The one thing that happened in the third quarter, just before I start sharing slides, is obvious—not obviously, but to me—the rise in interest rates, especially the 10-year, and that to me is squeezing the rest of client portfolios, because now yields are getting to be relatively attractive. Especially, you know, like high-yield munis and things like that are in the high single digits. That's competing with annualized equity returns of 8% a year.
Welcome to Thoughtful Money. I'm Thoughtful Money founder and your host, Adam Taggart. Welcome you here for one of our quarterly updates from Jan van Eck, the CEO of VanEck. Jan's getting really tired of seeing my face, because he's been giving us a lot of time to the Thoughtful Money audience. He just spoke at great depth at this past weekend's conference about what's going on with higher bond yields. He then came in for Q&A, and he's back again this week to give this quarterly outlook and give us a sense of what's going on in AI. It's very interesting though. It's going to be both on the public side, which is what most of us see and read about, but also we're going to be talking about what's happening in the private markets related to AI, and Jan has brought one of his esteemed colleagues in for that. So Jan, thanks so much for joining us today.
Thanks, Adam. And just to remind people that, you know, what I try to do is I sit with our investment personnel here at VanEck in Midtown, New York, and we see thousands and thousands of charts and slides and data points. And what I try to do is, consistent with our firm's outlook, select slides that maybe are somewhat unique or hopefully provide an interesting perspective on the markets. So, you know, that's the goal. And I think, you know, excited to have Christian join me this quarter. The one thing that happened in the third quarter, just before I start sharing slides, is obvious—not obviously, but to me—the rise in interest rates, especially the 10-year, and that to me is squeezing the rest of client portfolios, because now yields are getting to be relatively attractive. Especially, you know, like high-yield munis and things like that are in the high single digits. That's competing with annualized equity returns of 8% a year. So to me, you know, what we see is a slight shifting of people's portfolios towards fixed income, and that's what we've seen in our funds over the last month or so—that four out of the five top inflow funds have been fixed income funds here at VanEck. But let me pull up the slide deck. So that's why I call the bond squeeze this quarter. I don't know if you like that title, but you know, it's the fact that bonds are now attractive, so they're sort of squeezing other parts of your portfolio.
So the kind of perspective that we bring to the market at VanEck is what I've started calling 10-year macro. So in trying to understand the markets, we start by asking ourselves these questions. So number one is fiscal policy stimulative or not stimulative, or is it contractionary, and is it helping the markets or hurting the markets? What I would say is government spending is still high and is still at least one foot on the gas pedal. I would say we had two feet on the gas pedal a couple of years ago post-COVID, but it's still pretty stimulative. So that is probably not going to change with the midterms, because it's unlikely, I would guess, that the parties would agree, because if the Democrats win control of either chamber, the parties would have to agree to change fiscal policy, that they're going to change it, agree to contract it. Hopefully they don't agree to spend more money. Monetary policy also is pretty much locked in. You know, we said this at the beginning of the year. But it always has a big effect on the markets. You know, what I would just say is I view monetary policy as kind of neutral now. I will say that there's quite a bit of disagreement at VanEck about that. But I think if you look at it, mortgage rates are definitely squeezing the housing market. I think we've seen a number of failed IPOs in Asia and in the US over the last two weeks. And the Fed raised rates. So I'm not saying that it's supremely contractionary, but I find it hard to describe monetary policy as stimulative right now. I view it as kind of neutral, but net on net, government is still sort of stimulating the US economy, which I think is helping, you know, at least short term with a number of different things—technology, we'll talk about, regulatory as well. I don't anticipate a lot of changes coming in the next couple of years despite how the midterms turn out, but the noise will increase obviously.
Okay, so today what I wanted to do is dig into AI a little bit with the theme that this mega trend continues. There's a lot of nitpicking, I would call it, on different facts in the AI trade, and we'll walk through those quickly. Some of the big changes are definitely happening. It's a changing ecosystem. The big picture is the same. There are different points of the ecosystem that are feeling pressure, and they have before, and we'll talk about that. And then we'll get the private market perspective. We are structurally underweight bonds and fixed income because of our deficit concerns. But you know, they're more attractive here in the shorter term. So in our models, Adam, we've started buying bonds. But that's kind of a thing to dig into, and you know, we did that last week. Want to revisit real assets as a structural overweight that we at VanEck have, just update on India. So yeah, so that's kind of the overall agenda for today, if that makes sense.
