SPEAKER_00 0:00

Eric, we've been calling for it for years. Market volatility is back.

SPEAKER_02 0:03

We could get market volatility. Volatility.

SPEAKER_00 0:06

Tech stocks plummet, then rebound, then plummet again. Is this a buying opportunity or is it more pain ahead? Eric, you used the dip to put money to work. I heard you bought 13 stocks. I'm excited to get the list of all of those, and that you have been going deep on the opportunity behind a new buy recommendation. The next Claude model, Claude Mythos, is finally here. Is it as good as advertised? What are we seeing, or did we get a nerfed model? A major AI site with three times the power needs of the entire city of San Francisco is put on hold. Should investors worry, then we'll cover a QA. Will we be buying SpaceX stocks and the 800V market plummets after a chilling new analyst research report? All that and more as we dive into the details on the AI Investor Podcast. But right after we filmed our last episode, the markets puked. There was basically a 5% plunge in the NASDAQ on Friday. Monday saw a lot of the same stocks that sold off then rebound. And then Tuesday it was down a little bit more. So the AI roller coaster ride, well, it's it's this is like this is like the premier part of the roller coaster at the end, right? Where you get the most volatility and it's going up and down and up and down, and really people are getting their money's worth on this trip. But but what's happening? Well, why did we see this nearly 5% plunge last week? And what did you do with the news?

SPEAKER_02 1:34

Yeah, it's good. It's good to be back, Austin. What an intro, by the way. Kudos. I was I was on the edge of my seat as you are introducing everything we were gonna say. Uh, that is a lot of stuff happening in one week. We are gonna do our best to cover it. But as you noted, the big news is market volatility, it's back. Uh the 4.8% plunge you referenced on Friday. Tuesday for a while, the markets were down over 4%. We're filming this on Wednesday. I'm looking at the numbers at the close. It looks like the Nasdaq's down about 2% today. Uh, the big picture, Austin, you know, you got to zoom out, right? So the big picture, the Nasdaq, it's still up 8% year to date. The SP 500 is still up year to date. But the near-term picture is definitely a risk-off scenario. When when you have the market kind of doing these, normally the market goes down faster than it goes up. We we saw the opposite recently, which is not normal, but normally it goes down faster than it uh goes up. And when you have those scenarios, the most appreciated stocks, they're just gonna get hard hit harder faster. You know, a lot of the stocks that were up something like 100% year to date, which again is not normal to be up 100% year to date in January to May, they're gonna be down in a scenario like this. If the market's down five to 10%, they're gonna be down 20% or more. And that's exactly what we're seeing. So, Austin, we've said it over and over again that this has been an exceptional scenario since March 30th. I saw the stat that U.S. large caps had outperformed the SP 500 by six standard deviations over the past 50 days. Now, Austin, I I know many people aren't there, aren't they wanting to subscribe to standard deviation podcasts, but six standard deviations, that's a big deal.

SPEAKER_01 3:24

Is it is that one in a million? Is that more than that? It's it's a vanishingly small number.

SPEAKER_02 3:30

Well, yeah. See, you're gonna get me into math nerdery. It would be one in four million, but you can't apply Gaussian principles to financial markets. And we did have some six standard deviation movements, but only last in the dot-com bubble. See, look at look at what you made me do there, Austin. Sorry, should I get the whiteboard out for the next one?

SPEAKER_00 3:49

My apologies. I need my TI need my TI 83 calculator with a little Claude plug in to get through this episode.

SPEAKER_02 3:55

I know, I know. And and you know, if people think this is bad, they should wait until the end when we go for co-package optics and 800 V in one segment. But Austin, at a high level, really, it's it's actually pretty simplistic. You're gonna read all these explanations for what's happening every single day, but it's simple. The market's rotating. You need to, and I've talked about this time and time again when we've gone through these waves. When the market is selling off like this, you need to look at sector performance. You need to say, is this actually something really germane to this sector, or is this the market rotating into some of the sold-off industries, unloved industries, some of the more conservative industries, and out of some of the risky industries? And that's what we're seeing. On Tuesday, nine out of 11 sectors were positive, even though technology was extremely deeply negative. The only other industry that was negative, well, energy. And I think people can see on the news what's going on with Iran. So, what's happening with this rotation? Number one, right after our podcast on Friday, uh, we had a jobs report, and the jobs report was good, which you might say, oh, that's great. The jobs report was good. Oh, no, it's not good. It is bad because when jobs are good and inflation's elevated, that means an amplified risk of higher rates because the Federal Reserve's mandate is to balance job growth and inflation. So, what we've seen is interest rates go beyond 4.5% again. So that's number one. Number two, Austin, market froth. Uh, we've got the SpaceX IPO coming up. There, there really are some hallmarks. I'll cover it later in a QA question. We've got on the agenda. Um, there are some hallmarks of kind of what we saw in the dot-com era, such as um projections for what the company that can do that would be relatively unrealistic. We've got open AI putting their own filing this week. So there's just a lot of froth, and that just makes people who are managing money a little nervous. We we talked about last week, even beyond IPO markets. Marvell jumped 40% last week on a quote from another CEO, Jensen Wong, uh, just saying, hey, I I think this company could be a trillion dollars, jumps 40%. That's not normal. Third, Austin, some specifics with AI. We're gonna talk about these throughout the episode. But Google last week, they had announced a secondary offering to finance their build-out. Um, it was received pretty well. I said that was actually really positive. But then Meta immediately followed. Their stock dropped. Reports were out that Amazon was lining up. And again, Austin, these are just factors that spook the market, and they spook the market even more when we've been on a continuous rally that's lasted for over two months. But you know, at the end of the day, Austin, I I know it's gonna sound counterintuitive. I know we have new listeners out there, they're gonna think this is weird. I feel a lot better right now. This is good. This is good. I feel a lot better right now because Austin, when we started this podcast, the beauty of it, and why we had no listeners, was AI was non-consensus. Across the time we launched this podcast, we saw the search volume for AI bubble explode by 3,000%. Recently, we see everyone agreeing. That makes me uncomfortable. But but now, Austin, you know, we're seeing all of a sudden questions about this all coming back. We see news, as we'll get into this episode, being framed in the worst light. We see things like reports on the AI supply chain leading to panic selling. It's beautiful. I love it. I love it because you know what, Austin? When markets get irrational, you have a higher opportunity to be able to dig in and find the true gems. And we've been in an opposite environment, as we've talked about in recent episodes, where many of the most low-quality stocks were kind of rallying on this uh kind of um unbridled enthusiasm, as you want to call it. So, you know, what we're seeing right now, it hurts. I I realize, I believe the portfolio we had hit as high as 200% per average recommendation right now. When I looked at the close today, I was down to 150. That's about what you expect in this environment, especially since you have leverage due to the previous gains. We are seeing that euphoria squeezed out of the market. So um it it hurts in the short run, Austin, but all the same, I think we've done a good job of teeing people up to the fact that uh a reckoning like this would happen, where there is a bill to be paid on the fact that stocks genuinely make more than any other asset class. And that bill to be paid is volatility and we're experiencing it right now.

