SPEAKER_01 0:00

Eric, we are back with the AI Investor Podcast. And so is market volatility. Momentum stocks have collapsed. And we're going to answer the question whether or not today is the time to panic, or this is just normal AI investing market noise. We're also going to look at what stocks and industries are the most attractive today. And is NVIDIA's next major system delayed? We're going to dig into all those rumors as well as meta on the comeback trail. All that and more is next on the AI Investor Podcast. What are you seeing out there regarding this broad momentum change?

SPEAKER_00 1:15

Yeah, it's good to be back, Austin. We had filmed the last two episodes back to back. So we we filmed and uh we kind of looked at a point in time, and then right away the entire market changed. So, well, it's only been a couple weeks. I I was kind of off the grid last week. I did bring Starlink uh kind of out to Appalachia where we were to get some connection. But coming back and looking at the market, there's this TV show named Community where there's a scene where a guy leaves to pick up a pizza and he comes back and the whole apartment's on fire and everyone's fighting. The market kind of feels like that at the moment. Like you said, if you look at only the overall index values, it's holding in relatively well. But you look beneath the surface, and what we are seeing is kind of a historic rotation out of uh specifically momentum stocks. For the momentum factor, I saw a stat that this July, July 2026, is the worst month or worst July, I could say, on record for that factor. And um the 17-day rate of change in tech momentum is the worst in the past 27 years. So, you know, that gets you back right to the dot-com bubble. And, you know, Austin, what's happening, right? Um, I think a lot of people who are holding various stocks, especially ones that had appreciated a lot, are kind of confused why this is happening at this moment. And I think the bottom line, Austin, is we're just seeing a significant rotation out of a lot of stocks that had seen those really steep increases in April and mid-May. We had talked at the time how um almost unsettling it felt to have those consistent of gains, where we saw uh at one point, I think the semiconductor index increased something like 18 days in a row versus a prior record of nine, because stocks generally go down much faster than they go up. And we are just seeing such consistent gains day after day. So I think what we have now, Austin, I talked about what was happening in various markets, such as South Korea, where we had so much money piling in to leverage ETFs. Well, the news this morning, Austin, is they are now banning new leveraged ETFs from being created. Uh, we we see margin buying near its peaks that had prior, you know, hit in 2000, 2007, 2021. We we all know what came next. And we saw significant crowding from, you know, not just retail investors, but the funds themselves. I saw a stat from a fund manager survey that said semiconductor investing had become the most consensus long in the survey history. So, Austin, the end result of all this is we have volatility and we have a lot of air coming out because we have a lot of leverage. We have a lot of crowded trades. And that's kind of what we're seeing. You know, you look at the data, I saw one too that single stock buying from retail investors in the past few days is at its lowest amount since March 2020, which I I trust we all remember what happened in March 2020. That's that's kind of a remarkable comparison. But the bottom line is what we're seeing in the market, and we'll deconstruct a little bit, is a lot of stocks that were AI winners that were in this momentum trade are down sometimes 20, sometimes as high as 50%. But the indexes remain within 5%. And we're seeing the most concentrated beatings taking place in those areas that had risen, areas like memory, uh, brought across semiconductors. Um, you know, Austin, we talked about robotics kind of having their moment the last episode. Well, that moment only lasted a few days, right? Because it quickly retreated afterwards. So we'll break down the actual news today. Um, you know, we've got things like NVIDIA in the rumor mill. Uh, you know, we've we've got some other areas I think is really interesting to dig into, but from a very high level, that's kind of what we're looking at in the market right now.

SPEAKER_01 5:20

So I wanted to touch on a couple things there. One, yeah, one, let's let's tie together a few things you and I have discussed in prior episodes. Um, levered ETFs, not a vehicle we recommend. They're sort of a structurally bad way to invest in our opinions. Um, that's not to say no specific commentary on how the Korean market is approaching them, but certainly seems more akin to gambling and reckless behavior. So if you want to call back to that, we discussed that in our prior episode. We have also discussed in our prior episodes that we like sell-offs like this in this market because it's creating opportunities. So maybe we'll get into this in a little bit, but the does this feel like it has to you in the past, where when we are seeing these moments of weakness and we we come back to the deep seek moment, but really there's been three or four of them over the last two years in this sector where you're seeing opportunities created from the volatility, or is this a more structural rotation which won't create as many opportunities because it's people sort of moving from one long position to another long position? So that's my question. And then you had said something in there that puts me on pause a little bit, which is semiconductors being the most consensus long trade in a while. Why would that put me, uh why would that give me pause? I forget what episode we had discussed this in, but one of the most difficult things in investing is once something becomes consensus, the opportunity is no longer there, right? So you and I have talked about this a lot, that you that doesn't mean you always need to be contrarian. You shouldn't be contrarian just for the sake of it. But if you're to the point where everybody knows the trade is an opportunity, there's no additional opportunity left and it gets priced accordingly. So that gives me a little bit of pause. But I'm curious if you could answer that question about market opportunity in this sector and if this relates to other prior AI drawdowns or if it feels a little different.