Sounds like a great agenda. Can't wait to dig into it.
Okay. Well, hopefully these AI slides are going to be slightly different. Some will be obviously familiar. So we haven't yet hit earnings season, but it's worth just kind of quickly going through the overall numbers. We are in a tech-dominant equity market, and you know, I guess what I want to push—I'm going to get on a soapbox a little bit here, Adam. You know, a lot of people, including some of our mutual friends, you know, are really negative on either Nvidia or parts of the tech ecosystem. And I just want to remind people as they think about their portfolios, the normal equity benchmark, the S&P 500, is very much more tech dominant than it used to be. We at VanEck are actually overweight tech above that. But think about the reverse. So if you listen to the naysayers or stepped out of this mega trend trade of tech, you are leaving a lot of money on the table. So it sounds smart, but it's been very expensive for you year-to-date over the last one-year and two-year period, sort of since the ChatGPT moment. So, you know, I just wanted to kind of emphasize that aspect of it. When we talk about AI, we're basically trying to feel comfortable with the equity markets as they are today and comfortable with a slight tech overweight, which we do. This just puts the rest of the market in perspective. Obviously, energy has got a big year-on-year return.
And then, specifically the last column, expectations by some of the giants in the market. What you see here is a slight slowing down in year-over-year earnings growth, and a little bit less even. I want to stress this excludes their one-time capital gains, because Christian's world has been contributing to the quarterly earnings of public companies as they mark up Anthropic and SpaceX and other holdings. But I take that out, right? This is not including one-time gains, which I think are real numbers. Nvidia is obviously leading the markets here.
All right. And Jan, I just should interject. We didn't really ever formally introduce him, but when you mentioned Christian, you're talking about your colleague Christian Monofo, who's the head of private growth strategies there at VanEck.
Yeah. Well, I'll give him a little bit more, but sorry. Thank you for doing that. So, over the last two years, Adam, I've used simple hand gestures to define the AI trade, and I say demand is here and supply is here, right? And we can look out for two years. Don't overthink it. What I'd say is our analysts and our colleague Angus, who digs into this and has helped support our kind of 10-year belief that AI is one of these mega trends that will distort and lead the financial markets, is starting to say, okay, I don't think of the world only as being just supply constrained—like how much TSMC or how much the memory companies can produce, etc., etc. It's—I can now get enough data on token usage to model the demand side of the equation, and that is how we normally look at projecting earnings growth, right? For energy companies or consumer companies or every other kind of company. So this sector of the economy is starting to transition a little bit from this like crazy early jump-start phase of shortages on the supply side—and there still are shortages in supply, don't get me wrong.
But I'm just saying the new voice or the new harmony in the orchestra of AI is looking at demand as well. I would say the only thing—the black part here is what's called inference. So a lot of the use of compute so far has been the gray, which is the AI frontier labs, or the Mag 7 frontier labs training their models, right? Because when Jan or Christian asks an AI chat, you know, what do you think about something in the financial world or whatever it is, you want a fast and accurate answer, and that's what everyone has been racing to provide users, and that is—they're spending a lot in R&D. Looking forward here, the projections are—and this is an important assumption—is that a lot of that will be the use of trained models, so the way we use it normally, I guess I would say, and that's called inference. So anyway, I'll pause there to see if you have any questions.
None, except to say it looks like, you know, demand is going to keep growing a lot—five times from today. I don't know anything about the technology here. So this is a very naive question, but how safe is it, an assumption to make that training and inference are just going to be flat for the next four years?
Well, we think training held flat is a conservative assumption, right? So we're trying to be, I don't know, reasonable, if you will. As long as those companies continue to grow, I would like—I probably we would say that it would grow, but just from a modeling perspective, we just are going to hold that flat.
Okay. So, in other words, if you're wrong, you're wrong to the upside on demand. Got it.
