Hello and welcome to The Download. I'm your host Dave Richardson and we are joined by another Dave. It's a David and David special. Back by popular demand is our good friend David Tron. David, how's the summer been?

Summer's been excellent. It's been volatile in the stock market, but it's been excellent in Toronto where I live.

Excellent. It's been a hot summer in Toronto. I'm out in Victoria, BC. I tell you, the Pacific Northwest late summer is one of my favorite places to be. Not too hot, not too humid, little breeze, no bugs. It's beautiful on Vancouver Island. But it's been a nice summer in Toronto too. And as you say, it's been a pretty good year in the stock market, but an unusual one, maybe just in the volume of things that have happened so far this year. And we were just talking before, you described it as historic. So what are your thoughts on the market and where we've been coming into the middle of August here in 2026?

Yeah, I would definitely frame this year as historic. We started the year like any other year, really. Consensus was expecting 12% or so growth in the S&P 500 earnings stream, which is fairly typical to start the year. As we sit here today, or I should say at the end of July, consensus is now expecting north of 30% earnings growth. So in a typical year, we start around 12% and we get revised down to high single digits or so. So I think an interesting question is: why on earth are we expecting so much earnings growth in the S&P 500 given everything that's happened not the least of which is this Iran conflict, but the AI buildout also? So on the surface, the average investor would see a tremendous number of headwinds as it relates to earnings growth in the S&P 500. However here we are today, north of 30% expected under the surface. And we can talk about all that, but I think the real driving force is AI and it continues to be AI. And we can unpack that a little bit if you're looking to do so.

Sure. So, when you say AI, you're not just talking about the specific stocks that are involved in paying for the building out of AI, the companies that are the beneficiaries of that buildout, the picks and shovels putting all that stuff together. But you're talking about how AI is impacting the economy broadly. Is that what we get a sense of?

There is some element of that, and it is still early days for that impact in terms of productivity benefits. But of course, there are a handful of companies that are writing extremely large checks to build AI data centers, which is providing a boost to all sorts of pockets of the economy, not the least of which is construction. But really, a lot of that capital that is flowing from those check writers— Meta, Google, Microsoft, Amazon, Oracle, predominantly— into the hands of a small number of stocks in the S&P 500 that have had historic returns this year. And the reason they've had historic returns is because they're selling the thing that those guys want, and those spenders are fairly price insensitive today. And by virtue of that, the sellers of that stuff can increase price, and the buyers are more than happy to pay the increased price today. So that flows right to the bottom line in terms of higher margins, higher earnings.

I was going to say something I've wondered about, because a couple of people who have been guests on the podcast made that comment about price insensitivity. I would think regardless of how much money I've got, I want to get value for what I'm buying. And again, I understand that sometimes there's a shortage. I might have spent way too much on World Cup tickets. That would maybe be my personal example. But why these companies don't seem to care about how much they're paying for these particular areas of the market where you've got a bottleneck and shortage?