SPEAKER_00 8:41

Yeah, and one of the things uh you had talked a little bit about some of the money managers behind this and the way that they think about capital. And you said, you know, they're managing risk. First of all, that is correct. They're often trained to think about risk and they are trying to manage these portfolios for endowments or institutions with smoother returns. That's the goal. And they see risk as something to manage, whereas we as individual investors see risk as an opportunity. So personally, I like this cooling off a little bit for the reasons you mentioned, takes a little bit of the steam out of it. And there's all of the obvious ways where interest rates can sort of tamp down on this in highly leveraged industries like we know, like real estate and autos definitely take a hit here. But more specifically, and you can go back to our last episode where we talked about this, debt is now hitting the arena to fund some of these build outs. So it might seem weird. Wait, you guys are the AI investors. Why do you want to cool this down a little bit? I don't necessarily want to cool down the build-out, but slightly elevated rates a little bit, it forces people to sharpen their pencils on the financing here, right? If money is too cheap, financing is too loose. So, you know, I'm I'm not saying I like high or low rates in any environment, but here's a case where, like, as debt is now starting to enter this arena, and we just had Apollo's $35 billion, somewhat you know, interesting Broadcom chip deal for anthropic that we had talked about on the last episode. You can go back and listen to that. It's not a bad thing at this point to really have the debt markets really need to think carefully about these planned and underwrite at a high level, which higher um interest rates force you to do. But we are alluding that, hey, you know, these this volatility is a feature, not a bug. It's only a feature if you take advantage of it. And I want to get to some of the stocks that you're buying at this moment. I also did some buying in the sell-off, but we know that the market is down. Of course, let's take a little perspective. I think a lot of these stocks are bombed back to like where they were a month ago. So you know, they're there they're they're still up quite a lot. But beyond all the noise, was there, you know, what stocks were you buying? And was there anything that made you more concerned or raised the risk of AI stocks in the the last, let's call it, four trading days here?

SPEAKER_02 10:45

Yeah, and so let's first unpack that risk part. I I think what you said there's very pressing, very good. That interest rates are one of the things that worries me the most. And it is a little bit because you almost need some levers, Austin. Um, we we we almost need a little bit of a release valve sometimes to something that's growing the way that AI is. I think of back we we worked together so much during the period that smartphones were really becoming the big trend of the day. And I think about Google versus Apple. And Austin, you know, you filmed so many videos with me on Apple. You probably never want to talk about Apple again uh in your life. But they achieved basically the same returns between 2011 and 2017, but one return was a lot more stressful than uh the other, and that was Apple's. And that was because Apple would rise, and then there would be a fear that they had essentially grown too far too fast, and they would plummet. And this pattern repeated several times, and they were really hard drawdowns for investors to get through. Whereas Google just kind of rose the entire time because Google's Google's growth rate was never as high as Apple's, but it was consistent. And what we're approaching right now is I know this is gonna sound crazy, but the AI market would probably be better for investors in a way if it wasn't growing this fast.

SPEAKER_00 12:12

It's the durability point. It's the durability, it's the durability point that you made last week, which I think was really, really good, which is when you start to when durability is seen to be a feature or possibility in this industry, it is easier to underwrite them to a higher, more stable multiple. Um, so and and that was what Gavin Baker was talking about, durability, I think.

SPEAKER_02 12:31

Yeah. And yeah, I mean, I I think either way, we're probably gonna reach the same point in five to 10 years. We're probably gonna reach the same point in five to 10 years. It's just a question of whether or not we have how many severe corrections that really wash people out we have. Um, and you know, the the faster it moves in the interim in a way, the higher probability that's gonna happen. But Austin, that's that's the macro news. In the weeds this week, I thought there's some really encouraging stuff. We can put this up in our show notes, but Noam Brown, uh, he works at OpenAI. He wrote a very technical blog, but basically he he posted data on why the dominant technique for improving LLMs, which is test time compute, basically we still don't have the capability. We can't measure the ceiling of it because we don't have enough compute, which speaks to this need for kind of this enduring build out and and at what point you reach these diminishing returns on intelligence and and how far we have to get there, which is which is really bullish for these build-outs at the most base level. Another storyline we'll talk about a little bit more later, but Claude finally releasing Mythos, they rebranded as Fable. Uh, the actual launch, uh, it's far from perfect. There's a lot of uh concerns about guardrails, but the actual benchmarks of it. Um, everything from Austin, you can test on coding, knowledge work, reasoning, biology, cybersecurity, health, it's across everything the best model every ever released. And you a last one, and you actually started alluding to this earlier. I I don't know if you're gonna know what I'm gonna talk about, but we we kind of had the birth of a new hyperscaler as well. And a lot of people say, okay, um, you know, we can get this much money from Meta, this much money from Alphabet, this much money from Amazon, Microsoft. There was a very material piece of news last week that I don't think a lot of people scratched at or realize as much as it was. So when I look at the, and I Austin, I know that this is a recurring thing. When I look at the actual structure of AI news this week, I thought it was very positive. When I look at the market conditions, it was very negative. And, you know, when that happens, um, it's gonna lead to when when stocks drop in the interim. I'm I'm gonna be looking for new buying opportunities.