Correct. And then on the training side, I think the important insight—forgetting all the complicated text on the right—what Angus did, and you remember we shared token usage at VanEck, is he studied individual by individual at VanEck their token usage curve, their adoption curve, and what he found is like, you know, from your first two months to 12 months later, your token usage goes up like eight times, and then after that it goes up at sort of like 25% a year.
So all he did is use that, I'll call it, model from VanEck data and transposed it onto the rest of the market, and he ended up with exactly the same number. It's a little uncanny, as the industry projection, which was in the green line. So, look, this is all subject to change, and obviously we will revisit this quarterly, but I thought it was interesting. His main point is you can start to model demand, right? I mean, this is guessing at the future and not certain at all, but it's just kind of nice to have things line up. So I thought that was an interesting perspective worth sharing.
Super useful, and it lends support to your outlook, which is that the mega trend is likely to continue for the foreseeable future here.
Yeah. And listen, actually Christian has posed the question well, and I'll turn the mic to him in a couple of minutes here, but when will the, you know, kind of AI revenue curve start flattening or slowing down, right? We know it's heading up and to the right. And I think that's what investors are really worried about—like, is somehow these growth rates go to, I mean, god forbid, go to zero. That's probably not the right question, right? But anyway, I'll turn the mic over to him because he has his own perspective on that. But anyway, so here we go. Within the AI mega trend, you know, I think there's a newer way for investors to start thinking about this. I'm going to go really quick on these charts. Year-to-date, what I call the AI 1.0 trade has been the winning trade. So this is just a price of Nvidia. I guess it's not—we didn't label that sort of last minute. Oh, sorry, this is a semiconductor index, but you know, we'll see Nvidia in a second, but Nvidia is up, or semiconductors are up 50% year-to-date. Nvidia itself is lagging a little bit, but I mean, one wouldn't say that its uptrend is broken by any means. It's up about 16% year-to-date. And its valuation has come down dramatically. It's really been—I guess maybe 'de-risked' isn't the right term, but it's become a lot more attractive, and we've talked about that before.
But now compare that to—I guess the point I would make is within this mega trend there's a lot of price discipline in these public markets. So there were a couple of theories out there that, you know, the new world of AI compute needed the old world of electricity generation and data center buildout, metal demand, all that kind of stuff. And there's been a big, you know, correction in those trades this year. Partly, I think they were overvalued. So like utility companies that, you know, seemed to be perfectly positioned were trading above valuations. Here's a second company, Vistra, which is an independent power producer. I think again, core business, they have the contracts—long-term contracts—are in a really good position of having energy, which is what people need. Whether it's data center concerns or whatever, the prices have just come down to more attractive valuations. So, you know, it's just another part of what's happening in this ecosystem in the public markets.
And then, you know, the poster child of last week, which is CoreWeave, which is—I would call it the compute providers with less robust financials. I'll call it that. And again, I would call this maybe a reflection of the bond squeeze of Q3, which is higher interest rates. But my point here actually is that, and we talked about this literally a year ago, in the fourth quarter of '25—now the fourth quarter of '26—we saw a lot of price discipline on CoreWeave and Oracle where they both fell 50%. So that was this fall, you know, a year ago. So yes, it feels painful and it makes for great headlines, but it's the same harmony or same, I guess, bit of music that was playing a year ago.
So that's just kind of the... And then, you know, maybe another interesting data point, right? AI, we know, and we ask ourselves this all the time, is going to be disrupting different industries, and the first two industries are the software industry and then also, I would call it, the adjacent consulting industry, right? Which is, you know, implementing software projects at big companies, and Accenture is a leader in that, and you can see how its stock got destroyed. But actually it turns out that AI doesn't unemploy the whole world, right? And specifically it doesn't unemploy the people who are directly related to software, because as I say, AI has three components: data, software, and people. And now companies are pivoting their shift, and VanEck is as well, towards data, because if you don't have clean data, your AI is going to be garbage.
