on. And you know, this is a real challenge as an investor because when I'm in. If I, if I make an analogy for you, if I'm in a room alone and I have this great idea about how the future's going to play out, and I look at the stock prices and the stocks are down and the valuations are cheap. And there's really there's, there's, nobody doing podcasts like this one telling you about how great everything is. Right. You're sort of alone thinking about something. And as an investor, I love those moments. Right when I, when I, when I'm early to an idea and I have a good insight. What happens then, as that insight plays out is that the room becomes full of price momentum. And so you can think about it like you're in the room and, and sort of the finance bros start coming in, and in, their vibing and they're wearing their vests and they're jumping around right in there and there, and everyone's doing a bottleneck piece about where AI, you know, and you go on Twitter and it's like, it's really quite amazing. So then you find yourself in this room, so you're in this room and you're like, I was really comfortable in this room when I was alone, and now I'm in here with a bunch more people. And in fact, what these people do is every couple of weeks, they all run out of the room, and then they all run back in, and then they all run back out. Right? And it creates all this volatility around price. And that's hard for investors to cope with. It's hard for clients to cope with. And so a lot of investors would say I'm very uncomfortable in this room. I do not want to associate with these people. So they leave right.

Now the problem is that when you leave that room, if you're leaving just because of the company, right? Just because of the company you're keeping, you don't like the vibe, so you're leaving the room. The problem is that you could be walking away from solid fundamentals. So, the intersection of price momentum and durable fundamentals is a very uncomfortable place for a lot of fundamental investors. And I think that's sort of where we find ourselves today. Right. And even doing this podcast, I'm doing this podcast. And, you know, I'm holding these two ideas in my mind at the same time. One, I'm very uncomfortable with the price momentum. But it's uncomfortable because I'm now in this room full of vibe investors. So the question I have to ask myself as a fundamental investor, and based on the research we’re doing, is are the insights around the fundamentals, are the insights around the relative returns, are the insights around the next cycle in the data center? Do we have enough insights in there so that we can stomach the vibe? Investors running out of the room and running back into the room, you know, do we have, you know, or are we going to be swept with them either way? And, and, the other thing would be as a portfolio manager and running a diversified portfolio, I need to really be able to build the portfolio so that I can survive the vibe swings and stay on the right path, in the right amount, in the right time, right, with the right timing around how we run the portfolio.

Jennifer Martin

It's a lot.

Dave Eiswert

That's a lot. Sorry.

Jennifer Martin

No no no. Yeah. No, but I think I think that is sort of the magic right now. And I think that's sort of the psyche of a lot of investors today, of maintaining the conviction in a trend that seems obvious, but still having those insights to, you know, press a bet because the market really underappreciated things. So now, I have to turn to one of my other favorite sayings you've said. You've raised the question of when salad turns to garbage. And, and, I think we could we could use this in a many ways. It's one of my favorites. But maybe you want to ask the audience, what's the answer?

Dave Eiswert

Well, yeah. So I'm gonna say, first of all, a lot of, a lot of our sayings we get from different. I don't know where that came from. I did not invent that. So, I don't I don't know where that comes from, but I love it as a simple kind of analogy. There's lots of people that will tell you this cycle will end. Yes. And I don't find that very helpful, you know, and so a lot of times I ask the analyst, again, I borrowed this from, I'm sure, someone much smarter than I am. But I say, well, what's the difference between salad and garbage? And I always.

Jennifer Martin

I love the blank faces.

Dave Eiswert

This really blank look like, what this guy is talking about food? I thought we were like, you know. And I say, well, it's difference between salad and garbage. And it's a really, it's a really nice concise saying because it ends up being timing. Right. Like all salads are eventually garbage, right. Now, the question is, if you throw your salad out when it's perfectly good, you don't have a good lunch, right? And you're hungry. And if you wait too long, you end up with a bowl of garbage, right? And so how do you think about as an investor, you know, how you understand the duration of the cycle, right. And like getting that duration correct. You know, I think that's actually a level of difficulty that's really high. And, and, our job as investors I often think about for my clients, my job is often to eat the fear. And so sometimes it's in a down market where stocks are everyone selling all their stocks.