SPEAKER_00 7:02

Yeah, let's let's let's attack a few of those individually. It's interesting, Austin. I remember when we had first started the podcast, I think it was summer 2025. We had just come through another kind of wave of volatility, like what's happening today. I had brought up the point AI itself is something that is tremendously valuable to semiconductors because as we've talked about, it's kind of moved this value from software to um kind of this compute layer. And I had made the observation that at that moment, despite having probably the single biggest bowl catalyst to semiconductors ever, which was the launch of uh, you know, ChatGPT in in November 2022, they had actually underperformed the market from the release of that event. And at the time I said this is why investing so hard, because you probably had the biggest event ever for this one industry. It had underperformed the market. And um, you know, we we quickly after after that observation, we saw a massive reversion. Um, I'll I'll get a little bit more into semiconductors later, but you had asked kind of the broader setup for whether or not we are seeing kind of is this a rotation that follows news flow, or is this kind of volatility coming out of the market? So I would say, you know, as is often the case in moments like today, while stocks have sold off, um, especially momentum stocks the past couple of weeks. The news flow has been highly encouraging. You know, Austin, the biggest trend we've been looking at across the years, the growth of agentic uh uh AI. We saw the release of GPT 5.6 in the past couple years. It it keeps open AI at the frontier with Fable. And also it's a massive acceleration for open AI's business. Sam Altman put out a tweet a couple days ago saying that they are seeing a 2.5x increase in their agentic products across the last week. And third-party data validates that just to show how big the takeoff in agentic AI is. Codex, which is that product that Sam Altman was referring to, it had 1 million users in February. Austin, it's starting to jump by 1 million every single day. It hit 6 million on July 12th, 7 million the 13th, 8 million the 14th, 9 million yesterday. It's it's adding 1 million users a day. And and the other thing too, Austin, with all these models, and we'll talk about models from Meta and some other companies that have released, you are seeing them fulfill this promise of being much more broadly intelligent. One of the indexes I look at is something called the Remote Labor Index, which basically looks at how well AI models compare to a suite of tasks that you could send to a remote worker. And until recently with Opus 4.6, which is only a model that's, I don't know, what, five or six months old, it was 4%. Fable's now at 16%. And again, Austin, we'll we'll talk about a little bit more specific with Meta later, but they're getting back into the race. We've seen new releases from Grok, and we're seeing lab revenue. Um, who, you know, companies like Anthropic and OpenAI, their revenue continues to escalate at really eye-watering levels as they look to their IPOs. So, you know, Austin, the the important thing is these companies need this revenue to spend on compute, and they are getting this revenue at really high places. Second, I'd say the supply chain signs the past couple weeks are very positive. ASML, which we've talked a lot about, they produce the EU V machines that are at the heart of uh AI. They came out with their recent earnings and said that they're gonna raise capacity in 2027 30% and likely 30% again in 2028. The consensus is 25% and 14% before they reported. This morning, Taiwan Semiconductor came out and released their earnings. They said growth is gonna be 40% this year. That's above consensus of 37. And they had prior guided to uh, I think 56 billion on the high end of CapEx. They took that up to 64 billion. So, Austin, these are the building blocks that all the other future growth for the industry flows through, and it's extremely positive. Um, you know, two other things I'll just point out more quickly: hyperscalers, they keep moving up expectations. We've seen more chatter about numbers that could be coming out from Meta and Google on spend next year, which are um significantly above any expectations mere months ago. And finally, we've talked about macro worries with interest rates rising as one of the key things that could kind of derail a lot of the AI trade in the second half of 2027. But, you know, in the past couple weeks, we've seen positive news on inflation and CPI readings. And Kevin Walsh, who's the new Fed chair, you know, he he's been touting in in hearings with Congress the strength of the US economy, specifically citing artificial intelligence as the driver. So um, you know, we can we can pull this back. I want to give you a chance to react to this. I'll kind of talk about some stocks. But when I look at the high-level news from the past couple weeks, I see very encouraging signs.

SPEAKER_01 12:26

It sounds sounds similar to me as well. And particularly the agentic developments with codecs and OpenAI is sort of refocusing back on that, you know, having a more capable, autonomous model. If there was anything that could put wind back in the sails of the memory trade, that would be it, right? So and the scale of what you're talking about. So that's actually an interesting place to look at. And it's it's crazy to say, oh, there's an opportunity in the memory trade after they're all up, you know, probably still four or five hundred percent. But I think Intel, what back down now near 100, I saw Micron maybe in the 800s again. So um, if you missed that boat the first time, if there was one thing that could, you know, recap reinvigorate that, it's it's OpenAI entering the agentic race as well, where um and delivering on the thing where we have talked about where agentic usage, a thousand X is your need for memory because you have to hold the model in memory as you operate. Um I wanted to go back a little bit. Uh power management, uh this probably ties to the NVIDIA rumor that you have talked about, but could you say anything there? Because within the space, you know, I think optics took a step back, the semiconductors took a step back. Power management took seemingly the biggest of the, you know, the most brunt. So so why is that? There was some stuff down 50%, 40%. So is this because of delays with NVIDIA's new chips, or is this because of concerns around the 800 V transition? What are you seeing there? Or was it purely just that they were they were already speculative long shots and any small tremor in the AI industry, they were going to come down more?