It would be a slightly different answer for each company. But I'll just broadly speak to it. And this most recent earnings season at the end of July provided us very helpful commentary from these companies as to why they're doing exactly as you just suggested. The short answer is because demand is outpacing supply right now by orders of magnitude. So they're just trying to fill that demand. What is very hard to do and what makes this AI capital cycle so difficult for a public markets investor to forecast is this: we can forecast supply, we can know how many gigawatts are coming online, how many chips per gigawatt, how many data centers are coming online, we can model that, and we have modeled that. There's a lot of rows in that model, but we can get there. Conversely, on the demand side, call it 3 years out, we have tremendous difficulty modeling that. It is very difficult. We know usage trends today on ChatGPT or Claude or whatnot. But what makes it so hard is we don't know how many autonomous agents will be running in the enterprise 3 years from now. We don't know how many personal AI assistants there will be 3 years from now. So really what the stock market and even some of these companies are doing is just looking at current demand trends to facilitate how much they should be spending today, plus a bet on the future of future demand. So that's what they're betting on. It is really one big bet. And these companies would tell you it’s a big bet on demand filling that supply. So that's why they're spending on it. Now, there is some financial modeling at these companies, thank goodness, to suggest that the return on that capital is actually going to be quite fruitful for them, particularly the hyperscalers. Meta, Google, Microsoft, Amazon, Oracle. Meta does not have a hyperscale business, but it sounds like they might be interested in building one. But nonetheless. Andy Jassy on the Amazon earnings call did frame it quite well, I thought. When you build an AI data center, you build a shell— the building, the land, the cooling, call it the shell— and that takes 2 years to build. That shell has a useful life of 30+ years. It's literally a building sitting on land. And you can amortize that over 30 years. And after you spent 2 years building it, then you stuff it with the guts, the compute, the NVIDIA GPUs, the networking, all of that, which is what we call colloquially as guts. And that guts tends to have a useful life of 5 to 6 years. Importantly— and he framed this in the context of today's demand, who knows what demand will be 3 years from now?— but importantly, once you get the guts in the shell, you get a 3-year payback on those guts. So it has a useful life of 6. After 6 years, you strip it out, you put in new stuff, but you pay back that guts after 3 years. So when you run that math, you actually get quite a positive return on invested capital as it relates to this cycle. So that's why they're spending on it. It's a good financial decision today given today's demand trends, and they're betting on future demand coming to fruition. That's it.

So, if that's the case, David, why do we see so much volatility? We've got really big companies that we know are really smart. They made a lot of money, which has put them in the position to be at the forefront of this buildout. And yet it just seems almost day to day in the markets that people go back and forth on whether this is real or how big it could be and whether it's going to work out and whether they're making the right decisions. Is there something about this that creates this back and forth every day. I mean, what's amazing to me as a market watcher for the last 50 years, to see enormous companies move in value off one earnings report or one piece of news, 5% or 10%. I don't recall seeing that. Back in the dotcoms, you'd see news like that, but it was pets.com, and it would go up 100% in a week. What is creating all of this uncertainty around what's going on here?

There's a couple of things, in my opinion. Let's start with what I mentioned before. We can model supply. We cannot really model demand with confidence. So by virtue of that, we are tethered to some real-time, in-the-moment data points as it relates to demand. Two examples would be annual recurring revenue reports from the press for OpenAI and Anthropic, the two biggest frontier closed model labs. Those are private companies. They don't necessarily need to disclose information to the public, but sometimes it leaks to the press. And if the market is dissatisfied with the growth rates of that metric— which is typically just prior month revenue times 12— then the market gets worried about future demand if current demand isn't living up to expectations. So that would be point one. As we sit here today in August, this is literally playing out in the stock market with an Anthropic number that was leaked yesterday. So I think that's point one. The other real-time source of demand would be these GPU rental rates that we have access to in real time. I'm not sure if you've heard of these, but let’s say you want to rent an NVIDIA GPU for an hour, there is a real-time, it almost looks like a stock quote that you can monitor. So those are the two main points of demand. And any volatility in those, likely to the downside, can cause volatility in the stock prices because people just extrapolate today's demand forward 3 years from now. And if there's weakness in one of those numbers, and if there's even an inkling that demand will not fill supply 3 years from now, everything moves around tremendously. So that would be the fundamental reason. And then the second reason would be the stock market average daily volume is increasingly being driven not by folks like us with longer time horizon, longer duration views on some of these stocks, but more so the retail investor is quite involved in the stock market these days. Fast money hedge funds are quite involved in the stock market these days. So I think when you package it all together, there is volatility. Now, it should be said though, we do like volatility as fundamental active managers because it can create mispricing in the stock market that we can participate in. In July, the S&P 500 benchmark was down just a little bit, but there were some names in the stock market that were down 30, 40, 50% in some instances tethered to this AI theme, and they sold off on effectively nothing, which is hard to believe. But we were there to pick up some of the names that we prefer on a longer duration basis, and it was a good time for us, I would say. We don't mind the volatility but understand how it can be concerning for some looking from the outside.