SPEAKER_00 14:46

Well, look, let's get to those buying opportunities. I'm very excited. Most episodes, you might talk about you know, four or five different companies, one of which you're excited. And we're gonna quickly run through 13 stocks you actually bought with your own money. So this is very exciting. I know our listeners want to hear that. They're gonna have to wait another one or two minutes, though. Um, I am really excited by the the mythos fable release here. And I'll be the first to admit, when I saw the first rumblings about how capable mythos was, you were a little more optimistic about them than I was. And I was like, uh, this feels like good marketing. This feels like good branding. And one of my data points for that was in the first few weeks after that, when they were talking about the cybersecurity holes that Mythos was was identifying. We just we were not hearing from a lot of people inside the firms using it saying, oh yeah, this really did find all these. So it felt to me to be a little bit of smoke and mirrors. Then lo and behold, maybe four weeks went by. We started to hear those accounts, and people were like, oh no, it actually did exactly as it said it would. And if you're an active user of Claude, and we talked about the flurry of features that they were rolling out over the last three months, those were all powered and created by Mythos internally. So absolutely validating everything they said it was, even if Fable, you know, is the new name and it seems to be a little bit nerfed in um safety-wise, capability-wise, it is truly astounding astounding, uh, absolutely incredible there. And I know OpenAI is waiting to release their most recent um version that was trained on the most recent NVIDIA clusters, and they're already looking forward to. I think it was CO Sarah Fry just started talking about how much they're planning for uh what is the next chipset that's coming out to Ruben? It's at um Feynman, right? Um, and and so we're now gonna see their first big model trained on the new Blackwells, which I expect will be uh competitive with Mythos. So I'm excited for that. I don't know if you have an ETA on release date yet or if you've heard anything. Um, so if you have, let's hear it. If not, let's get to what everyone's here for, which is the 13 stocks you bought when the market plummeted. Where were you putting your money?

SPEAKER_02 16:40

Well, we'll we'll talk a little bit more about uh mythos slash fable later, but let's let's talk about the stocks, as you said. I I did 13 purchases, I actually probably did 13 purchases. I bought a few things multiple times, but 13 stocks. So let's just run down through them. From my personal portfolio, these aren't things I'm adding to um the AI portfolio. I did Ouster. I would give that a risk level of high. Austin, as we've talked about before, you we would love to make more robotics uh recommendations. A lot of the supply chain is in Asia. I believe Ouster remains a company with relatively differentiated technology. It it had run up on a lot of enthusiasm for robotics when it gets pulled back. I just want to get a little bit more allocation in my personal portfolio. I was a little underweight that um Quanta Services, which is sticker PWR. John Rotanti talked about that. I would I would say that's one's probably a low or medium risk. We had we had talked about that being um kind of exposure that's different than what a lot of people might have, especially if you're buying chip stocks or you know, some of the other plays that are popular. So I just wanted to take the chance to diversify it. I'm not putting in the AI portfolio yet, but we we may do it. I think I was down another 6% today. So the risk reward keeps getting better. CamTech, I've talked about how much I really like the semiconductor equipment play. So I just wanted to continue adding that. That could be one that we continue adding to the portfolio. Broadcom, um, which most people know is AVGO. I'm gonna talk about that a lot more in a minute, but I that was my most substantial investment of this week. And NVIDIA, I just keep stacking my ownership of it. I think it's relatively low risk level, especially in the AI space right now. Now, Austin, the other thing I've talked about on this podcast a lot, I had been unable to purchase stocks due to disclosure rules roughly between October and when some of our consulting work had ended up. I no longer have those rules. So with them gone, I'm catching up. So I bought on semiconductor, ACMR, Anto, AMSE American Superconductor, which has been dropping quite a bit. I think might be a pretty good deal here. Um IES Holdings, uh ST Micro, Nevitus, Wolf, and uh did I already say ST Micro? I might I might have said that one twice. But I purchased them all for the portfolio. And I am waiting on a few stocks people might know. So Liberty, uh, Axtron, Intel, Air, uh, Semtech. These all run up an especially high amount since we had recommended them. So I'm holding them, but I I did want people to know I I had said I will make this commitment to purchase them, even if I'm getting substantially worse prices. This this isn't a shots fired at anyone else doing this, but you know, you see so many people who go out and post things, they make no claims to buy after the recommendations. I will only purchase after I recommend things. I will only give the best price to my listeners. Um, so you know, if that means being taking an extra price.

SPEAKER_00 19:45

Shots fired at all those Twitter accounts. Eric, you're too polite, but they are cowards. You are a man of your word, and our listeners appreciate that.

SPEAKER_02 19:54

Well, so that's that's the way we'll always do it, though. This portfolio is never about how much money I make. I've got my own portfolio where I invest in other things. And um, you know, this has always been about getting listeners the best returns possible. But so we're Austin, 800 Vs a space with a Love Movement, Wolf and Navitas. We're gonna cover that in another segment. Um, I I do want to talk about, I mentioned Broadcom's my biggest acquisition. So I do want to go into a little bit more detail on that.

SPEAKER_00 20:24

Okay, so let's transition to it. Broadcom, your biggest acquisition on the dip. Uh I think the narrative here, I'm curious to hear what you say, and I know we'll get into the numbers here, but it sounds like this is a story of incredible earnings coming down the pike for the next few years that we can see, right? We have visibility into 27, 28, whatever earnings. And a lot of these stocks on a forward earnings basis, a lot of stocks in your portfolio specifically, I think Meta, Nvidia, Broadcom, Micron are probably, I'm I might be misquoting here. I think they're all trading for meaningfully less than 20 times forward earnings, like meaningfully less. Um, I'm interested that you are adding Broadcom at these prices because Broadcom was one of the very first positions you added all the way back in 2024. Have you bought it twice in the AI portfolio? Yeah. I see that you bought it twice in 2024, skipped out on adding to it in 25 as it run up. Very sharp pullback last Thursday. So, what is it that you're now seeing about Broadcom? And the bear case here, right? And we alluded to this a little bit in our last episode. The bear case here seems to be yeah, these stocks, the ones I just listed, are all very cheap on a forward earnings basis, but we only have earnings visibility out two or three years. So they look cheap for two or three years, and then what happens next, right? Is that fair?