SPEAKER_00 13:57

Yeah, Austin, it's a good question. I I think you have to there are continuing rumor mill battles around a lot of these stocks, but I think you do have to start with the momentum aspect that these were some of the stocks that saw the steepest appreciation in that March and April period. And and now they're kind of taking that breather. But, you know, the bottom line, and I know you alluded to this earlier, and any new listeners are gonna think we're weird, but I'm the most excited I've been to invest um since probably late March uh and April. You know, I came back uh from my trip and I kind of looked across the landscape, and there's a lot of things I think are are very interesting right now. And and you had alluded to it earlier that semiconductors being that consensus, it's it's something that is a bit of a red flag. And when we start this podcast, it's funny how how much research is out there, how how many substacks have spun up, and you know, it was all in late 2025 and early 2026. The fact we started in uh kind of late summer 2024, there just wasn't a lot out there, right? We were extremely non-consensus, you know. I I would say the consensus at that point was extremely bearish that, you know, this was a bubble that could never reach the points that has today before it would um kind of retract. So we we kind of, as you said, you don't want to be contrarian for the sake of it, but I'm much more comfortable being contrarian because it often means that you're gonna see something that's that's not priced into a stock. So when when when people kind of move out and uh it you're not in kind of that red hot center of the investing universe, often I'm more comfortable with that. So Austin, I would say, you know, right now, let's just talk about, you know, you mentioned 800V. I can I can touch on some of our themes, but what I'm most excited about right now, I've said in recent episodes, I think investors who who don't want to go down and and look at some of these, you know, uh stocks in very complicated industries, you don't need to do that to get um what will likely be outstanding returns from today. Because, you know, we made our large investment in Broadcom recently. We laid out our thesis for that. I believe this is a stock that looks very cheap relative to its growth potential and and also has material upside head into 2028. I think NVIDIA from these levels looks cheap. I talked about in the last podcast how Microsoft um looked looked pretty cheap and was a very unloved stock. And we've we've seen some rotation back into that. Uh, you know, Austin, specifically as we've seen the sell-off in momentum, hyperscalers. We'll talk about meta later in this episode, but meta is up 25% from June 25th. So it's an extremely uh fast move. You talked about memory, Austin. Um you know, when when you do the math on memory, it it I think people are really underestimating just how much. Yes, memory stocks are up a year a lot this year, but the demands for memory from the rise of agenc computing is also such a catalyst that didn't exist a year ago. So, you know, that needs to be baked into these stocks as well. And when I see the sell-off in memory, it it often includes a lack of discernment. I've talked about ACM research in recent episodes, told people keep that on your watch list because it might sell off if we see a memory sell-off. It's selling off now. If there's one swing that's going to bring more supply into the memory chain, it's going to be China. And that's a stock with direct exposure to it. So when I see it selling off with everything else, that's a stock I'm gonna put at the front of my watch list for things I might be adding in some future episodes. We we released in our latest episode that watch list on robotics. Uh, why did we not buy all those stocks? I said generally they were too expensive. Now we're getting 30% better price, and we might get even better prices in the weeks ahead. And we we um made uh that uh recommendation last week, which was Regal Rexnord, which is ticker RRX. You look at that pricing as even more attractive, and that was a stock I was liking because it's an industrial at a valuation I thought could be compelling with some catalyst into the future. So you only get better pricing on that. Industrials, Austin, you look at things like Modine Manufacturing, Quanta Services, these are companies with known structural trends. We really like the future with what's gonna happen with the need for grid upgrades and companies are at the center of data center build-outs. The problem was in past years, you know, you will have been paying 20x trailing earnings for them. Now you were suddenly paying 20x 2028 or 2029. Well, if they're stepping back 30%, 20, 30, 40, all of a sudden you're getting these names that you can really feel comfortable holding for long-term structural trends and much better prices. You look at energy, companies like Visdra's trading for 30% were traded last summer. And as you alluded to, memory, yes, it has run up a lot. But if we see a 40 to 50 percent drawdown, am I going to get very interested in the space? Yes. And you know, we'll we'll have an opportunity in future episodes to lay out our math and logic on that. But I do believe that this is going to be a buying period, especially if someone missed that, um, that you're going to have a chance to get out of place. You will have some what I believe high chance of good return. So, Austin, the the key idea here is often when you have moments like today, yes, there were companies that needed to take a breather, but the more and more this panic builds, the more it turns into what I would call indiscriminate selling. And that's a better market for kind of the stock picking market versus, you know, we all love seeing our portfolios go up in April and May, right? But, you know, we had warned at the time it was leading to some very inflated expectations that was largely agnostic of overall quality and valuation. And we're we're just kind of seeing a moderating on that. So, bottom line, there's a lot of industries I'm very excited about. I think we'll do some buying in recent weeks. And um, I think there's a chance to get some stocks that are really well advantaged, um, not just in this bottleneck trade that got so intense in April and May, but stocks that are going to be structurally advantaged. And it's gonna be a chance for investors to add them to their portfolio at a much better price than they've been able to across 2026.