So we're seeing a little bit of that today just as we speak. What are we on, August 18th today, and we're seeing some huge moves. You mentioned a number from Anthropic yesterday. Is this a real move, or is this another one where again you're just seeing this volatility based on a number that somewhat changes the way you're thinking about all this forecasting?

It's always tough to say. Keep in mind, Anthropic and OpenAI want to go public in the not-too distant future. Anthropic probably this fall, maybe OpenAI early next year. So there is a tremendous amount of gamesmanship that goes into the preparation of that. Some of these companies are just capacity constrained. As we spoke about before, demand is well outpacing supply right now. So when you bring supply online, in other words, when you add more guts to an existing shell or you add another shell with guts— these are my very technical terms that I like to use— it can facilitate some of that suppressed demand quite quickly. So one theory goes, Anthropic ARR number was leaked yesterday by someone, perhaps an investment banker looking to work on the IPO such that it sets a lower bar heading into the IPO as they bring on capacity over the next 2, 3 months heading into the IPO. We accelerate off that low bar and everyone cheerlead the business and the stock. So there's a lot of theories and gamesmanship that goes on to this. We like to be tethered more to some of the North Star metrics that we look at for this capital cycle. And I think there's a couple, one of which is: what are these companies actually trying to build here? What are they striving for? And if you do a time series of what these companies publicly say as to what is their North Star metric, you get some interesting data. Some companies started as just wanting to be an infrastructure provider and have now evolved into wanting to be a true frontier lab building the latest and greatest model that they'll sell to consumers. Others have started at wanting to build a lab and have decided to become more of an infrastructure provider. So it's all very interesting. But I think the key point here is you have Anthropic, OpenAI, Meta, and I think Google, as well as SpaceX, looking to be building AGI, superintelligence, intelligence greater than that of a human, and that is continuing to be their North Star. So you have these 4 businesses or so that are competing and believe they can get there under 2 years. So it is our belief that because they're stating this publicly, that they'll still be pedal to the metal trying to build what they believe is possible within the next couple of years. So I think that's an important point as it relates to this cycle. There's a lot of news, a lot of noise, a lot of volatility, but we have these companies that still seem committed to a North Star goal that requires a lot more compute and a lot more money.

You've already mentioned it, but it just seems like, again, the way things move and the scale of this, it must be just an absolute heyday for an active investment manager. Like you say, if you've got things modeled right, you understand, you've got a way of determining what is the actual value. You've got a model for where this can go. It just seems like an environment where you can do particularly well, relative to an index that is going to be is asked to bet on everything.

I think that's right. And I think we pride ourselves in changing our mind as the facts change. And there is a pocket of the market that we've changed our mind on as of late. And it's a fairly significant pocket of the market, and that is these hyperscale businesses. I think we spoke that earnings have been revised higher from 12% growth to 30% growth this year. For those businesses, earnings have also been revised higher. However, the multiples have compressed quite notably from the start of the year to today. And that's because for the first half of the year, even last year, perhaps even the year before, the market did not necessarily believe this was capital well spent as it relates to this AI buildout. We are now increasingly of the belief that it is capital very well spent, with a huge caveat that demand will fill supply. But if you just believe in that, this is capital very well spent. And what we're seeing is these businesses becoming effectively utility providers of compute. The thing that everyone will need if this becomes ubiquitous, they will be the sellers of, and they will be able to charge similar margins, we believe, selling compute for AI as they do selling compute for any cloud service today. And that is quite the margin. For some businesses, it's 40% operating margin with 50% incremental margins that we've seen this quarter. We think it's a very difficult business for anyone to try to compete with them because you need scale. You need scale in terms of data centers in the US, but also geographically. And you also need a custom ASIC chip, I do think. And that is a very difficult proposition to go at. They've been at it for 7, 8, 10 years, some of them. So if someone wanted to try to compete with them today, they are that far behind. So we are increasingly of the belief that the winners of this all, whatever this looks like in the end, are going to be the utility providers of scaled compute. And it's just the hyperscale businesses. And they won this battle long ago because they got into the cloud long ago. So that would be one aspect that we've changed our mind on recently. I'm not even sure if you asked me that, but I just wanted to share that because I think it's an important nuanced point that is playing out in the stock market.

