Talk Your Book: AI Is Not a Bubble

0:00Ben Carlson:Today's Animal Spirits Talk, your book is brought to you by Elger. Go to Elger.com to check out the Elger Concentrated Equity ETF, ticker CNEQ. That's Elger.com to learn more. Welcome to Animal Spirits, a show about markets, life, and investing. Join Michael Batnick and Ben Carlson as they talk about what they're reading, writing, and watching. All opinions expressed by Michael and Ben are solely their own opinion and do not reflect the opinion of Ritholtz Wealth Management. This podcast is for informational purposes only and should not be relied upon for any investment decisions. Clients of Ritholtz Wealth Management may maintain positions in the securities discussed in this podcast.

0:40Michael Batnick:Welcome to Animal Spurts with Michael and Ben. On today's show, we spoke to Dr. Ankur Crawford, Portfolio Manager at Algebra. We had Ankur on The Competent Friends back in April and were blown away by the breadth and depth of her knowledge and her ability to talk about her space. She worked at Intel. Not only is she a portfolio manager, she was an analyst, but she - She was in the field, right? She was in the field. It's so unique and important to talk to somebody that has actual boots on the ground experience. because I think a lot of the discussion that's being framed around the AI trade, the growth trade, the semi-trade, whatever, is framed through the lens of price.

1:20Michael Batnick:And I mean stock price. What does a stock price do? And then you make judgments, narratives based on whatever your opinion is. But to talk to somebody who understands the supply chain, the bottlenecks at such a granular level really opens your eyes as to why the prices are doing what they're doing. It's not just prices respond to the explosive earnings, which is obviously the case, but the backlog for as far as the eye can see. It's unique. It really is. It's, you know, I've never seen anything like this. I mean, stating the very obvious.

1:54Ben Carlson:We've talked in annual spirits a lot about the downside case of AI. What could go wrong? I think her case is, no, no, no, no. People are looking at this using the past as their guide. This is different. And here's what could go right. with AI. And so, I mean, she has a very bullish outlook on everything, right? So I think it's important to hear all sides of this debate. And I think it is worth, we've been saying, have an open mind that maybe this is one of those times that's different. And so she kind of lays the case out for it. And I love the discussion about the market making you feel right or wrong about these ideas and waiting for the market to catch up with you.

2:38Ben Carlson:Anyway, very interesting conversation. So here is our conversation with Dr. Ankur Crawford from the Elger Concentrated Equity ETF.

2:48Michael Batnick:Dr. Ankur Crawford, welcome to the show. Thanks for having me back, I suppose. Back-ish. Okay, so we did have you on the Compendent Friends back in April. And part of the conversation that I found so interesting was you thought you were wrong about the memory trade because the market wasn't confirming your suspicions. And these were more than suspicions. You thought it was so obvious that these were going to work and they weren't. So you thought, hey, I must be wrong. The market, like, right? Like you've been doing this a long time. You're humble enough to know that if you're mentally pounding the table, but the market's disagreeing with you, it's usually the market that's right and not you.

3:32Michael Batnick:Okay. This was the outlier in which you were so right. And April is, it was two months ago. It's two months ago. That's it. Micron was at $454 a share, give or take, when we recorded. It's now at$1 ,080. $450 to$1 ,080. It's up 135 % since that conversation. So now, what signals are we taking from the market? Gosh, what I would say, I think it was in the context of initially, I had thought that I was wrong in September when I had first bought Micron, that I thought that Micron was telling me I was, or that the market was telling me that I was wrong on the thought process. Look, there's been a lot of movement in the market since April.

4:19And I would say back at the end of March, there were, you know, Nebius, I think was at 95 and Astera Labs was at 95. and now Astera is close to 400. And what has happened is that people have, I think the market is coming to the realization that some of what is happening in the AI trade has duration. We have seen, for example, for the NeoCloud's pricing for the NeoCloud per GPU hour is rising because we are short compute. and you've seen Anthropics numbers shoot through the roof. I think they started at 30 billion. Now we're looking at 44 or 49 billion in AR. When growth investors are telling you that they've never seen anything like this, I think everyone needs to take heed because this is a different era and we can stop calling it a bubble.

5:13We can stop calling it one time and everyone just needs to pay attention to what's going on.