Right, and so I think this is again my point is there is a lot of price discipline and rotation within this mega trend trade, and we just say, hey, be diversified. Don't, you know, don't put all your hopes on one company, and own the 1.0 plays, the 2.0 plays, even though it'll mute your overall returns. You know, we think that's a good way to play this AI trade. So now I'm going to introduce Christian, who runs late-stage private funds for VanEck. We launched our first fund in Q2. But more importantly, he looks at this phenomenon from a different perspective. So, Christian, let me just turn it over to you, and you could—I'll flip the slides, and you can make your observations.
Thanks, Jan. Adam, great to be here. There's so much going through my head right now based on the slides you talked through, which I think is why it's so interesting to get the public and private market perspective, because there's just so much overlap. Like I was looking at your slide six, the Angus slide, and when you see that projected, expected surge of inference utilization in the private market, what we're seeing is the valuations of these inference clouds just skyrocketing, right? So like you could look into where things are moving in the private market, and sometimes that becomes a tell. By the way, sometimes it blows up, right? It doesn't end up being like a long-term sustainable durable trend. The other thing, you know, we always talk about, as you said, the old world building the new. That's absolutely true, but the new world we believe is also going to make this whole environment much more efficient, right? Whether it's like next-generational computational and chip technologies, and like photonics and quantum, whether it's more efficient interconnects and switching, right? So we can better utilize capacity which is not currently being fully utilized, less power consumption. So there's a lot going on in the private market that I think is directly overlapping with this. So I'll get off my soapbox and maybe focus on a couple slides, but I wanted to just kind of share that.
I think this slide here is something that we talk about a lot here, Adam, at VanEck, which is, you know, you have this ongoing debate around the long-term durability of these frontier closed models, right? The Anthropics, the OpenAIs of the world. And the debate is things are increasingly shifting, and token consumption is increasingly shifting to these more generic open models that are quote-unquote 'good enough', and they're good enough, and they're also substantially cheaper.
I was going to say they're way cheaper, right?
Exactly. They're way cheaper. And so this data was really interesting because, one, I think it validates that there is this active shift in terms of token consumption moving from these large frontier closed models to the more generic 'good enough', you know, much cheaper open models. But at the same time the majority of the revenue generation continues staying with these frontier models. And why is that, right? One, think about the average user of these frontier models, these closed models long term—they're from firms like VanEck, right? They're enterprises that want to have the ability to do, you know, advanced coding and software engineering and forecasting and, you know, agentic workflows. Those high-impact, high-value use cases come at a price. It also comes with security, right? The enterprise-level solutions also need security. And so the switching costs for VanEck, for other enterprises, is quite high. And so the ability to kind of switch for a firm like us to these more generic open models which are in many cases quite competitive—probably six months or so behind, based on, you know, research; some say it's four months, some say the open models are about six months behind—but still far enough behind where those high-impact, high-value use cases are willing to pay a premium, if you will, over the open models to be able to access the capabilities of those models.
Now again, this is something we're constantly monitoring. I think frankly, look, you look at Anthropic, right? I think the numbers you probably saw, Adam, their last quarter, they put up 16 billion, almost 17 billion in revenue for a company that just started generating revenue a few years ago. We honestly can't look at each other and say we have full conviction in 10-year outlooks here, because we don't. But I think so far, right, this debate continues to show that while the usage is switching, the revenue generation, the high value, is staying with these large frontier models.
So just to walk through the slide really quick, right? The slide on the left shows in purple the open-source usage in terms of consumption of tokens has shifted up to over 70%. On the right, the fainter line shows that the revenue share from Anthropic and OpenAI is still over 70%.
Yep. Yep. Thanks. Thanks for pointing that out. And I mean, I think it makes sense to a layman like me that even though the share of tokens is skyrocketing for the cheaper models, the share of weekly spend of the frontier models is decreasing at a much slower rate. But that's because they charge not just more, but a lot more than these free models, right? So the free model, you would need to use a lot of free tokens to equate to a small number of frontier tokens just cost-wise. Correct?