SPEAKER_02 21:45

Yeah, and I think the the case I'll build for Broadcom right here is that Wall Street is dramatically underestimating what their earnings can be when we get to the end of visibility. So if we look at Austin, as you alluded to, it was at 480 before. Earnings it's at 370 now, so it's 22% cheaper. So that's a big haircut. Well, the stock might look expensive, especially if you look at trailing PE. I think calendar 2028 EPS will be above $30 per share and potentially actually much higher than that. So Austin, that is when we get to the end of visibility that we have right now, I think Broadcom's trading for about 10 times sales of where you can plot out to, which is a significantly better value than the market, and it will be showing exceptional growth headed into that period. So as I note on the last podcast, Wall Street, the the big reason for the sell-out sell-off was essentially that they didn't raise their prior guidance. But as we had talked about, they had already raised it the last quarter. Um, all of Wall Street knows they're gonna beat their guidance. It's it's modeled into their forecast, it's a little silly, it's it's basically semantics. I think the CEO of Broadcom got caught a little off guard by this. And I've talked about this a lot, Austin. When we do our live blogs on 24 Wall Street, a lot of the time we do break major news that ends up becoming very material. We do catch things. And as I was listening to Broadcom's earnings call, Hoctan, who's their CEO, he sets up these big narratives and he waits. And when he finally releases them, the stock goes big. The downside of that is when he doesn't release one, the stock often craters, right? Because people were expecting something, a juicy little morsel. And they they didn't get their juicy little morsel. But what they weren't listening to was reading between the lines that he's setting up for a future major revision to 2028. He said, through 2027, that's all booked. We know what it is. It's coming into 2028 that that's where the crux is. That's what he's telling everyone. And he's also setting up that they have some major deals happening. And Austin, the background is Broadcom's major client is Google. Most people are selling Broadcom off because of this Google relationship. And the reason why is because they say Broadcom is collecting too much money. Google's gonna give increasingly their allocation towards a cheaper company named Media Tech, who's building their own version of the TPU, and that's gonna lead to negative growth for Broadcom. I believe they have this backwards. What Broadcom what's happening is Google has had a question of whether or not they're going to actually sell their TPUs externally. Austin, the TPU is the number one driver for Google Cloud. Google Cloud grew 63% last quarter. Google Cloud's probably a trillion dollar asset. If Google is going to allow TPUs to be sold externally to Anthropic, to other companies, well, it better be worth their while, right? And to be worth your while when you're Google, that means 20 billion, 30 billion, a lot in profits. They're letting this happen. I believe they're only letting it happen because the payoff is so big. And what Broadcom has formed is its AI XP platform. They're trying to enable 20 gigawatts of compute to Frontier Labs through 2028. Just consider this, Austin. Broadcom can make something like I think $20 billion a gigawatt when you factor in its TPUs and also networking. They're talking about 20 gigawatts. I I mean it might even be 30 billion. I I don't know the specific number, but the point is they launched a partnership. You mentioned it. It's with the biggest PE firm, some of the most respected, Apollo, Blackstone to get started. I believe Google's actually gonna put financial backing into this. And and what we saw is in this past week, Mizuho, they put out a very controversial note. But what their findings were is potentially up to 35 million TPUs in 2028. A $600 billion opportunity. Austin, Broadcom's entire revenue in 2026 is 106 billion. Wall Street expects 228 billion in 2028. You know, as I said, there's a lot of debate. Is is this forecast nonsensical? Well, you know, Austin, it it might be too high. With the delays we've seen, that is such an enormous amount of compute to be built. I don't think it can happen. But even if it is shooting too high, the other side of this is even if this is fractionally correct, yeah, it certainly seems like Wall Street is undershooting 2028.

SPEAKER_00 26:49

I was just going to say that. This might maybe if this is even partially correct, uh, it's huge for investors, right? And maybe it's wrong, right? So but the butt the outcomes here are binary, right? Either Wall Street's right and this is maybe too frothy, or they're wrong on the upside and this is enormous opportunity. But so, so it but but that upside continues to grow, right? So on a risk-adjusted basis, it's worth buying that that ticket. But also keep keep in mind the structure of the TPU deal that Apollo and Broadcom and Google and they all just rolled out that we talked about in our last episode. I believe it was Broadcom that was backstopping the price of these chips to guarantee the loans. And that who is gonna understand the value of these chips in the market better than Broadcom? When they have customers coming to them saying, we're gonna pay this price for this, we're gonna pay this price for this, we're gonna pay this price for this. They understand what these chips are worth and they're in the best seat to make that call. So if they're willing to backstop this debt in order to get the deal done, it's because they know that the price they would have to pay if the world goes to hell in a handbask and they actually have to pay that price, they will still make money, right? The reason they're backstopping that is not out of desperation. It's because they know that the price they're paying is still lower than what they could sell them for.

SPEAKER_02 28:02

Yeah, and it certainly introduces some risk. You know, I don't want to be blasé about this, but Austin, here's the thing. I I think when you look at published Wall Street estimates that have them doing something like $24, $25 in EPS a share in 2028, that's mostly modeling of Google. That's mostly modeling of their scaling relationship with Meta that's happening right now. That's that's mostly their open AI. They they have six different major partnerships going on. And XPV is a substantial cream on the top that I don't think is even modeled in a substantial way. It was just created. So I think what's gonna happen, Austin, I think on their next investor call, I think Hocktan, he's gonna give a lot more specifics. He's gonna basically give Wall Street a here's exactly how you should model it. And I believe it's gonna re-rate where you're gonna have a company coming into 2028 at absolutely astounding growth rates. Um, that's that's gonna be priced for, you know, 12, 13 times whatever it might be at the moment, 2028 earnings. And I believe at that time it's gonna force a pretty compelling re-rate for Broadcom. So maybe I'm wrong, but my conviction is high enough. It's now my second largest holding behind NVIDIA. I want to add today another $20,000 to it in the portfolio. And Broadcom is not a stock that you expect. I mean, it's a $1.7 trillion company, something like that. This moment, it's it's not going to be the high flyer of the portfolio, but you need things that are ballast, right? You need things that are going to be outsized positions. And I believe this pullback presents extremely attractive opportunity to add Broadcom today because I I do believe they have made a major pivot towards their business that is so new and so underappreciated that investors have what amounts to a temporary opportunity on it.

SPEAKER_00 30:04

Okay. I love that. And this is certainly not portfolio advice, but when I hear that narrative, and you've talked about the CEO of Broadcom and how he he's got like a showmanship-like quality, right? And Wall Street expects that. And then when they don't deliver, the shares sell off when he delivers on it. You often, you know, they get excited. It's the one more thing moment with Steve Jobs. This is not portfolio advice uh or direct investment advice. But when I hear that, it feels like given the risk-reward profile of a Broadcom right now, optkins aren't a bad way to play it, right? You could do a position in Broadcom, or or if you wanted some exposure on the other side of this, maybe in the next earning call, you expect Hocktan to maybe, you know, do the one more thing, reveal, um, and they re-rate. Put a, you know, you could you could have a have some fun in a retirement account with some options. Again, not portfolio advice. I I pre I generally don't have success with options. This is one that I will probably buy the underlying shares of, but hearing this moment in time, that might be that might be a little fun way for our listeners to, you know, experience, you know, have have a good day after earnings. We'll see.