5:20Ben Carlson:It does seem like the bubble talk is getting a little exhausting. It's funny, I looked, I wrote a blog post in 2017 and people were talking about bubbles back then. It's been a long time coming. Do you think that people are just so reliant on history that like, listen, every time there's been a big CapEx boom and we've built out a big innovation like this, there's been a bubble. Is it just that people have too small of a sample size? Like, what do you think that people get wrong about that idea? Because anyone who uses history as a guide says, listen, we've seen this checks all the boxes. So what is it about, I guess, this that is completely different?

5:54I think it's exactly that, using history as a guide. Because we have never seen, there's no historical precedent for what is happening today in terms of how fast the innovation is occurring. I mean, was it just last week that we were talking about the Nelm Brown tweet about how models will begin training themselves? and we've heard this before. And, you know, I had, I said this a couple of years ago on CNBC, I think, where I said, you know, when software begins to write software, innovation becomes exponential. And we are at an exponential innovation curve. You know, we're not used to thinking exponentially.

6:38And when innovation, it's also, this is digital innovation, right? It's so, it's very easy to adapt and adopt it. So think about, I mean, if the innovation came in the iPhone, right, then we needed to have dispersed all the hardware and we needed to make sure everyone had an iPhone. And that penetration curve is a little bit different than a digital penetration curve where everyone can adopt it immediately. I think that's what's happening right now where the digital penetration curve, there is no barrier to entry to adopt it and to start using it. The usefulness of the productivity of the AI tools has become incredibly clear.

7:23And therefore, there is no compute because everyone wants to get their hands on intelligence.

7:32Michael Batnick:So Ankur, you're the manager of the Alger Concentrated Equity ETF. The ticker is CNEQ. But you've managed this portfolio in what, a mutual fund or SMA wrapper for quite a while? It's actually an ETF and mutual fund and an SMA wrapper. Okay. And it was accepted about two years ago as an offshoot of some of the other funds. Okay. When you just said that, it reminded me of when people promote their podcast, wherever you listen to, right? Apple, Spotify, YouTube, you can get it in any flavor that you like, but if you want Encore's stocks, you can get them. You're a long-term investor. You are not, right?

8:11Michael Batnick:Like these are concentrated bets that you're making on these companies. How hard is it to ride these winners? Because you're right, we are linear thinkers. I wrote this a while ago, many a moon ago. Anybody could do six plus six plus six plus six plus six. But if you ask them to multiply six times six times six, I mean, your brain will melt. We just, we can't think that way. So as we're transitioning from linear to exponential growth, All right, Micron is up 136%, whatever. It went from 400 to 1 ,000. Can it go to 4 ,000? I'm being silly. But how do you think about the long-term prospects of a stock?

8:50We know that it's discounting something a lot.

8:53Michael Batnick:But people have said, well, this is funny. Micron is the largest stock in the Russell 1000 value index. Like what? Because the market is probably rightfully putting a discounted multiple on it eight times. because we know we've been through these cycles before. But the question is, is this cycle of memory different? What do you say? How do you think about this? Let's talk about this in terms of not just the memory cycle, because I think that there is something much bigger going on than just the memory. And memory is just kind of an example of what's happening. If you look across the supply chain for semiconductors, semis were kind of a forgotten child.

9:35I remember I was a semiconductor analyst for Alger when I first started my career. And it was kind of like we were the ugly stepchildren to software because software had these beautiful business models. Like, you know, semis were cyclical. They didn't get any pricing power. They would like beat each other up every year on getting sockets. And I remember at some point thinking, God, I used to work in a fab, right, in a bunny suit. Like, it's really hard to make these things. Why is it that semis get no respect?

10:03Michael Batnick:Wait, hold on, Ankur, can you just, just don't gloss over that for a second. So we're talking to somebody that prior to her career as an analyst and a portfolio manager, you literally worked inside of these, these buildings that you had to put a hazmat suit on. Yeah, well, it was a bunny suit. It wasn't a hazmat suit, but it was, I did. And I did my PhD work at Intel and, you know, had to build my own wafers there. And so it was just like, I had this like, this appreciation for how hard it was to build this stuff. What happened? You know, semiconductors became kind of a GDP plus grower, plus plus grower to a GDP plus grower to basically almost a GDP grower.