Yeah, I think that's right. And I think, look, over time the frontier models are going to continue lowering their cost to be more competitive, right? The example is, look at what AWS did in terms of increasingly decreasing. But that doesn't mean, though, it's not going to be a sustainable business from an operating metrics and performance standpoint. But yes, over time, I think we can expect to see pricing come down to become more competitive. But if you also have broader adoption, right, there's an offset there, like we're just starting...
Make it up on volume. Yeah.
Yeah. We're just starting in earnest, right, to really look at what the capabilities are of this technology. It's the early innings, and the entire market's clearly not using it yet.
Right. So, quick question on the lag that you talked about. The frontier models—sorry, the open-weight models—you said are basically a half year, four months to six months behind the frontier models. Over the past year, has that lag been increasing or shrinking? Like, are they getting faster and faster to be able to copy what the frontiers can do, or are the frontiers extending their lead over time?
It's interesting. So there's a nonprofit organization called Epoch AI that kind of tracks these types of trends, Adam. And it's been fairly consistent, to be honest with you, in terms of that four-to-six-month gap, because while the open models are improving, guess what, the frontier models are as well. And to some extent, they're likely not releasing the most powerful models that they currently have, for a variety of reasons that we all read about.
One point I wanted to make, Adam, because I spent one day last week at a conference with just individual investors, and they don't live inside a corporation. And so I wanted to describe Anthropic kind of in a completely different way to them, which is, if you think about Jan's desktop at work or Christian's desktop at work, right, it's valuable real estate. What they say in the real estate industry is it's all about location, location, location. So if a software company like Anthropic can get that positioning on our corporate desktops, that's really valuable. And that's what everyone is fighting for, right? And I tell them that I'd rather go to Anthropic, I'd rather go to Claude as my interface, even for my Microsoft suite, to ask questions about my calendar or my email or all of that kind of stuff. And so that's the kind of fight that's going on. But that's why there's that value to Claude, and what that means is like Claude is connected to the different databases at VanEck. Now, could we replace it? Absolutely. But you know, people just—it takes time, and then people say like, 'I don't want to spend that time just to learn software that's the same.' So anyway, I just found that people that don't live in a corporate environment don't quite get that dynamic of the real estate, if you will, of the corporate desktop.
Sure. But I mean, we get it, right? It's like you were saying, there's a switching cost to VanEck at the enterprise level in, you know, pulling out one AI and bringing in another, whatever. But it's the same at the personal level. It's like, you know, I have an iPhone. I'm just so used to Apple products now that, yeah, maybe there are Android apps that do a better job, but I'm just so accustomed to this. The thought of just relearning it all just feels like too big of a hurdle for me. Yeah.
Yeah. Yep. It's a great point. So, why don't we move on, because I want to be efficient and not take up too much time. I mean, we'll spend a couple minutes on this in the next slide. I think this is pretty much us just trying to make sure that the audience is aware. Because it's interesting to me. I've been investing in private markets for over 20 years, and you'd be surprised how many folks just really still have not yet learned about how significant of a shift there's been between public and private markets over the past few decades. And if you look at the number of US companies on listed exchanges, that number has essentially been cut in half over the past few decades, while the number of private companies has soared to 50, 60,000. And there's a lot of reasons for that, Adam. There's regulatory changes that have happened. There's administrative reasons that companies will look for ways to extend the runway before they have to go public. There's also been a massive unlock of capital that's been made available to private companies over the past few decades. But what this ultimately means is that these companies are scaling in private markets into much larger and larger market cap companies, quite similar to what many investors used to look for. Frankly, when you think about a small-mid-cap growth level of exposure, to be honest, a lot of public companies I think would kill for some of the operating metrics that we see the companies that we look at generating.