You, you, spend a lot of time, and your job as a firm, as T. Rowe as a firm, is to have the resources to be able to eat that fear and understand, okay, this is what's real and this is what's not real, and make a decision. I don't think investors think often in an upmarket when things are going well. That's also a time where you have to have the, the resources, the sophistication, the frameworks to eat fear around when the cycle will end. Right. Because nobody, I do not come into work every day saying, I would like to blow up my clients and look tremendously foolish by riding this off the cliff. Right? So, yes. So, I mean, we spend a lot of time thinking about, okay. How long will this cycle end? What are the fundamentals? What would turn the cycle? How do we manage risk? But a significant amount of that is also on volatile days or days when there's some headline story that's driving the vibe investors out of a sector, is your ability to say, okay, this is noise, this is not, this does not affect our fundamental thesis here. Here are the correct moves we should make in this scenario. And so that you don't you don't think about it. But in up-markets there's also a certain skill and expertise that goes into eating that fear and staying with, you know, the good fundamentals at the right time.

Jennifer Martin

So I would I would say that, you know, we've spent a lot of time talking about your frameworks and also talking about semis, memory, infrastructure. And I think one of the more interesting takeaways from the recent industrial strip that we had in in London just last week is that AI is, you know, becoming such a physical infrastructure story, while you touched on it a little bit, but it seems like the opportunity is becoming a little bit more global over time. And so maybe you want to talk about, you know, what separates companies that are merely have that AI narrative attached to them from, to really the return profiles of those businesses?

Dave Eiswert

Yeah, I mean, certainly, you know, the supply chain is global and it's very focused in Asia, although I think there's always unintended consequences of things. Right. So one of the great unintended consequences of the AI cycle has really sort of been the saving of Intel. Right. And you know, Intel.

Jennifer Martin

Yeah. Lip-Bu was just on our podcast.

Dave Eiswert

Was just on our. Lip-Bu was just on with Eric Veiel. I thought it was a fantastic interview. Really. I have a tremendous amount of respect for him as a CEO and as an entrepreneur, and a thinker. It's a great story. I encourage people to listen to that. But like there are, this cycle is so big that it's going to create these unintended or unexpected consequences that are going to really reshape the supply chain. So there are going to be companies that go from being, you know, sort of a commoditized component player, to being a sophisticated scaled provider of a certain component that that actually creates its own competitive advantages. And so we're always looking for that. What are these companies that, you know, they lucked out to a certain extent of being in this space. They took advantage of that luck. Right. And then they created some competitive advantage that will be sustaining, and create high returns in the future. I think, in general, in the United States, there's the, more physical infrastructure we build, the more externalities we're going to get from that, which is going to allow us to do things that are unexpected.

Health care is another area we're spending a lot of time on, in an AI, I think. I think one of the things about, about health care is, it's very interesting because most people involved in health care. And I think one of the things in healthcare is healthcare is a is a field that is incredibly precise. And we've spent a lot of time focused on precision and on science and on data and on experimentation and, you know, and that's wonderful, really that and that's really the nature of, of, the way health care and health sciences have developed. But what's really interesting now is that AI is allowing us to simulate healthcare and then to sort of take much more risk in guessing outcomes digitally and guessing relationships, and then going through all those guesses and then moving on to doing physical commercial trials. But so I think, you know, what's going to happen in healthcare. And I think with the innovation in health care, I think that's going to be a super interesting space.

Because with AI you really have to say, well, if the innovation cycles are going from, let's just make a simple, they're going from ten years to four years, you really have to be prepared to think that a business could change fundamentally in that four year period. And so you have to be open to that, right. You have to be open to that kind of that kind of change. That's really hard.

Jennifer Martin

It's very hard. And I think I referenced you in the intro as an AI wizard, because you're not just investing in AI, you're using it. And maybe you could give us a few examples of how AI changed the way you research companies. Test your ideas. It's incredible actually, what you've built. And I love that you call your harness Jevons.

Dave Eiswert

Yeah. You know, it's, it's, a great question. Maybe I'll share too much for the audience. But if you go back 24 months, I would say I was pretty glum about our industry. I was pretty glum about, you know, I've been doing this for, I don't know, 26 or  27 years. And I was really sort of like, this is kind of boring, right? I mean, there's like ten companies that dominate the indexes. Everyone wants to be passive. Passive is generating strong returns. You can't argue with that. And, you know, it's quite challenging for active managers who are looking for something nuanced, or different. Right. And so, and then I would say the spring of 2025, and when the O3 model came out, and we really started to understand reasoning. For me, that kicked off a really critical part of my career. And again, I don't want to share too much with our audience, but just between friends, I really thought, you know, I, this is a major, fundamental change and I need to decide as a person. Am I going to participate in this change?