SPEAKER_01 20:47

And one of the key things to watch uh is the hyperscalers here, right? A big part of this trend and the opportunities created has been the transfer of cash flow from the hyperscalers to the AI build-out, which you know you were very early to. Uh, and we continue to see that the hyperscalers are signaling additional capex, right? And that's been the question all along, right? Yeah, how much more can they really spend, right? Oh, they're all they're through their cash flow, now they have to go to financing, but they're still spending more and they're continuing to signal that they're spending more. So at every moment where people are breathlessly saying this can't continue forever, and right in in some way that is true, but it has exceeded every expectation, and they've and the hyperscalers have continued to raise their capex estimates. So, can we talk a little bit about Meta and Google and some of the moves that they've been signaling for 2027? So I'd love to hear about that, as well as the one big black swan that we come back to, which is interest rates, particularly as these hyperscalers move to debt financing. How would interest rates impact this industry? So, two questions there. What are you seeing with Meta and Google's big signals? And then what do we need to watch with interest rates, which are now looking like they might be higher for longer?

SPEAKER_00 21:59

Yeah, you know. No, it's it's interesting. I I think the news has been uh interest rates are a major concern, but within the past two weeks, I think the news has been generally more positive than you will have seen in a preceding, you know, two months before it. So I think that situation is looking a little better. You know, we need some things to happen like commodity prices to come back down. We've seen that, which is is that's gonna kind of help this overall CPI picture. Um, second, as far as the kind of hyperscaler spend, as you noted, the big question 2027 is going to be eye watering growth. You know, I I saw a Morgan Stanley estimate that we're gonna go from um something like 770 billion in spend this year up to 1.2 trillion next year. The big question is 2028. So what what is what is the level that you can get to before there's there's just really no capital available? Um, you know, I I think it sits probably somewhere at the current moment that 1.5 to 2 trillion level. But Austin, the other side of this is so much of this capital is going to investments that are going to have a payoff period measured in years. So what we need to start monitoring as well is what kind of revenue is coming from generative AI. And as that's happening, um, that's going to lead to kind of an increasing amount of this hyperscaler spend that can happen. One one interesting kind of point of validation this week was Warren Buffett on CNBC saying uh they had recently invested in Google Secondary and that secondary investment is going to be used to finance AI chips. And Buffett said he actually directed the investment, which, you know, if if um I saw uh someone putting kind of a you know just a tweet out saying Buffett had done in 2008 by banks, by American banks I am, that's and by data centers I am. So I I I think overall, in terms of the picture of how data centers can be financed, it it remains kind of the number one overall area I think about and and uh determining factor of how long this trade can go for. But like I noted, I haven't seen something materially worse at this in recent weeks. In fact, I think the news flow has been relatively good for the durability of this trend.

SPEAKER_01 24:29

Yeah, and you know, uh capital always finds a way. You know one of the things that hasn't really entered the arena, you know, let's say these companies do, which is shocking to say given how profitable they are, but let's say these companies at some point do start to hit the, you know, their ceiling of tolerance on debt. There are other ways to raise capital. There's secondaries, like you're talking about. There's also plenty of um, you know, money in the Gulf, which would probably be eager to invest in some of these um you know models and these build-outs as they're proving out. So um there is capital to be had. It's just a question of where it comes from and under what terms. But you had said something, right, which of course gets me and every other listener excited, which is it's the most excited you've been to invest since late March or April. A lot of high-quality stops stocks are looking cheap, even cheaper than they were when you recently recommended them, including Broadcom and NVIDIA. We've talked at length about how NVIDIA is cheaper than the market, despite having eye-watering levels of opportunity and growth in front of them. So what are you know, just talk to me about some of the sectors where you're seeing opportunity? What has you more excited? You talked about memory, we talked a little about robotics. What specifically has you excited about the AI investment opportunity today?

SPEAKER_00 25:41

Yeah, and you know, I I went through some of the sectors earlier, and you know, I I think areas like industrials, it's it's it's not a matter of whether or not they're going to have structural tailwinds for a long period, especially you look at what John Rotanti brought up with the quanta services. Um, this is this is going to be an electrification of the grid trend that's lasting more than a decade. The big concern is valuation, right? So as long as we're getting better valuations, you know, I'm going to have more confidence in these companies. Robotics, I was most concerned about the valuation of a lot of these companies because, you know, the Austin, I've brought this up a lot as a central idea in recent podcasts. But if our area of uncertainty goes to 2028, that's where, you know, we might be spending 1.4 trillion and we're kind of at that line of what could probably be financed from the companies in this build out. This moment, um, there's just a little bit more uncertainty. If you're having to pay really high multiples for that moment, it just introduces more risk, right? And that's where we've seen these momentum stocks sell off. The more that you have lower valuations, the less risk that you're going to have buying a lot of these companies that are really high quality in trends. So I think, you know, again, in in future weeks, we'll we'll have some potentially like momentum stocks that maybe have gone sold off more. And we we feel like, hey, they're down 50%. It it's hard to buy stock like that in this moment. And we'll add them, but we'll probably, you know, make some investments as well of some of these more boring companies that we just like the valuations a lot more and continue giving people more diversified portfolios across different areas. So you don't have to feel like you're just in something like optics that's run up a lot but can get rug pulled. You know, I I mentioned in a podcast, you know, I I don't know which one it was, but it was four or five podcasts ago that I was personally when we reviewed my portfolio, I was kind of taking back my optics positions because I was I was afraid of how crowded the trade was, um, how how concentrated I had personally gone in. And and I did do that. Um, so you know, coming out the other side though, we might have opportunities again now that we've seen a lot of these stocks retrenched. So I I'm I'm pretty excited across a lot of areas, Austin. I think I think if you're looking at more conservative stocks, there's opportunity. If you're looking at some of these momentum stocks, there's probably some that have fallen and needed to fall, but there's probably some that do have outstanding opportunities that the market has indiscriminately sold.