So you have this massive surge in value. There's roughly around 11 trillion dollars in total value today, just looking at US venture companies. That's a big number. And the challenge is that there has not been enough exit activity to be able to recycle that capital back to investors. So we have this glut we call in the private markets, this TVPI glut. TVPI means total value to paid in. So you may, as an investor, Adam, you may have invested a dollar. It may be worth $4, but if you did not get any of that money back, what real good is it at the end of the day? So you have this TVPI glut, which this slide basically talks to. And so what we really look at in the end is what we call DPI, so distributions to paid-in value. And what we're seeing is that on a historic basis, that ratio is actually quite low. And there's a lot of reasons, right? We had, you know, COVID, we had the interest rate hikes from 2022 to 2024. The IPO window in our view is far from open. We've had obviously SpaceX get out. We'll see if Anthropic gets out, but most of the exit value we've had is generated by a select few companies. So we have this supply-demand imbalance. So on the one hand, you know, we think it's really important for investors to understand that more and more economic growth is happening outside listed markets, and you should really think about ways to access them. There's increasing accessibility, which we could talk about at a different time, in terms of how the average investor can gain access to these types of strategies. But you also need to understand that these strategies typically are illiquid. And so you do need to have a longer-term time horizon to benefit from what this type of an asset class brings to the table.
Christian, can I just put a pin in this, because I think this is such a really great opportunity. And so folks watching, if you'd like us to do a deep dive on how to invest in private equity as a regular retail investor, let me know in the comments section below, and if there's enough demand, we'll make it happen. Christian, I'm saying that almost just to be polite, because I know there's going to be enough demand to do this. And I've talked to, in my talks with private investing executives like yourself over the years, the message that I've got is that the private markets are now so big and so deep that if you're a promising new company, you have no problem getting funded in the private space. And there's a lot of benefits to being private, right? There's a lot of restrictions that get placed on you when you go public. And so like the best and brightest new businesses are going private and staying private for really as long as they can. And so, you know, in years past, if you wanted to get in early on a new startup, you know, you'd hope it would IPO and you could get in early at a company like a Yahoo or early Google or whatever. Now, those companies spend way more of their life cycle in the private space. So if you're not in the private space—I'm saying this with a, you know, big generalized broad brush—but you're kind of getting sloppy seconds if you're just a public market investor these days. You're nodding a little bit as I'm saying this, but am I capturing the spirit of what's going on?
Yeah, you're spot on. You could do our pitch for us. So, you have an option.
You'd probably come up with something a little nicer than sloppy seconds, I'm sure.
Yeah, but I mean, just some quick data behind that. If you look back a few decades ago to like technology venture capital-backed companies, the average company went public within about four years of inception, and the average market cap was about a half billion dollars. Those companies today are staying private for 12, 14, 15—SpaceX, right, 24 years. And they're going public now at market caps that are multiples higher, like multiple billions of dollars. And SpaceX is a right example, you know, over a trillion. And so that's value, that's return potential that you're missing out. And so, you know, you heard Jan talk about the pain of missing out and sitting out the AI trade. There's a pretty big pain of sitting out the private market trade. And so I'll kind of leave it there. By the way, again, there's illiquidity, there's valuation concerns, everything comes with risk, and we can talk about that. A lot of the assets now are priced to perfection, which gives us some indigestion. So you have to be really careful and really disciplined, but we believe it's a really big opportunity.
So you can switch to the last slide, Jan, and I think this one just quickly says that you have this large growing market and you have all these VC managers and private equity managers getting funded, but similar to the public markets, this power law remains true in the private markets. You have a select number of companies increasingly driving more and more of the value, and you have a decreasing number of managers that are driving more and more of the value. And so if you kind of look at it on this slide, this is just trying to give a visual picture of first-time fund managers. So Adam, Jan and I go out to raise our first-time fund. The current environment is telling us that we would be having a very hard time, because investors are looking at historical performance and they're saying, you know what, more and more of the returns are being generated by the incumbent VC managers, growth managers. We're better off parking capital potentially with those folks. And if you look at some data that we have, roughly 90% of venture returns over the past few decades has been generated by 5% of the managers. Right. That's an important number. And if you look at the 3,000 VC managers that have been active the past couple decades—maybe I'll ask you a question. Of those 3,000 fund managers, what percentage of them, Adam, do you think put up more than 3x returns consistently over that time period?
What percent put up over that? A tiny fraction, I would think, because my understanding of this industry is that the big dogs get first view, and so they snap up the obviously good deals, and then everybody else kind of increasingly is fighting over the scraps where there's a lot more risk.