SPEAKER_02 31:06

You know, Austin, you this is why you're my muse. Um when when I kind of brought up the risk reward profile for Intel, I remember you immediately were like, actually, what you're saying is perfect for options. I just think about buying stocks first. I I do buy options, but I I just biased towards stocks. And um, I think you're right on this. I do think you're right.

SPEAKER_00 31:29

Okay. Well, that's not an official call in the AI portfolio, but if someone wants to, you know what? Some of our listeners out there who do who missed the actually, you know what? I think if we go back to that episode, we also said buy call options on Marvell. Um, and that would have been a hell of a ride. So if anybody decides to put options on this one, let us know. Uh only if you wouldn't know. Uh okay, enough on Broadcom. Let's talk, but but speaking of TPUs, right? Staying in the theme of TPUs here, let's talk about the most recent Claude Mythos release. I'm just gonna keep calling it Mythos. This is sort of like Twitter, right? Twitter was rebranded to X. It's always Twitter in my heart. I'm sorry, it's never gonna be X. This model is always Mythos, um, which was as far as I believe this was largely trained on TPUs. So a nice segue there just about the how incredibly powerful this chip is. And I remember maybe a year ago, I know I'm really tying this back to the last segment a little bit, but one of the debates was like, you know, NVIDIA versus you know, NVIDIA's approach versus TPUs, NVIDIA's chips being way more capable, TPUs being much more narrowly capable, but extremely energy efficient, which are in sort of like different paths to training a model. I I believe this most recent cloud model was almost exclusively trained on TPUs. Um I'm not sure if that's correct or not, but what are what are the benchmarks that we're seeing about this model? Yeah, yeah.

SPEAKER_02 32:46

You know, I'm I'm yeah, I'm actually not sure on that off here. It's interesting, you know, Apple, they just had the relationship with Google for the new Sierra AI. They're gonna use NVIDIA chips. It could partially be to that versatility you were talking about. But Austin, I I do think talking about mythos is is really important. Like you, I am just team Twitter forever. This X stuff, no way, no way. Um, from a from a top level, we we mentioned this earlier. The benchmarks are incredible. Um, they're better than any model ever. That's just it's it's simply the point. I talked about all the different disciplines. Um, there are some downsides. It's gonna do some things like hallucinate more. Models, they they've gotten a little um lumpy in the fact that they don't universally improve. That that could be part of a problem. We're gonna need a underlying architectural improvement. But hey, that's that's a topic for another podcast. Second, Austin, I would say the model is it's really locked down. Um, consumers are gonna hate this, but there is some background here. I think we're gonna see more of this. I think the companies are gonna say it's a security problem, and there's some truth to that. And then there's some truth to the fact that, well, Chinese models like to distill them. And uh, if you want to get around distilling, uh, you lock them down. And actually, Anthropic has explicitly said this. They basically said if you are doing any queries that would help you in construction of an LLM, we're blocking you. And I think that's just what we're gonna see over time. Um, it's IP, you gotta protect it. There's no IP protection from the main open source competitors, which are all Chinese. It's just probably a fact of the future. Uh, so Austin, the third point I would want to make about this is anthropic, they're winning. They're winning. If they were a public, they're probably worth. I I think Gavin Baker had said he would 2x. I think that's pretty accurate. So if they're right now just under a trillion, they're probably worth 2 trillion if they were on the open market. A key reason for that is they specialized in coding. And that coding advantage was fed back into model development. And what's happened, Austin, is they've had better models available internally, and they've used those better models to build all these products that get so much attention that are very good, right? Like cloud code, it's great. Cloud design, blown away how good it was. Cloud co-work. They've built all these things because they had this model advantage, because they specialize in coding. And OpenAI realized a little too late. Coding was the big thing. They push into codex, and now they're having to play catch up. So, what does this tie into a big picture, Austin? Well, we've talked from the beginning of this podcast of the you could call it holy grail, you could call it whatever you want, but recursive self-improvement models that make themselves better. And in this past week, we saw Anthropic put out a big white paper on how much recursive self-improvement progress they've seen. There's there's a lot more room to go, but how much they've seen. And just this week, as we alluded to earlier, open AI filed to go public. And and what did we see in an email today? Well, there's probably a little bit of, there's probably a little bit of uh, oh well, look how valuable we should be in this. But Sam Altman, he said the one thing that would keep them from IPOing is um if the progress towards recursive self-improvement speeds up, uh, you know, they would rather then stay private. Um, so Austin, it is just a matter of we've talked about what what reaches this end state of the value creation of AI and what could it mean to the economy and world and recursive self-improvement is the holy grail. And um mythos is definitely driving in unprecedented ways towards that. And uh it's starting to get very real. And if that gets very real, the compute build-out um is gonna accelerate in a way I think few people would ever expect.

SPEAKER_00 36:52

And going back to the compute point, um, I did confirm on that. So Mythos was trained with one gigawatt of TPUs, and they have announced the intention to expand the next model train up to 3.5 gigawatts, plus some plus some Amazon training in there. So the amount of power that went into this model would roughly 3.5x, plus all the additional capabilities you have and develop along the way in terms of software and more efficient training and recursive self-improvement. So, what might be a 3.5x increase in gigawatt usage might actually be more like you know, four or five X once you layer on the additional capabilities you gather along the way. Pretty immense to think about. I mean, when we think about the leaps that we've seen in one model generation to the next and anthropic really breaking out with cloud code in this most recent generation, uh I'm a little dumbstruck to imagine what comes next when you three to four X the amount of gigawatts that go into that. And I'm particularly excited for when we start getting to like robotics-specific model training, which we're not gonna, you know, we're not gonna talk about in this episode, but you know that's my you know that's my thing. I want to see AI in the robots. I want it to be able to affect people's day-to-day lives and you know, break away from the keyboard and start folding laundry and mowing lawns. Um, so anyway, it it's humbling to think about the scale of power that we're talking about here. And speaking about the scale of power, there was a major AI site that has three times the power needs of the entire city of San Francisco and it's now been put on hold. And as you know, Eric, Eric, you you've had some great rants on this podcast. I'm not gonna try and go toe to toe with that, but I'll say the the announcement around these really bother me because these are high-paying, high-quality jobs. They're a national security uh you know beneficiary. We're not talking about these manufacturing facilities that are putting uh coal uh fumes into the air. We're not talking about these massive manufacturing facilities in Shenon that are ruining air quality. We're talking about like possibly one of the best gifts to all of these local states and jurisdictions. And I understand maybe not wanting it directly in your backyard, but at some point, like you just gotta say, like, we got to do this. The town that I'm working that I live in has a preemptive negotiation uh conversation right now about um preventing data center build-outs. Let's be clear, they're not gonna try and build data centers here. I'm in Vermont. They know better than to try and build anything larger than a chicken coop in this state. You can't do it with Act 250. But Wyoming, come on, Wyoming. Wyoming, you're supposed to be where people can do these things. Anyway, Eric, what's what's what's going on with this uh 1.8 gigawatt data center? Why is it getting blocked? And should investors worry?