But you're not worried, because they're enormous generators of cash, that they move from spending their own money to spending other people's money in terms of that build-out by going out and borrowing? And any of the concerns around finance across this this space?

Financing is definitely a concern. The buildout of this AI infrastructure is of such scale that the financial markets or the borrowing aspects have gotten a little prickly, to put it politely, less so from the hyperscalers, but increasingly so from the AI suppliers. Funny enough, the semiconductor stocks, some of them at least— NVIDIA being the most notable example recently— are backstopping some of the customer purchases, some of the data centers being built. They're effectively saying we will backstop this if it doesn't necessarily work out. We're that confident in it. Which to me is in the same zip code as a price decrease for their customers. But as it relates to the hyperscalers themselves, as you say, they're fairly cash generative. At least their core businesses are. And they have relatively unlevered balance sheets. They have issued a fair number of debt recently. They will continue to do so because they need to do so to fund this buildout. One has issued equity. We expect others will start to issue equity also. And that can be a scary thing, obviously. But if we go back to Andy Jassy's comments: 2 years to build the shell, 30-year useful life, 3-year payback on a 5-year or 6-year useful life guts of the data center. Once we reach steady state of this data center build— and no one knows when steady state will be, but we model it to roughly 2029, 2030 or so— then you don't need to build shells as much anymore and you can refill the existing ones that you already have. And then these companies, these businesses, these AI data centers, token factories as we sometimes call them, become self-funding mechanisms to the extent that AI is ubiquitous and demand does fill supply. So yes, funding is a concern. We're not concerned about this year, next year, or the year thereafter as it relates to tapping the requisite amount of capital to fill their needs. 2029, 2030 gets a little bit more hairy in our math. But if we're right on demand, then they should begin to start self-funding some of those data centers. So that's how we think about this. Now, there are a tremendous number of assumptions that go into all of this, and we could be wildly incorrect on some of them, but I feel like we're relatively conservative on some dynamics. So we feel good about the hyperscalers in terms of supply, funding, and if demand shows up, those businesses will really be humming.

Wow. What about what we hear? We'll get a headline from time to time about what's going on in China and the way that they've attacked AI. And this is a battle between the US and China, I guess, from a geopolitical standpoint. But is there anything that's going on in China that's could disrupt your view of how this all plays out?