SPEAKER_01 28:17

Let's let's talk about one of those, the opportunity hiding in plain sight. This is NVIDIA. We're seeing some concerning news uh out of your favorite research tool, semi-analysis. Uh input Eric Bleaker's promo code for 10% off at registration. Um, that there's going to be some delays in NVIDIA's system, right? So this is right, we're looking forward to like Ruben, which we've talked about, and then beyond that, Feynman. So what is the nature of the news that we're seeing out of semi-analysis and just broadly in this space? And is it something that you put weight behind? If it is true, how much does it change the thesis here? And does it just become a timing issue as opposed to a trend issue?

SPEAKER_00 28:55

Yeah, this was one of um, there's so much confusion right now. It it's funny because we went through this with Blackwell back in you know, late 2024, where you you've almost got this fog of war of a of a new kind of change coming from NVIDIA, where kind of the architecture of this next generation of chips is coming. And there's just a lot of uncertainty throughout the supply chain. And NVIDIA is so big, they have so many sales, it it impacts dozens or hundreds of companies. So awesome. The news is that some of the analysis, they they said that NVIDIA was essentially going to be delaying some of their next generation systems. This led to a lot of selling pressure, but it's also something that probably wasn't entirely represented correctly by most of the headlines I saw. So just taking a step back, because I know we we have all kinds of nurdery that we get to on this show. But one of the hardest might be we we bring up all these systems from NVIDIA, and they often have kind of complex names. You you know, might have Vera, Ruben, Ultra, NVL 144. People might wonder what that means. Essentially, NVIDIA, if you talk to a person on the street, they would probably think this is a company that sells chips. The truth is they're a system seller, and these systems all have names that start with the CPU first, which is going to be Vera for the next generation. And we've talked about on the show before their their CPU businesses now just the standalone CPUs are 20 million, so this is extremely large. Then they'll have Ruben, which is the generation of GPU. Um, but the last thing that we have, Austin, I I mentioned the name is something like NVL 144. And that that's how many GPUs are in each system. Because the big idea that we've talked about is when you're doing a scale up, scale out, scale across, scale up is how many GPUs you can get to work as a single system. Because if you can get more GPUs as a single system, you don't need to do all this communication, all this networking that leads to some of your GPUs being idle and you having lower utilization. I mean, it it sounds it sounds low low table stakes, but the reality is we're talking, you know, this is hundreds of billions of dollars um worth of value if you can keep all of your GPUs working. And that's why so many of our networking stocks have worked so well. But behind this all, Austin, they're gonna go to this next architecture and they're moving from something called Obern to Kyber. And this is basically to get more of these GPUs in a single rack, in a single system. And basically, what would happen? We've talked about how copper makes more sense within these single systems, but they were going to need about 20,000 cables. So you can imagine 20,000 cables and the challenges from that. They decided to do a different strategy, which was they're gonna make this giant circuit board behind it. And it's got all these layers and it's got all these copper traces. The problem is just it's really hard to make. It's it's just really challenging. And what semi-analysis reported is basically that this big circuit board, this this mid-plane, they're they're having problems making it, which is gonna push these kyber systems back 12 months from 2027 to 2028. So I I know I had to walk through a little bit there because we do need to detangle why many of the headlines look scarier than the reality. You know, on the surface, if Nvidia is delaying their systems by 12 months, that looks very bad. There, it looks like they're losing their lead. It looks like their sales growth's not going to be as strong as it would be. But, you know, Austin, the reality of the situation is that these systems were supposed to ship in the second half of 2027. So a delay into 2028, it's it's presented as 12 months entire year. It could be a matter of months, right? This could be November 2027 to January 2028. Second, by semi-analysis' own admission, they can ship Vera Rubin just off the old uh architecture, which is Obron. So it's it's not that NVIDIA would necessarily even be losing sales, they would just be selling a different type of these racks to their customers. And you know, the third part, Austin, is and this is what we saw with Blackwell, there's so many rumors heading up to it, so many things about this isn't working, that's not working. Well, whenever you're doing a really ambitious kind of product, which this is an ambitious product, well, they're a year away. Yeah, if they're having a hard time making these mid planes, they're gonna have a year to work on it. They're gonna have a year to get the yields up. So I think they're in a relatively, you know, good position here. So, Austin, the bottom line, we need these new generations to happen. Right now, we're talking about the models that are happening, like Fable, like GPT, you know, 5.6, um, like the new models from Meta. These are the models that are increasing in capability and making agentic AI a reality. And they're happening because they are all exclusively trained on Blackwell. The Ruben generation will allow AI to continue improving to its next generation. So it's important to the entire AI trade. It's important to NVIDIA, it's important to NVIDIA suppliers. And the bottom line is well, the headlines were scary and caused, I think, a lot of selling. They didn't represent the situation. So I I I had to catch up on this after getting back from my trip, and and and it looked a little scary and it's got some details, but it's honestly not anything I'm I'm concerned with.