SPEAKER_02 39:32

By the way, I I love the idea. You know, you're like, I want robots to fold my laundry, mow my laundry. We're gonna have robots doing that. I'll be like, hey, Austin, robots are mowing your lawn. What are you doing? You're gonna be like chopping wood in a shed somewhere.

SPEAKER_00 39:46

As you said, as or you know what, Eric, as you said, I am the bridge between AI investing, an Amish lifestyle. I don't hide from it. I don't hide from it, right? I drive a vehicle from the future, but I insist on chopping wood by hand. I can't explain it. It's a barbell approach to life. Get over it.

SPEAKER_02 40:04

You did get me into the ribbon, and that's been a godsend. So I appreciate that. That that'll be your technology contribution. There we go. So, Austin, the the news here I wanted to talk about, as as you allude to, um, there's a 1.8 gigawatt data center plan in Wyoming. Uh, on Tuesday it came out, plans were no longer moving forward. We've talked about a lot. If you start with the all-in costs for a data center, something like 50 billion per gigawatt. Well, 1.8 would be 90 billion. I believe this project was supposed to scale to potentially 10 gigawatts. That would be a lot of billions. Uh, you know, it is three times the power consumption of San Francisco, one billion times approximately the power consumption of South Burlington, Vermont. Uh, so Austin, there's a lot of conjecture about why it was canceled. People didn't know. Uh, they and that leaves people an opportunity to fill in the blanks. Uh, you know, Google and Meta announced their secondary equity offerings and their share prices fell. Is that why it was canceled? Was it being canceled due to lack of demand? Was it being canceled for regulatory issues? People didn't know, and it was adding to kind of this echo chamber. And what we've seen is with its cancellation, a lot of infrastructure stocks, especially stocks in the vein of what Rotanti was talking about, we've talked about in recent episodes, Blue Energy, Quanta Services, Fluence, they've taken outsized falls in the wake of this news. Well, today, Wednesday, June 10, we we got some more details. What's happening is there's a developer of this project, which is Crusoe, which is a developer of many major projects, they're a developer of uh the Stargate project I went to. They're essentially getting the boot on it. And the energy company behind the project, I think it's Black Hills Corporation, they say they're still moving forward with it. So, Austin, a few things on this. Number one, the AI build-out can exceed all current expectations with a lot of projects getting canceled. We are going to see so many projects getting canceled. We'll never get tired of seeing them canceled. Because Urcot, um, which is the Texas, um, the Texas, I I don't even know what you call it right now. I I I woke up 13 hours ago and I've had like 32 ounces of cold brew, Austin. So if I get too out of line, you gotta stop me right here. But they have 438 gigawatts. So if you do the math, 438 gigawatts times 50 billion approximately per gigawatt. That's $22 trillion. Austin, $22 trillion of data centers is not getting stamped out in Texas. It's not getting stamped out for years and years and years. The US economy is $32 trillion. Crusoe alone, Austin, their pipeline, their development pipeline is 40 gigawatts. So a 1.8 gigawatt project, that would be like uh Arby's saying, well, we're canceling our location in New London, Ohio, and everyone going, oh, the Arby's build out's done. No, it's just New London maybe don't work. So the point here is be careful about extrapolating a single project as the direction of the infrastructure build out. Austin, we saw this before. Remember when the Microsoft projects were getting canceled last March and everything tanked. It was, it was like actually worse than Deep Seek. And on this podcast, we kept saying over and over again there are a lot of reasons for these projects. This does not infer the infrastructure build out. This is actually a time to buy. So, you know, Austin, it's just when volatility is higher, people read into this news in different ways. And um, this this seems relatively healthy. It's it's a single project, it seems like it might be going forward, um, etc. So that that's all I gotta say about this.

SPEAKER_00 44:02

Nature and capital abhor a vacuum. And this in this industry is too important right now. These are going to get built. It's just a matter of when and where, right? And at some point, if there's a community that is open to them, they will they will reap the rewards from it. And you know what? I uh speaking of vacuums, maybe capital doesn't abhor a vacuum because this actually starts to make the space-based data center thing kind of make sense. If it's like we don't have to deal with all the permitting and all the the pitchforking and the and the local municipalities, we can just put it in space. I mean, damn it if if I'm not suddenly going to become like I'm gonna go from making fun of these uh data centers in space to actually understanding and supporting them on account of this. But I get it, right? If it's like, well, we don't want to deal with uh, you know, the the local local town of uh 2,000 residents here who are pitchforking, we can just put it up in in space and get it running in half the time, then we should do it. Okay, Eric, we're running a little long here. Do you still want to do an 800 V segment?

SPEAKER_02 44:58

Yeah, let's do it. Okay now. I it will be my greatest challenge ever that people have gone through 45 minutes. I would like an advil counter on the episodes on YouTube where it's like how many advils have been taken as people try and get through all our complexity to end with a Sigma co-package optics.