Potentially. China is approaching this from a position of being behind in all of this. And when you're behind in something, sometimes you don't necessarily play by the same set of rules. So they're doing some things and they're being pretty darn scrappy in a good way, I guess, in trying to catch up. So first and foremost, I'm not sure if anyone on other podcasts has explained distilling a model, but what they're effectively doing, or some of the labs over there are copying what's being built by Anthropic or OpenAI. And if you recall, Anthropic could spend $20 billion to train a new model. A Chinese company could distill that model, copy that model using a variety of approaches and come out with a model that is not on par in terms of quality, but not far behind and effectively give it away for free. So that's a scary proposition for Anthropic. Now, I am of the view that that will continue. It's very difficult to stop. Maybe the US government will try. I'm not sure. I would love to see the US just fight head-on with China. I'm not sure if you've heard of this debate in terms of closed source versus open source, but effectively they're giving away for free open source. Claude, we don't know necessarily the ingredients in the recipe. Closed source. What I'm more concerned about, specific to Anthropic and OpenAI, is US open-source models and the one company that is likely going to spearhead this approach. This is news as of a week ago. Meta is making tremendous progress in terms of their quality of their models and they're pricing them at a very low price, effectively taking no gross margin on some of their models that are very good, not the best, but pretty close to the best. And the pace by which they're advancing is, I think, exceeding people's expectations. So I think the one company that could really shake things up for OpenAI and Anthropic, who sell tokens at a premium because they have premium models, is Meta. And why is Meta doing this? Well, Meta has effectively nothing to lose. They don't have a hyperscale business today. Remember, Google, Amazon, Microsoft have hyperscale businesses, and a good chunk of their RPO, their backlog, promised future revenue for future services, roughly 40% of that, we think, is either or, and OpenAI and Anthropic. So you have two companies that are a good chunk of future demand at those companies. That's a scary proposition for those hyperscale businesses if OpenAI and Anthropic all of a sudden go into a price war with Meta. Meta does not have a hyperscale business. And secondly, Microsoft, Amazon, and Google also in varying degrees have equity investments in OpenAI and Anthropic. Meta does not. Meta has nothing to lose and everything to gain by going into a price war— instigating a price war, I should say— and trying to steal share that way. And there has been rumors that they're going to start a hyperscale business. I think it's not a bad idea if they did. I think the stock market would applaud it. But yeah, there's a lot of things going on. Price wars, conflicting channels, customers. It's all very interesting and it's playing out in real time. But I still come back to this point that I made at the beginning of the podcast, and I said a couple incremental times, this is one massive bet on demand coming to fruition. The way we use ChatGPT and Claude today— most people, I should say— is not entirely dissimilar from how we use Google search. We need as an economy, as a capital cycle, to be rolling out agentic agents, consumer agents. We need these things working 24/7, doing tasks that we don't like to do because they're too mundane. That is what we need built, and enterprises are going to be the ones that need to build it. So there are some demand bottlenecks that need debottlenecking for this to all work. So demand is still something I'm laser-focused on.

As we walk through this, we keep going through all the different things that could shift. Who leads? Who follows? And the prize is enormous. We can agree on that. You can see why there's so much volatility and why it's so hard if you're buying individual names of these stocks and you don't have the expertise, the background, the information to really know what you're doing. The scale of what you need to know to be able to play in this space without really just effectively making bets instead of investments is really incredible. And to think that I might go out and try and buy a stock and go head-to-head against you and all the information that you have— or a bot, by the way, but I'll take you over the models— is kind of a scary proposition. Because when you're buying and selling, you really know why and what, whereas a lot of people, I imagine, who are trading these stocks or buying these stocks aren't really that sure exactly what they're buying and why they're buying it, in a real sense.

Yeah, I think that's right. We're in such a fortunate position in terms of access to varying opinions from very smart people. But also we get to ask some of these companies these questions. I'm going to San Francisco in 2 weeks. I'm going to meet with a whole bunch of these companies and get to ask them different questions. So we're super fortunate here that we have that. So, yeah, I would agree. It's tricky. It's tricky for me. I imagine it's tricky for anyone.

Yeah, there's just a lot to piece together. I've been out with customers, I'm going to be out with investors this evening. I'm sure they're going to be interested in this topic and trying to make the comparison between the build-out of the internet and the build-out of AI. We knew 25 years ago or 30 years ago that the internet was going to be big. And there were companies spending a lot of money on it. And when the market was most optimistic, you'd see these big runs in these stocks and a big run-up in the market overall. And then you could get pessimistic. And you'd see that volatility. And then ultimately, even though the internet became something that was life-changing— anybody who's sitting listening to this podcast, as they're sitting on a subway, looking on their phone, watching it on YouTube. By the way, subscribe and give us a glowing 5-star review because David Tron deserves a 5-star review for all the content that he brings us regularly on this podcast. So the internet is a real thing and everything that built out around it. AI is a real thing. This is going to change the way we live. But there's going to be some ups and downs along the way. We had that big hiccup between 2000 and 2002. Is that something we need to worry about? Or just because we're dealing with companies that are just so much bigger and just make so much money this time, you're not going to see that?