10:49As that entire supply chain became a proxy for GDP, it all consolidated. So across the supply chain, whether it's the PCB guys, whether it's the foundries, the memory guys, the semiconductors, they all, as semi-cap equipment, it all consolidated between, you know, 2010 and 2022-ish. So what happens when you basically get a consolidated market for the entire supply chain and a demand signal that feels exponential? It's really hard to bring on supply. And therefore, you have many different things that end up being fantastic, which is you get better margins, better pricing. all of a sudden the power has shifted back to semiconductors and this entire supply chain for semis including the foundries tsmc global foundries um you know the semiconductors themselves even like the networking equipment like so the entire supply chain chain is stretched right now and this is just what's happening in memory but it's happening happening across every other aspect of semiconductors and this AI build as well, including power.

12:08Now power didn't consolidate, but that power is also short. We're basically short, I would say, soup to nuts, the entire supply chain for artificial intelligence.

12:20Ben Carlson:It is funny how the market sometimes tells you if you're right or wrong, and that's kind of the ultimate scoreboard. I'm curious how many times that you've changed your mind since the chat GPT moment back in 2022 about who the winners and losers are going to be. Because obviously there's been a ton of different narratives. Have you been pretty down the fairway on this and pretty resolute? Or have you changed your mind a lot on how this is going to play out? Look, I think that there's some big picture things that I haven't changed my mind on. There's a paper that we wrote in early 23, March, April of 23.

12:51It was called AI and the declining cost to create. And that talked about how software and any digital assets are at risk and value should accrete to the hardware and semiconductor supply chain. That big picture was something that we held on to. However, the components of that are a little trickier, i.e., when should you have sold software? We sold a lot of our software then, but there were kind of rip-roaring rallies, like CRM won from 160 to 300, right? There was a doubling because people thought it was an AI winner. There was definitively a points in time where we would question the hypothesis.

13:32However, we did stick to our guns on understanding that, you know, there are swaths of the market that you really don't want to be invested in. They could be trades, but not investments. individually in names inside of sectors. I think what we have seen is there's been not necessarily like TSN is something that we've owned for a very long time and have only gotten more convicted in, in part because you've gone from a seven-player market from 15 years ago to effectively a one-and-a-half-player market today. Maybe I would say one-and-three-quarters because Samsung and Intel right now are trying to find their sea legs.

14:17Michael Batnick:The compute shortage that everybody's talking about, is this the type of thing that Taiwan Semi, who actually manufactures these chips, are they being the responsible stewards of the entire ecosystem? Or can they literally also not manufacture as fast as demand wants them to? Let's study that supply chain as well. TSM is not the holder of all equipment, right? So they need to ask ASML to make them equipment. They need to ask LAM Research and AMAT and KLA 10 core for that equipment as well. So the equipment that goes into these buildings to actually. That's right. So, so let's, if you go across the supply chain for even TSM, even if TSM says, you know what, I'm going to add 3X the capacity, which they never would do that if they wanted to do that, the supply chain is going to limit that growth.

15:14Because even the semi-equipment manufacturers cannot manufacture at that rate because their supply chain cannot manufacture at that rate. And here's the beautiful thing, right? Everyone is afraid that this is a bubble. Everyone is afraid that we're growing too fast. And what I'm telling you is that we are in fact rate limited by the supply chain, which is a really important aspect of what is happening. So we're almost capping the growth. So we can't get into bubble territory right now. We can't spend as much as much as we may want to, but that gives you duration.

15:54Michael Batnick:What if somebody would say, but I'm just playing devil's advocate. Everybody knows that supply is constrained as far as the eye can see. And therefore, we're going to bid up the prices of these stocks in anticipation of an endless runway. And of course, the impossible, impossible part of this is, well, when do prices outstrip future expectations of what can realistically grow into? And how do you know that? I know, again, I'm asking you a literally impossible question. So what would you even look to in anticipation of, okay, perhaps prices have outstripped what's reasonably, what these companies can deliver on the fundamentals?

16:35Now, what prices are you talking about? TSM prices?

16:38Michael Batnick:Oh, good question. So I was thinking of my brain stock prices, but if there's like actual business prices, like wherever you want to take that. The most common question that I get asked is, oh my God, the valuations are so high. How can you invest in AI when the valuations are so high? And my response to that is on what metric are the valuations too high? Because the first thing you need to get right when you think about valuation is the E. Only then can you come up with a PE. So if the numbers keep moving up because we are in a supply chain shortage through the end of the decade, then what is the right earnings?

17:24Now, I will give you an example. There is a stock. It's called GEV. I remember looking at GEV, and this was, I don't know, a year ago. There was a certain thesis around GEV. They make turbines for utilities and for CCGT plants, which are required for electricity, which is required for data centers. There's only three companies that make CCGT turbines. GEV is one of them. They have about a third of the market. And their pricing has gone from like$1 ,250 per megawatt to something like$2 ,500 inside of the last year.