Yeah. So you're spot on. The percentage is less than 1%. Less than 1%. So, you know, from our perspective—and there's other data we could talk about—but from our perspective, what this just tells us is that there's increasing value concentration at the asset level, increasing value concentration at the manager level. Again, we think investors need to or should explore how to incorporate private markets into your overall portfolio construction. But you have to be careful and you have to be disciplined.
That's all I got, Jan. I'll leave it. I'm going to call it.
Okay. I just want you to agree with me, and then you can sign off, which is: in the private space, the AI trade is still booming.
It is, it is booming. Unshakably booming, I would say. It's hard to wrap your head around the valuations in some of these situations for companies that have not yet demonstrated any durability, let alone... I mean, on the hardware side, even beyond AI, we have companies that are on the hard-tech side that are getting rewarded for demonstrating they've executed at scale manufacturing, if you look at how they're being priced, and they haven't—they have some prototypes that they've been able to create. And on the AI side, as Jan said, there was a company—I won't name it—but they were valued at close to $25 billion and they had not yet released their model. So it's exciting, to Jan's point. There's a lot of focus and attention, but those of us who've been doing this for a while also are being very cautious.
Okay. If I could just ask one quick question, and then we'll wrap up on AI. So to Jan's point, the appetite for ownership of these AI companies is just voracious still right now. Now, a word you guys have not used yet is bubble, and I don't want to get into that because that could be a much longer discussion. We've had a lot of discussions around this. But it does seem to sort of have a mania intensity to it. We'll find out if that's justified or not, you know, as time goes on. The question I have—and Christian, I don't know if you had a chance to watch Fred Hickey's presentation from my recent conference. I think Jan at least saw part of it. And I'm just curious, Fred, as you know, Jan, is quite concerned, I mean, about valuations today, but it's largely because of the quality of the earnings so far in this space. And I'm just curious, as you leave Christian, or as we get off this topic, Jan, what thoughts, if any, you guys have on the quality—the earnings for sure have grown tremendously here, astoundingly. Future projections are still really robust, but what thoughts do you have on the potential quality of these earnings? And it could be because of circular financing or a lot of the other issues that Fred brought up at the conference. Let's go ahead, Jan.
Well, I would just say the end consumer here—and that was the breakthrough of Q1 of this year—is corporate America, right? Anthropic plugged into corporate America, and corporate America is saying, 'I want to spend money on AI.' So yeah, there's little parts of the ecosystem, like I said, in the orchestra, the violin player may have broken a few strings, or maybe someone has a stroke or whatever, but you know, the overall trend is the same. And there's a lot of discipline in the public markets, and probably I would say—and I don't have a metric, and I didn't ask Christian about this ahead of time—there's more discipline in the public markets, I think, in a way, than in the private markets right now. I hate to use the word bubble, but that's why I wanted to bring Christian on, because it's like both feet on the gas as far as the private market AI trade is. Now, that may change, and that may not be the only fact to know, but it's just a different perspective, Adam, that you don't hear all the time.
Yeah. And the majority of the earnings—I would just push back a little bit. They're not startups. To Jan's point, most of the revenue generation—if it was all coming from startups, right, that are thinly capitalized or incredibly debt-laden—I think you'd have more concern. I mean, a lot of this revenue and earnings generation is coming from very large, established, well-heeled businesses. Doesn't mean there's not risk, but I think that provides us with some comfort.
'Doesn't mean there's not risk,' but you're not ringing a bell saying, 'Hey, this all seems like paper here.' Which is great. And I think Jan's point is the key one I'm taking away from this, which is: bubble or not, the juggernaut is still rolling, you know, and perhaps even still gaining speed. So if you're going to take an oppositional position in front of it, like, you better really be right, or you're going to get flattened.
I mean, let me just go back to Nvidia. I consider it to be the heart of the AI hardware ecosystem. I mean, obviously they have software and other ways to protect their moat. It's not the Nvidia of prior cycles, right, where it was just a chip manufacturer for games or for Bitcoin mining. It's just completely—they happened to be in the right spot, and the AI train came along, and they were...