Of course, we worry about a lot of things. That is certainly one of the things we worry about. We are building supply and the companies that are building it are hoping demand fills that supply on a relatively steady cadence. Adoption of this is impossible to forecast, and perhaps it happens quicker, perhaps it happens slower. It's difficult to say. The internet buildout is not a bad analogy. The analogy we prefer to use here is the cloud buildout. Before the 2010s, a lot of companies just had data centers in their own backyard. They ran them themselves. It was costly. It was not necessarily the most efficient way of doing things. And then these hyperscalers came along and said, we're going to build data centers, and you'll be able to rent a piece of our data center and you'll do it effectively over the cloud. And that buildout happened not without its set of skeptics, I should add. That was also very expensive and the scale was quite enormous and demand continued to exceed supply throughout the 2010s. So I think that is an interesting analogy because this AI buildout, at least for the hyperscalers, is an extension of the cloud buildout. That's how we think about it. It's really tough to know who are going to be the winners of this on the lab side, the software side. Is it going to be OpenAI? Is it going to be Anthropic? Is it going to be Meta? Google? It is so tough to know. But one thing I think I have reasonably good confidence in— again, if demand fills supply; a big «if»— is that the providers of the infrastructure, the scaled compute providers, the cloud companies are going to be the ones that will facilitate that infrastructure, I am fairly confident in saying. And relative to the internet, we have real-world use cases today generating real revenue. The Anthropic number that came out today in terms of ARR was $65 billion of revenue. Again, that's last month times 12. That happened in very short order. So there's clearly demand today for real dollars filling that supply. Fast forward 3 years, what will it look like? That's where it gets interesting. No one really knows. But if you believe in the technology being of high utility to knowledge workers or populations as a whole, I think you can believe that society will find use cases for this and generate a return for those providing that technology. I think it's as simple as that. But I am not going to sit here and pretend I know what it's going to look like 3 years from now. It is remarkably murky. The amount of capital that's being spent without a very confident view on what demand looks like is something we're going to look back on the history books and wonder about whether it works or it doesn't work. That is what we're going to look back on, I think.

I think someone was saying— and you can correct me— that adjusted for inflation, the build-out of the railway networks back 100 or so years ago was on a similar scale. And you see ultimately how that transformed life and made many millionaires back then. It's now billionaires and trillionaires nowadays. It's just at a completely different level when we start adding the zeros on after inflation. David, what's the question you're most excited about asking when you get to this conference in San Francisco? And by the way, we had David on last year after the conference. So this year we're getting him on before and after. We're going to get him on when he comes back next month when he's got some time. But what is the big question you're going to be asking these companies when you're there?

Well, it depends on the company. But I think as a stock market, we need a handoff to take place inside of the stock market from the companies that are benefiting from the CapEx dollars. So call it NVIDIA, Broadcom, all those guts. We need a handoff towards the companies that are truly harnessing this technology for internal productivity optimizations, ways to make their company run better than it was before. I have a theory, and I'm not the only one, that inside of certain profit pools, the companies that do this better, harness this better than their competitors, will be able to steal share from those competitors. So I really want to dig into outline your use cases, be very specific. How is it transforming the way you guys do things internally? How do you think you're doing this better than your competitors? I really think over the next 2 or 3 years, a lot of alpha could theoretically be earned by owning those businesses who are doing a very, very good job of using this technology internally. So on that respect, I'm going to be asking a lot of those questions. And then secondarily, I really want to understand demand. Where do these companies get the confidence to continue this buildout when it's so difficult to forecast demand? Help me understand your forecast. That's really the crux and the heart of this whole thing. And I want to hear an answer that is more than «today demand exceeds supply». No, that's not good enough. I want to hear 1 year, 2 years, 3 years from now, how are you so confident? Because there's a lot of capital at stake, shareholder capital, our unitholders' capital is at stake. So I'm not going to leave those rooms without getting a good answer to those questions.