SPEAKER_01 34:57

Uh let's turn our attention to one of the biggest positions in the portfolio, and that's Meta. So, this is a position that I have also added to recently, and you and I have discussed Meta's opportunity in the AI industry being multifaceted. So it's it's not one-dimensional here, but primarily Meta, in the worst case scenario, can build very capable AI models simply to optimize their own platform and get more ad yield and you know return on row ads and ad spend for their marketers. So that is the worst case scenario, but of course, Meta's ambitions don't stop there. Uh Meta's been leading the open source model charge. Fair to say, I don't think that's gone very well so far. It just hasn't gotten the traction domestically at least. China's definitely out in front there, but Meta's continuing to plug away at it. But they also recently transitioned to a cloud business. We talked about that opportunity, maybe making turning something that has a cost center into a profit driver, right? Being able to rent your excess compute. So it seems like Meta's in this little bit of a renaissance comeback kid moment. Um, I'm rooting for them. I own them. I think they have a lot of opportunity out there. However, I will also state of the large hyperscaler companies, Meta has historically, in my opinion, had the worst track record of doing things outside of their core business. So if we think about their launches into, you know, Luna coin or hardware or the metaverse, they're always willing to try new things. Fortunately, to Zuckerberg's credit, he seems willing to fold those investments and opportunities and capture that cash flow when they don't work, which is one of the things I like about it. Um, and this looks like it's another one of those. So is this the one that will work? Will Meta be able to make this, you know, come back in this transition as they launch their cloud business and they have their new models that they're releasing? And there's definitely some echoes here of Alphabet and Google, right? Uh Alphabet and Google left behind early on in the AI race, dumped a ton of money in, caught up with Gemini. Um, and then they had they were able to make a new business line out of their chips business. One of the other things about Google's models they've and chips, they've been very, very energy efficient, which is also something we're seeing with um Muse Spark 1.1, which is Meta's model. So sort of a setting of the table there with the current sort of opportunity and concerns around Meta, but what are you seeing as they launch this new cloud business and their new Muse Spark models? Is this them breaking through and they finally got a meaningful new business line here? Or is this the next Luna coin um hardware misadventure?

SPEAKER_00 37:22

Yeah, I like that Google comparison. That's you know, because you think when we first started the podcast, I was very negative on Google. We changed our tune on them. Um, and then you saw how quickly narrative flipped. And um, we we might be seeing something relatively similar with Meta. A key part of that narrative flip was um Google becoming competitive with its frontier models like Gemini.

SPEAKER_01 37:46

So yeah, they and energy efficient, which is one of the been the big bottlenecks, right? Energy production and and and if you can have more efficient chips in the form of TPU or more efficient models, right? That's worth something to the bottom line in plenty of use cases, which looks like Muse Spark is quite affordable compared to Grok and GPT. So that's the thing that's got me a little interested here. Um, but but please keep going.

SPEAKER_00 38:08

Oh no, I I think the big picture, I so just the news. We we talked about them announcing um kind of their cloud offering in our last episode or the episode before it. Um, you know, they're gonna do bare metal sales where they allow people to rent their infrastructure. They're also gonna give access to Muse Spark. Um, we we have now seen uh Muse Spark. Uh I believe it's it's in wide release, but it's definitely something that people are able to go out and test. And Austin, it it's it's not probably quite as good as GPT 5.6 or Fable, but but it's close. And as you mentioned, it's affordable. Um, so you know there's always this performance versus affordability gap, and and and they are going to have their own lane, which is they're going to push for being more affordable. I think the big picture, two things. Number one, for Meta, it's it's we've we've we've been saying over and over, I don't know if I want to hold as large of a position as we have in them. We had initially purchased $100,000 um back around the uh deep seek moment. I I just didn't love where it was trading. I felt like it wasn't fully reflecting its value. We are now getting it back to around the price where we bought it. So I am heartened to see that. On the second half, how this works for the AI trade. I think for the AI trade to be its most successful, um, you want competition across all of these models. And it it looked for a bit like Anthropic might be running away because they had that early lead into coding and and they really pushed that advantage. But I talked about the usage numbers for open AI's products. Um, Grok is uh just released 4.5. We have News Spark, which released 1.1. And there is this kind of narrative emerging at this moment as well, um, about how companies um protect their data and their IP and how they're going to use all these models. We've seen Palantir CEO, Alex Carr pushing on Microsoft CEO, Satchin and Adela just put some out on it. Um and it is really interesting. What is going to be the orchestration layer amongst all of these um models as suddenly, you know, agent usage escalates? So there's a moment here, and this moment's happening as we're getting a lot more model diversity, which is which is a really interesting development. I've been researching some stocks, Austin, on this, um, you know, like one like ServiceNow, it's one that we've talked about in the past, it is extremely beaten down, but it could have a significant play here. And, you know, sometimes when these moments are happening, I do love putting out to the community that um if if anyone is looking at plays in this area, um, especially getting some more software stocks in the portfolio, I would love to hear um them because I do think that this, you know, so much of the revenue was flowing to anthropic and open AI. And now we've got more models happening. And we're we're going to need kind of some control layers that feel like they are outside just those model companies. And that might be a really major opportunity ahead into the back half of the year as kind of people come to terms with where we're at. And um, you know, if that allowed us to get into some, you know, stocks that that might be a lot different than this kind of hardware and infrastructure, I would love to continue adding them to the portfolio. So that there's there's the side that's been very good to Meta, and there's the side that we've seen something really good developing to the overall AI trade.