But I would say for me, I really thought about it like like like evolution again, not to use science but like a fish out of water. So you're a fish. You learn to swim underwater, you grew gills. You're kind of enjoying your life, although it's getting kind of boring in the lagoon. Right? And then suddenly a wave comes and kicks you up on land. And that's sort of what AI is like for a lot of people like me, mid-50s in their career. And suddenly AI is happening and you find yourself on land and you really sort of ask yourself. Am I interested in growing lungs? Am I going to take this journey and learn about AI? And am I going to embrace this and really learn to use it? And so so for me it was sort of like this decision I'm going to grow lungs, and growing lungs is painful. It's taken a lot of work.

I, my wife would call me an AI vampire. She's like, you just want to sit with your laptop and talk to Claude. And use Claude code and use Codex? And what are you doing? I have rules now in my house where I can, where I can and cannot take my, my, laptop. Because it's very. Once you've decided you're going to grow lungs, and once you sort of go down this road of learning the power of AI. And if you're someone who loves data, and if you're someone who loves computers, right, it's suddenly now. And if you're curious again, I think curiosity and imagination. Any idea I have about investing? Technical analysis, fundamental analysis, some new sector I never learned about, you know, testing hypotheses about what leads to the best alpha across stocks. Anything I want to do now, I can just ask, and I can then build up this data. And I think what I love about T. Rowe Price, and I think, and I think we actually think about this in the long run as a competitive advantage. I mean, we are embracing technology and AI. We are rebuilding our data sets. We are using, you know, different forms of software to build databases, to give this access, to give Claude code and codex and really pushing those limits because we want to attract people that have those characteristics because we think this is so profound. And so I sort of went down this road of, you know, trying stuff, running experiments. And you, it's just this process where if you're curious, and you're using your imagination.

So what we're doing. So we're using so we're using all the different tools. I personally don't love wrappers. So I really want to get very close to the data. And I want to see the code. And like a lot of people, are there's a lot of companies that are trying to wrap the models, and I'm more trying to be close to the model, but we're building out a whole structure. Jevons is our, our, our, sort of orchestrator agent. And, you know, it's from Jevons Paradox, right? Which is the idea that the combustion engine would use would end up using less coal in the UK. Of course, you ended up using way more coal. Right? So that was the, but that's the paradox. But, but, you know Jevons job, he's orchestrating like the, all the different elements of the framework. Right. And so the way we I've spent the last 27 years kind of building up, what do I fundamentally believe about investing? And some of those things were just feelings, some of those things where I, this was, worked for me. So I'm going to keep doing this. And what we're trying to really do is build that into a systematic architecture and to test that architecture, and to, and to make it better.

So, I think the best thing for me is that really I'm seeing our framework improve. Right? So I'll use a striker analogy like you've got a great team, you've got a great striker. You know she gets the ball 15 times in the game. She takes seven shots and she makes two goals. And you're like, well that's amazing. We have a really good team and you're like, great, now let's lever her up with AI. So let's get her the ball 100 times in the game. Let's have her take 50 shots and let's have her make 20 goals. Because she's a really good shooter. She, she, has a way that she invests that we've seen over time be effective. And I think we have a lot of people like that too at T. Rowe Price. And now we're like, let's get them the ball more. And so that's kind of how we're trying to embed AI into the into the system.

Jennifer Martin

It's perfect. So maybe the last question would be what do investors most misunderstand about AI today? And what is the one idea from the conversation you want listeners to take away?

Dave Eiswert

I mean, I would just go back to that. I really would go back to this is a time when imagination and creativity, it really is compounding, right? So having an imagination and being creative. That doesn't mean being undisciplined, but in terms of valuation, in terms of what drives stocks. But I mean being open to change versus, versus, hating change, I think is a is a good characteristic. You know, around AI, I think the most fundamental thing that I find when I discuss AI with clients, or analysts, or portfolio managers, or even companies, is that AI is not software, right? So, so software is incredibly scalable. AI is a capital intensive thing. So the more AI you use, the more you need to build. I can write software and share it to everyone on this podcast, and the marginal cost is zero. If everyone on this podcast is trying to build their own models or do their own work, they're all going to consume in AI. They're all going to be consuming DRAM, they're going to be consuming NAND, and they're going to be consuming CPUs and GPUs. And so that infrastructure has to be there.