Well, you know what? I actually could have guessed that question just based on listening to the podcast so far. That's how clear you are in explaining all this stuff. I just love having you on. I know the audience loves having you on because we just learn so much about this. It's talked about in one way in the media and a lot of the nuance of it is just glossed over. And we get into the guts of it with you instead of just thinking about it from the perspective of maybe the shell. And that's what's fantastic. So, David, let's finish off, let's just step out broader market because you're not just involved in technology. You're looking at the broader market. And so, we're coming up on now, we're going to have 4 years of pretty outstanding performance across stock markets all around the world. How long can this continue? Is it just AI, or how long can this go on? I don't know if we've ever seen this level of performance this long. Maybe the late 1990s were along these lines, but it's an unusual period.

So a lot of it is AI. I'm not going to sugarcoat it. There is a good chunk of the earnings stream that is definitely AI that is likely unsustainable in certain pockets. However, the average stock in the S&P 500 is still having a good year. And in fact, many businesses that have nothing to do with AI are also having a very good year. So I think it's important to understand what is going on there. Why now? There's a couple of dynamics at play, I do believe. Remember, COVID happened 5 or 6 years ago. There was a huge overbuild in certain pockets of the economy. As an economy, it took this many years to digest a lot of that demand. In certain pockets, we are only now just escaping some of that demand digestion. So demand has met supply finally last year and this year in certain areas of the stock market such that we can now grow again. So I think that's a very important point. Secondarily, the tariff dynamic that has played out relatively recently under the current administration has really incentivized businesses, or at least caused them to think very deeply, where do you build your next wave of capacity? Are you going to do it in Asia? Are you going to do it in Europe? Or are you going to do it in the US? I think what is happening is a lot of companies are playing it safe and leaning towards building that capacity in the US, which is an accelerant to growth. And then I think another important dynamic is under this current administration also, they came with a tax bill last year, the one big beautiful bill that incentivized from a tax perspective all sorts of positive dynamics in terms of building now as opposed to building 5 years from now. So I don't think it's just one thing, but you have a confluence of factors. Again, nothing to do with AI, that is just creating growth and stimulative growth inside of the economy. The consumer is still bifurcated. It's K-shaped. The more well-off folks are having a better time than the less well-off folks, and we see that in company earnings and whatnot. So we suspect that will continue to be a trend, although we'll see what Donald Trump has up his sleeve when it comes before the midterms. But nonetheless, I think there's a confluence of positive dynamics that are playing out for now as an accelerant to growth. The midterms will be important. Most important will be the elections 2 years from now where you could get a changing of the guard. It's unclear. All that being said, we're laser focused on the 10-year, the 30-year inflation. That has massive implications for what we should be valuing equities at going forward. So it's not all sunshine and roses per se. There are risks galore, but there are certainly confluences of factors that are an accelerant to growth outside of AI inside of the US economy right now.

Yeah, and another factor driving what's going on in the market today is you've got a 30-year Treasury in the US that's at a 19-year high. You've got the 10-year, which we talk about quite a bit on this podcast sitting above 4.7%. And we had Eric Lascelles on a couple of weeks ago talking about, yes, it should sit in a range between maybe 4.40 and 4.80% for a bit here. But whenever you start to creep towards the top end of that range, it's always a bit nerve-wracking. Inflation numbers are mixed. But this war drags on, and that's really the thing that I guess none of us can predict, exactly how that plays out, because it's taking another turn over the last few days that I'm not sure any of us would have predicted at the front end.

That's right.

David, I think we'll maybe end it there. We've used up a lot of your time, and we always appreciate it. Have a great trip to San Francisco. I know how excited you get about going to this conference. And again, the one thing I love about this podcast is we can just bring on investment professionals like you and give a sense to investors of the types of people that are involved in this business and the passion that they have for just digging deeper that I think any of us would dig as we're making investments. And it just brings a sense of what a true professional is in a particular space. You're certainly a great example of it. So, thanks again for your time and have a great trip and we'll see you in a month or so.

Love it. Thank you, Dave.