SPEAKER_00 45:15

I want to I would be I want to figure out what the correlation is between Eric cold brew and listener advil consumption. I'm betting they map pretty closely, right? The more cold brew that you take, the longer we go, the more advil they need. So I don't know. You know what? If there's a data analytics nerd out there who wants to try and put this together, by all means, please do it. But we'll we'll we'll share with you Eric's cold brew consumption so that you can uh you can get us a report on this. Speaking of reports, you've been talking about 800V and the need to manage power differently. I'm not gonna re reiterate that thesis here. People can go back and listen to half a dozen episodes to get that. Um, but semi-analysis came out with a report that NVIDIA is delaying the 800V transition, that they're struggling with co-package optics, and that the launch will be delayed. Now, similar to the build-out point you just made, delay is the crucial word here. This is not terminated, this is not stalled, this is not tech sealing. This is delayed due to complexity, or, you know, in the data center case, maybe delayed due to objection or permitting, but it is inevitable that this is coming. This 800 V architecture is coming. It's just a matter of the timeline being extended a little bit. Hopefully, that gives us a buying opportunity here, including some of the more speculative stocks in the portfolio, like Wolf, which I know you're gonna talk about in a moment here. So talk to me about this report from semi-analysis, what you extracted from it and and why did the market react this way? It seems outsized to me, given the fundamental point here, which is just this is incredibly technologically difficult rollout, and it's gonna take a little longer than thought, but it's still coming.

SPEAKER_02 46:44

Yeah, I I think you nailed it there. Uh, I will do my best to simplify this, but um, so we've got two parts of this, Austin. Um, number one is the 800V delay, which the most speculative stocks. We we've said since the beginning, the the most speculative stocks for 800 V are gonna be something like Wolf and Nevitus. They dropped 24-28%, respectively. There's a second part of this report, which is about co-package optics, and we saw a drop in everything optics-related Lumentum, applied optoelectronics, coherent. As far as our portfolio recommendations, it that had a bigger um impact. So, as I mentioned earlier, Austin, and this is what we need to unpack. When market instability is higher, you see often, in a way, you see illogical reactions to the news on the way up. Like I we called it out. I mean, like Marvell going up 40% from Jensen Wong saying nice things about them is illogical. It should have been 100%. It's logical. It should have been 140 in my mind. Well, you see illogical stuff on the way down, right? So, all right, quick explanation. We've done this a few times, but I'll do it one more. Um, for reasons we've explained, the future of networking, it's gonna increasingly be optics. We've talked about limits to copper, etc. Right now, most sales in the industry are pluggable transceivers, which you simply you plug into a rack. The nice thing, Austin, about this is when they break, pop it out. The problem, optics, it's all about transmitting light, right? That's that's the core reason you run optics. But when you hit the back of a server, when you've got these pluggable optics, it requires a run via copper. Well, copper's not transmitting via light. It's now an electrical signal. And also it might sound trivial that you've got a run of a few inches where you're going behind this ASIC switch, uh, that you're gonna be able to take it to that next level and your pluggable transceiver, but that run over copper, because of the networking speeds we're working with, it's tremendous signal degradation. So you need something called a DSP. And those are very signal, or I should say they're signal hungry, they're power hungry, is what I meant to say. So the drive to co-package optics means you have an entirely new infrastructure for how you handle this. You basically mount an optical engine right next to the switch, so it's gonna use a lot less power, but it's just it's it's tremendously complicated. Um, it it's new technology, it's difficult to package. And also, last but not least, I kind of brought up the fixing angle. If it breaks, it's it's not easy to fix. So you need it to be incredibly reliable. So, what semi-analysis report says is effectively co-package optics, they're just proving really hard. They're hard to produce, they're hard to be reliable, and NVIDIA is gonna move their launch back a little bit. Um, so in the near term, what they said is they're gonna move to a new approach called near package optics or NPO. So we've got co-packaged CPO, NPO, near package optics.

SPEAKER_00 49:56

I wonder how they came up with that branding.

SPEAKER_02 49:58

It it is it is quite beautiful, Austin. Um, but yeah, it's just it's a little less it's it's the same components, it's just a little less difficult to package, essentially, is what it comes down to. So the companies that many people are buying for co-package optics, which we've talked about, Lumentum, Coherent, um, they're gonna be suppliers to MPO as well. Um, in in fact, there might be scenarios they actually make some more money under this. And second, if co-package optics are delayed for reasons like it's too complicated, well, think about a company like Applied Optoelectronics. I know many people might buy it and not totally understand what their technology is, that's okay. But they're going to be big unpluggable transceivers. If their market is threatened by co-package optics and co-package optics are potentially moving back, well, that's actually maybe a positive scenario for them. But Austin, what we saw yesterday is just everything's selling, everything's selling. Applied optoelectronics is selling just as much as Lumentum, is selling much as much as coherence. So I I think my point here is, and this is gonna sound a little like snooty. Call me off for snootiness.

SPEAKER_01 51:12

Uh, it it felt like a little bit of uh technologically illiterate sell-off in the sense that Whoa, whoa, whoa, too many cold brews, too many cold brews. Another cold brew moment. Eric and Eric strikes.

SPEAKER_02 51:27

You've got people piling into the industry, and those same people who pile in because it's a growth thing, they they pile out when there's bad news. Um, but Austin, you know, it's it's worth noting. It's Wednesday, June 10th. I'm pulling up um applied opto uh AOI stock.

SPEAKER_00 51:47

They were up about 7% to 7% and they were down two and a half pre-market.