Perfectly placed, and now they are, like I call them, the IBM of that system, and they're paying dividends. You know, they have a lot of cash that they're generating, and they're buying back shares, they're paying out dividends, and doing all the right signals as far as I'm concerned. They extended guidance for a year out. Let me contrast that with the memory chip area, which is an area of prior pain for a lot of tech investors. There we actually put a cautionary email out to our clients in Q2 saying, "I don't know." Right? And also remember, in summer of '24, I did say Nvidia at 50 times sales, at 50 times revenue, was overdone. I won't go back to the chart now, and I said on your interview, like, I'd lighten up for that, and that turned out to be the right call. So it's not like I'm a perma-bull. It's just that I think Nvidia is just a completely different animal. It's like comparing a 21-year-old to a 2-year-old, you know? But they're just different. So those are tough calls, because I'm saying no, the memory chip companies, a lot of them are 21, but they are still like a teenager, you know, and I wouldn't overvalue them. So anyway, we'll see in the future. Okay.
But again, just want to make sure folks are clear on the key takeaway you want them to get from here, which is: it sounds like all signs point to this train continuing to run hard, maybe even gain speed in the foreseeable future.
The big picture trend looks the same. There is a lot of discipline in the public markets. There's less discipline in the private markets, and that will continue to change, but there's nothing so systemic that happened in Q3 that would lead me to say this trend is over. And it's very important to portfolio construction: if you're going to say I'm going to underweight equities because I'm worried about tech, or I'm not going to overweight semis, which we do in our models, that has been very expensive if you get that wrong. So that's why — probably people are sick of talking about AI, but it's so important for financial markets and your portfolios, which is why I think it's important to question assumptions. Okay, so be really cognizant of the opportunity cost of leaving the party.
All right. Hey Christian, this has been really valuable having your private market perspective here, and I will get back with you on, hopefully, scheduling at some point in the not-too-distant future a deep dive into how the viewers of this channel can learn about potentially becoming more involved in private markets as retail investors.
Yeah, it's a pleasure. Thanks for having me on, and sorry, Jan, if I spoke too much, but I'll give it back to you.
Okay, he didn't speak enough.
Yeah. Well, lesson learned, Adam, that when we have guest speakers, Jan should use fewer slides.
No, not at all. It's all great.
Let's, okay, in no particular order. You know, I wanted to revisit a couple of different trades that you and I have talked about. So I'll call it: I was right coming into Q3 to be bullish semis and equities. There was one trade which I'll just highlight here, which is — let me go forward. Oh, I'm sorry. So I thought that private credit managers' equities had kind of were attractive buys in Q2, and I felt like a genius when Blue Owl and others rallied up 30% from like $9 a share to over $12 a share. But again, squeezed by bonds, I would argue, they gave back all of that value. And so now we'll see where we're at, but I certainly didn't anticipate the rise in the 10-year. So, TBD. I'm not selling my personal position there, but when, you know, we get it short-term wrong, we do try to be clear about shorter-term undervalued opportunities and longer-term trends. This was something that has not worked so far. So I wanted to kind of get to that.
Well, I appreciate that. Love the fact that you report on your hits and your misses, although "so far" is the key word here. Real quick before we get off the slide, Jan: just — do you or VanEck have an opinion right now on the health of private credit?
I think what I got wrong also is that Blue Owl got more redemption requests for one of their biggest funds. So I think it's just going to take a while for the wealth community to get right-sized in these positions, number one. And number two, I think the market is saying — which is the theme of this outlook, right — the bond squeeze. All of these value propositions: it's one thing to offer 9%, 10% yields in a zero interest rate environment; it's different if the 10-year bond is offering over 5%, right? The value prop is starting to become diminished. And so that's my point: there's more pressure on the rest of our portfolios, but I don't think it's so bad, given the health of the economy, that I would predict lower lows, if you will.
Okay. And yeah, the spirit of why I asked the question was: six months ago or so, in the headlines, there were a lot of dire headlines, and people were worried about a contagion and maybe a collapse of some of these big private credit lenders. It sounds like you're saying — and don't let me put words in your mouth — but the pressure that part of the market's under right now is largely due to interest rates. It's not due to imploding holdings.