SPEAKER_01 41:50

ServiceNow is an interesting one. They were they were on my radar for a little bit in the SaaS meltdown, and I actually never pulled the trigger on them. I may need to do another little investigation there. The thing that that company has always done remarkably well, right, is their net dollar retention rate, which famously I think their retention's like 95 to 99%, and then their net dollar retention exceeds 100%, which means they're getting more out of their existing customers, which is possibly the greatest signal you could get from a software company that if your customers are getting value, they continue to not just stay, but spend more with you. So that's an interesting one if you're looking for a SaaS falling knife play. Um, but speaking of companies that are doing well, we talked a lot about how a lot of the positions in the portfolio have had a very rough last two weeks. One company that's shooting the lights out is Air Test Systems. I think they were up, what, 25% on earnings and a forecast. Uh, was it yesterday I saw or the day before? Um, so this is one that you've had in the portfolio for a couple weeks now. And yeah, talk to us about what the market saw that they loved, which is particularly interesting at a time when they're selling off many of the other companies in the portfolio and that with exposure to semis.

SPEAKER_00 42:57

Yeah, we had we had recommended air test systems March 13th. Um, it was trading around $35 somewhere in there. It got all the way up to $120 per share. Um, it had fallen, it looked like uh down below 70 into the 60s recently. It kind of, you know, again, this is this is a stock that's a momentum play. Uh, they had revenue, I think, last year of 50 million, and they were trading for more than $4 billion. So you expect some error to come out of it. They released earnings and they were very good. So, just quickly on air test systems, it's a burn-in test company. They were previously levered to EVs. What we liked about them was their products are going to quickly become very applicable across a broad set of uh kind of AI industries and trends. Um, as power density increases, they're they're gonna be applicable to uh AI processors, um, they're gonna be applicable to optics, they're gonna be applicable to memory. Also with something like Like advanced packaging, we've talked about how you know the value of chips and a lot of the progress is going to come from instead of just being you know going down this kind of five nanometers to three nanometers, it's gonna be how they're packaged up. If you have failure in chips, when you're putting them into these kind of advanced package solutions, the cost is going to be higher. So you want to invest more into testing to make sure you won't have failures before you package them. So Austin, just a lot of trends coming together for one company. And we loved the tailwinds behind it, but we knew there's a risk. It's it's got revenues of 50 million, it's it's got a market cap that's soaring, and and they're going to need to execute. They're going to need to show some inflection. So the report this quarter, it's still a tiny company, but their backlog went up to 100 million. And the revenue forecast for this year, it's 150 million. So 200% growth on the high end. They'll probably beat that high end, would be my guess. And the drivers are kind of the thesis that we talked about back when we recommend them in March, taking off new photonics customers, momentum and AI processors. So, Austin, a really encouraging report. We talked earlier about how we're in earnings season now, um, which is which is exciting. I I get a little like kind of like weird. You know, I'm like a junkie needing my hit or something while I'm outside of earnings season. It feels good to be back. And, you know, Taiwan Semiconductor, really encouraging numbers. ASML, really encouraging numbers. Here's a smaller company, but very encouraging. So you you mentioned their returns. They opened up yesterday up 50%. They closed the day up 20%. They're down again today. So what we're seeing is really good earnings, really good news. But you know, it is, we are in a huge sell-off on this momentum factor. So, you know, the bottom line, Austin, I would love, I I know people are gonna think this is weird for me to say this. I would love for them to sell off this entire earnings pop um because this is a company that, you know, I I don't want to just be recommending the riskiest companies, especially at a moment that feels as uncertain as today. But this is a company with just such stellar tailwinds behind it. If we can get it for cheaper, it might be one that I can, you know, re-recommend and and make a larger position after we see kind of the promise of this stock begin transitioning into execution, which is the story of what we saw this last quarter. So so great quarter for air test systems. It's it's one of the kind of higher momentum stocks I would definitely consider adding if we if we keep seeing a market sell off.

SPEAKER_01 46:49

Eric, let's uh let's continue your earning season junkie analogy here, because you know there are the highs and then there are also the lows, right? There's the come down. So the flip side of earning season is sometimes uh companies can get whacked pretty badly. And one of those companies is IBM, which was down more than 20, I think it was 24% after earnings, more or less. Um IBM total renaissance in the AI era. That's one point the company was up 170, 200% or so um on this uh AI tailwind. And boy, boy, are they coming down hard, right? So, so what's going on here? I mean, did do we need some Narcan for IBM? Like what do we do from here? And and and and why were they down so much?