So that, that, just understanding when someone says to me, like the link to this kind of misunderstanding is people will say, Dave, you're too bullish on AI, the Chinese are going to have free LLMs where they don't charge for the LLM. So free LLMS does not mean that the infrastructure trade isn't happening. It would change that. I'm not worried about free LLMS. I'm worried about scaling laws. Right. So. So AI is not software. And you know this free model idea. Just understand that it's, so think I, just one thought experiment. Imagine that LLMs are free and that they are also brilliant. So they are PHd, or beyond, a level of intelligence. So, I would just I would posit to you, first of all, there is nowhere near enough infrastructure in the world to support the use cases of that.

And the second thing would be if there were, society would so fundamentally change. I think this is sort of where Elon Musk talks about this idea that no one will be working in the future. I think what he's imagining is, what if these models were so great and the price per token were so low? This is why I talk about scaling laws being the barricade. So the real bear case is that models end up being a college freshman. So, the best model is a college freshman, doesn't get any smarter than that. And the ROI we start to question, because this college freshman, yeah he's pretty good. He can do some, you know, he can do some simple things, but he also makes a lot of mistakes. And the ROIs not that good. And in fact, I have people that are just as good as this college freshmen. And so, you know, that I think is where the AI infrastructure trade would fail. And I think you would want to get out of infrastructure, because the ROI wouldn't be there. As long as scaling laws hold and we keep getting better. You know, I think the whole, the whole equilibrium we've talked about here sort of holds.

Jennifer Martin

That's perfect. So I heard creativity, AI is not software. And I also want to bring in a thought you said earlier, which is that AI beneficiaries are obvious. Some of the best opportunities sit in less glamorous parts of the supply chain, or could be parts of the market that are now getting derated. At some point we might want to look at them again.

Dave Eiswert

We spend. So the two areas that I spend a lot of time on are high-quality companies that are being crowded out by AI, and I actually use AI to think about those companies a lot. So, Claude and I have all sorts of projects have sort of, you know, analyzing, you know, what are these really high-quality companies that seem to just keep derating. So it's not that they are really being disrupted by a, they're just being crowded out by AI. And so there is a, there is an equilibrium there to where these, these investments will become very attractive. And so we're we're actively watching that.

Now we'll go back to our salad and garbage example, like just buying high-quality companies when they're halfway through their derating is not a way to keep your job, right. So, you need to get the timing right in terms of, of, those. There's no extra points in difficulty, like if you can find the place where the relative returns are the strongest, the valuation is reasonable. So, stay with the areas where you know your framework delivers consistent solid returns. And so, yeah, we're spending a lot of time there. And, and, yet not doing a lot, right. A lot of time, not a lot of portfolio decisions. Because you know we're trying to maximize our returns with the, with the environment we have today.

Jennifer Martin

Perfect. This has been a fascinating discussion. You know, I think there's so many things you've highlighted, particularly AI is not just changing technology, it's taking the economics of intelligence. And you've really made a great case for investors’ need to follow where scarcity, capital, and value creation actually move. So, thank you so much Dave for the time, and we really appreciate it.

Dave Eiswert

Thank you. It's been great. Thanks for having me.

Jennifer Martin

Again, I'm Jennifer Martin. Thank you for listening to The Angle. We look forward to your company on future episodes. You can find more information about this and other topics on our website. Please rate and subscribe wherever you get your podcasts. The Angle. Better questions, better insights. Only from T. Rowe  Price.

DISCLOSURE

This podcast episode was recorded in June of 2026 and is for general information and educational purposes only. Outside of the United States and Australia it is for investment professional use only. It is not intended to be used by persons and jurisdictions which prohibit or restrict distribution of the material. This podcast does not give advice or recommendations of any nature or constitute an offer or solicitation to buy or sell any security in any jurisdiction. Prospective investors should seek independent legal, financial, and tax advice before making any investment decision. Past performance is not a guarantee or a reliable indicator of future performance. All investments are subject to risk, including the possible loss of principal. Discussions relating to specific securities are informational only and are not recommendations and may or may not have been held in any T. Rowe Price portfolio.

There should be no assumptions that the securities were or will be profitable. T. Rowe Price is not affiliated with any companies discussed. The views contained herein are of the speakers as of the date of the recording and are subject to change without notice. These views may differ from those of other T. Rowe Price associates and/or affiliates. Information is from sources deemed reliable but not guaranteed.

Please visit http://www.troweprice.com/theanglepodcast for full global issuer disclosures. This podcast is copyright by T. Rowe Price, 2026.

CON01504670

202606 - 5630009