18:08Michael Batnick:That's insane. Yes. And it's continuing. Like I was just looking at the numbers the other day, and this is happening across all of these AI stocks. I mean, there are stocks that on 28 trade at a low double digit earnings multiple because our numbers might be quite differentiated from the street. But again, if you stop thinking linearly and thinking about the exponential, then you have to figure out what the right earnings is. Then we can talk about what the right PE is. But I would say this is happening across the supply chain where the numbers are just low.

18:44Ben Carlson:Before you mentioned like, hey, there's going to be some trades here, too. in some investments. When you're creating your portfolio, which is concentrated, I think, what is it, 30 names or something, are these all longer-term investments? Do you look to any trades in this space that you try to get too cute on? No, these are like, I don't know, you told me the number, three, five, seven-year hold. What is your holding period when you're trying to buy these stocks? I mean, ideally, you're looking at things that you can hold for three years. So I always think about, in three years, will this be a triple?

19:11Could this double in three years. That's the mentality. Occasionally, there might be a trade in there, but it usually is on a pretty short leash.

19:22Michael Batnick:One of the things that you just said that I would totally agree with is that, and maybe I'm projecting, maybe I'm making this up, I don't think I am, that a lot of investors are asking the questions, as I did. I started the show with this. Like, is it too far, too fast? Are the valuations too high? Blah, blah, blah. There's still, again, this is like anecdotal, but there's still a lot more disbelief. I do think that investors continue to just be stunned by, and they're talking, they're looking at the stock. You are looking at the business. So you have a, to say that you have a better handle on what's actually happening is comical, but you also are talking to investors and you're hearing disbelief.

20:05Michael Batnick:And you're probably saying like, these people have no idea. They have no idea what's happening. Yeah. Honestly, Michael, I love the skepticism because if everyone was not a skeptic, then, um, you know, everyone would be on the same side as us. Right. Right. And so I love the skepticism and I love the debate because again, it makes me question my own. Like it keeps me, keeps me grounded as well and makes me question my own thesis over and over again to come back to what I'm telling you. There's nothing really that's deterring me from my viewpoint. And look, you see it. Look, the most basic thing, I think at some point today you asked me, what do you look for?

20:44One of the things that we've been looking at is we own this company called Nebius. And Nebius is a, I don't even want to call it a neocloud because it's progressed so much beyond that. I would say it's like the first AI hyperscaler, AI native hyperscaler. that is, that has been formed in this, in this era, their pricing for their old chips is up 30 to 40%. So that's their pricing for like hoppers. The hopper is the first generation of AI, NVIDIA AI chips. They were supposed to be going down 15, 20 % as soon as you got the next generation. As soon as Blackwell's and, and Grace Blackwell's came in, hopper pricing was supposed to go down.

21:33Yet pricing is up 30 % to 40%. What that tells you is that there is simply not enough supply.

21:41Michael Batnick:For now. For now. The thing that people keep coming back to is, well, let's see what ROI the buyers are going to see. And I think what they mean is the hyperscalers. So does that all rest on the shoulders of hyperscalers or are these orders broadening out? I think what you have to do is look at the supply chain of AI. So we started the conversation talking about Anthropik. Anthropik's revenues are, or their ARR is approaching almost$50 billion. Their rumor is that they're actually profitable in Q2. And if you take a look at, again, I'm just talking about the publicly available information. If the rumors that they are profitable in Q2, then this whole idea of, oh my gosh, no one can ever make a profit is out the window because clearly the leader in the space is making a profit.

22:38And I think if you take it just back to fundamentals, the hyperscalers are a bit more confusing, in part because some of the capex that they're spending is for their own internal means. And it's hard to deconvolve how much is being used internally versus how much is being used to lease or to rent out to other people.

22:59Ben Carlson:I'm curious, what would cause you actual concern, though, for the people that keep just shouting from the rooftops, this is a bubble, this is a bubble, this is a bubble? They don't seem to quantify anything. They just kind of say, hey, valuations are high and stock prices are up. What would be actually a cause for concern for you to say, all right, we've overdone it here? Okay, if I could magically wave a wand and we could spend$3 trillion in CapEx and place it in the ground, I think that would cause me to worry. If I saw some sort of an algorithmic change such that we could do a lot more with less, that would cause me to question.