SPEAKER_00 47:31

Yeah, I'm looking, they're down, it looks like down 25% across the past year. It was company, you know, we've talked about them a little bit on the podcast, just saying one of the real unexpected winners of kind of the AI age was um from 2023 to the end of uh 2026, they had grown, it looks like over 100%. But also, I I don't have a lot to say about this. It was just a little interesting to me. Um, they had essentially said we missed our results because our customers are shifting spending to things like memory and AI compute. So on the surface, it's it's kind of bullish to a lot of the stocks we have in the portfolio. Um, but it it did cause a little sell-off in software of people saying, okay, if if all this spending shifting there, this is another example of where software might be under pressure. I do think one thing that's interesting, Austin, is you know, one of the facets no one probably will have expected if you went back to the 1990s, is how durable all these technology companies are, right? IBM, they've they've gone through so many transitions and through it all, they've had their mainframe business, you know, and of course we've seen reinventions in Apple and and uh Microsoft and and so many of the companies have have had to reinvent themselves. But the question is, and I think Ben Thompson brought this up, who who does the uh newsletter Stratekri. Are we seeing something where IBM mainframes, which have been a durable product for for so long, is it finally that AI makes things like porting code out of them easy enough that they're finally starting to see these headwinds? So I I think that's a really interesting question, that it's it's a company that's survived so many, so many shifts in technology and and might finally see a franchise that's existed through them all breaking. So I I think that's something to watch. And and I think um it it has really sweeping implications across a lot of companies that have had defensive technology that's lasted for a long time and and and how quickly um it could potentially go away in the coming years.

SPEAKER_01 49:42

Well, one of the most interesting old guard companies to me that we haven't talked about in this podcast, but their stock chart is phenomenal is Texas Instruments. You know, they're a quarter of a billion dollar company now. And you know, we all of course know them from our TI-83 calculator days, but they do a lot more than that. And you know, what what an incredible reinvention that company has has gone through many, many, many times, right? They they boomed in the tech bubble, they bottomed out, and they've had to reinvent themselves a phenomenal number of times. One company that's also done an amazing job reinventing themselves and being relevant is Oracle. We've talked about them on this podcast. Interestingly, they're a tale on the other side, though, where I think at one point their shares were up 40% in a day because they were starting to forecast 2030 revenue uh and earnings back in 2025, I think, or 2024. Um, and now they're actually trading back to their June 2024 pricing. So um there's opportunity in these old companies, and Oracle's reinvented itself many, many times. I would never bet against Larry Ellison and that company. Um right now, looks like they're they've got the wind in their face, so to speak. Texas Instruments, opposite story. Totally riding this AI boom, phenomenal company up like 13,000% or something, um, close to all-time highs. So um IBM, sorry, sorry for your loss, but um, you know, we wishing you the best out there. Uh Eric, where do you want to go from here? Do we do we call this a wrap? Do you want to put people on uh throw the football to the next episode and tell people what we'll be talking about, what they can look forward to?

SPEAKER_00 51:11

Yeah, I think I think we'll wrap it up. We're at 50 minutes, which is about how long we we normally go on episodes. So uh, you know, a couple things before we go. Number one, I talked about in our prior episode, but um Investicon in August. Uh, we do have a promo code for that, which we'll put underneath in the comment section across uh YouTube and other uh players. If if you have any interest, if especially if you live in Europe, it's taking place in Dublin, and you want a discount code to go to that. Um, we we've got it below. I would love to see some uh podcast members there if you had a chance. Second, for coming episodes, you know, Austin, I think it's really gonna be, we're gonna see how long this volatility lasts. I'm I'm looking at the market today and uh it definitely opened choppy. Um sorry, I'm trying to see if I can load it. Yeah, the NASDAQ's down 0.8% momentum stuff's gonna be volatile. So as this continues, right? Um we always joke we'll we'll have fewer listeners because people are gonna tune out, but um, we are going to be much more excited for the market. And I I just as I noted earlier, I I haven't personally been this excited in months. So I think it's really going to be a lot of names that we've maybe missed in the past or that we deeply respect. We're gonna have an opportunity to add them at uh cheaper prices. So if you're looking right now, if if you are saying, yes, I understand that there is volatility, it feels uneasy, but this is this is a period that I would I would like to add some best names for this moment. I think we're going to be able to provide that in the coming weeks. And and I I haven't been more excited to do research on things. So I think I think we'll have plenty of research on industries, on companies, and it as the market gets worse, it's it's an even more exciting time for the podcast. So I hope I hope we have all of our listeners uh uh journey along with us and and I I think we'll have some good ideas in the coming weeks.

SPEAKER_01 53:06

Wonderful. Well, I personally look forward to all of that, and in particular, looking forward to our AI and robotics trade, which I know you know I'm excited about, and I'm hoping that this recent volatility creates some opportunities there for you, your portfolio, and our listeners. But Eric, until then, thank you for your time. I'll see you next time. The AI Investor Podcast is for educational purposes only and should not be considered investment advice.