SPEAKER_02 51:53

So, you know, I I I imagine that's just some for lack of a better word, like rationality coming back where there's probably a lot of Wall Street notes saying, calm down. This is actually fine. If you're selling them off for this news, it's actually positive for them. So that's one. Could could you believe that's just one, Austin? Now, number two is um, well, actually, there's something in between those. It gets even worse for people saying I've had too much technical stuff. Uh just showing where the market is, semi-analysis this week, they had another report on memory. I think the headline was like a pun about memory stocks. It was actually relatively good news, but the pun read bad. So you saw memory stocks sell off. And by the time people realized it was a pun and the news actually wasn't that bad, they kind of rebounded. But Austin, that that shows where the news cycle is for this. Well, and how efficient markets are. I mean. Well, I mean, semi-analysis, you know, I've talked about them a lot. They will move stocks as big as Amazon. They put out a note on Amazon, they will move two trillion dollar companies. And the last one I wanted to talk about is 800V. Um, I I should note this note, you can't go read it, even if you pay the $500 for their, you know, normal service on Substack. This is for institutional clients only. I've seen a little bit of it, I've got the gist of it. Basically, the 800V push, which we've talked about a lot, that had been planned for 2027, it's it's they're moving it back. And NVIDIA has denied these reports, they've also denied co-package optics. Partners have denied that. Do with that information what you will. Other Wall Street research firms have denied it. Austin, if if anyone here, we're we're talking about this next NVIDIA cycle. If this is the first time they're investing through it, buckle up. When Blackwell was coming out, the amount of research notes that the sky was falling, that everything was bad, that liquid cooling was dead, that this and that and that were just like out of control. And you know what made money? Everything those reports said was going to pot. Um, so so just keep that in mind. If this is your first rodeo, we're we're gonna see a lot of this, especially with some of these like newer architectures. And also, I think it's happening a little bit, as we've talked about with the 800V, it requires a new architecture and requires an architecture starting at the data center level, how you build data centers. Nvidia is working with a lot of hyperscale customers. A lot of these hyperscale customers want something intermediate, they want 400V. It's it's something that they've been working on in the interim. So, what NVIDIA is going to do is they're gonna accommodate them to get Ruben rolling out. But Austin, at the end of the day, the the thing with this 800V transition, it it doesn't matter in a way if it's 400V, 800V. The what's happening is the density of these racks is getting higher and higher, and we're going to need to solve that. And the way to solve it is eventually going to be a lot higher content of these power management chips. So at the end of the day, Austin, this is a transition we expect to last over several generations. We expect the companies that are in the catbird seat to benefit over all of them, and we're not buying them because of their revenues in Q3 2027. We're buying them because of a transition that should take place between 2031, 2032, etc. So I am not overly bothered by this. I'll say for the companies we've recommended in this trend, ST Micro on Semite were our more conservative plays. I added ST Micro to the portfolio this week. I did the buy. The more aggressive ones, Wolf, Nevitus, I actually added those as well. We've got air test systems, which is also risky, but has a few other trends to play with. Once again, we've talked about this so many times. If you want the full breakdown of this opportunity, go back to our March 13th episode. But the the absolute TLDR of it, Austin, is I would kind of love this sell-off to continue. I would, and I know that hurts. And I know if you're someone who recently bought Wolf and you're saying, I'm down 25%, it's going to be extremely risky. Um, Nevitas will be the same. Uh, air test systems will likely be the same. But if this is what we believe we're buying it for, which is a long-term trend, I would welcome some better prices of it because it had gone extremely overheated. It had had a lot of money rushing into it and sucking a lot of that money out for short-term reasons. For some, that's gonna be long-term good. I'll always take that. That's how you win the market.

SPEAKER_00 56:37

Oh, sorry, I fell asleep during the technical section of that. Um I'm I'm gonna try and do our YouTube listeners a favor right now. You know, you know, I'm a I'm a little bit of a memer. So I'm gonna try and simplify everything you just said about co-package optics down to an image. And Brad, if you can find the image of the evolution of the SpaceX Raptor 5 engines where it goes from extremely complicated to uh the really beautiful simplified engine at the end, that's the advantage of co-package optics, right? It's it's fewer moving parts, it's fewer things that you have to connect, but it's new technology and you have to iron out all the difficulties on the way there. So uh I hope our listeners appreciate that visual explanation of the benefit of going to co-package optics so they don't have to listen to the technical version again. Uh Eric, we're that this is a longer episode. We had talked about doing some QA. Do you want to answer those now or do you want to try and move those to a QA episode for next week? What do you think?

SPEAKER_02 57:27

Yeah, let's do it because we moved last week. So we'll do two questions and then we'll push some of the other ones to the next episode.

SPEAKER_00 57:33

All right, let's do it. So QA. We got one from uh Henrik, which said, Great show, guys. Couple of questions I hope you can address. Uh, is the thesis for Sienna broken? Ticker symbol C-I-E-N. The stock has been hammered since their earnings call, but I can't figure out why. So that's part of the question. And the second one is what is the general framework you would recommend for allocation into and out of the AI trade? For example, is the AI infrastructure trade over when the rate of change in CapEx spend by hyperscalers is going down. Really good questions here. I'm curious to hear your answer in both. And uh the rate of on regarding the second one, the rate of change of capex is an interesting thing to home in on here because the scale of it is still so big, but also law of large numbers. At some point they slow down. But anyway, Eric, take it from the top. Let's look at Sienna rate of change on CapEx. I'm curious to hear your thoughts there.

SPEAKER_02 58:23

Yeah, and Henrik had a third part of the question, too, and I didn't quite understand. So if you want to refine that, and we'll maybe do it again in another episode. Sienna, uh, this one I think is relatively simple. They were $600 pre-earnings, they're $430 now. Uh earnings, honestly, solid growth, margin improvement, exactly what you want to see with the company. New uh product traction, they've got a new product named HyperRail. Uh, it's a scale across solution. We we own Sienna because we want them for scale across, which again, it's it's actually connecting multiple data centers together into one networked whole. This is this is at their core, what their thesis is. Their backlog Austin continues growing. But as we talked about with someone like Lumentum, Lumentum could grow a lot higher. There's just so many cons supply constraints within optics. You are just fundamentally bounded in how much you can grow. So, you know, Sienna the bottom line, you gotta zoom out. They're still up 76% year to date. Um, they've got 654 and expected EPS this year. That's up 150% year over year. They're expected to hit 1437 by 2028. I would imagine the vast odds are they will beat that by uh reasonably high margin. But Austin, $600 a share divided by 1437 in 2028. That's 40 times their PE two years later. Um right now, with their sell-off, they're at 30 times. So, you know, Austin, sometimes it's just you when risk is flowing, we talked about Broadcom being 10 times potential 2028 numbers. And and maybe with even higher growth rates than this, and this is 30 times. So I think it's just people kind of flow out of situations like this. Would I consider investing more in Siena if they um continue selling off? Uh potentially, yeah, because because the scale across um opportunity is so large, um, and they are really a dominant company in it, but I think it's just Austin, it's just it's a company that was up substantial margins in 2026, and we're getting a relatively uh just it's it's extremely high expectations. They it can only grow so fast on the back end of it, we're getting a fierce uh market sell-off. So I don't think to any degree the thesis is broken. Now to the second part of this question, um the rate of change question is good. It's it's gonna come down. And this comes back to what I talked about earlier in the episode, which I I kind of wish AI was growing slower. Um, but I think over time within our portfolio, we are investing in infrastructure for this reason I've mentioned of uh compute broadly being realized as uh the industry that replaces what used to be software in some ways, right? The value has shifted. It historically was commute, compute was commodity, software was unique. We're we're seeing a little shift in that. I do think, you know, we wish we could have invested more in some things like software, although I'm glad we didn't because it hasn't played out well. But we are looking, you know, as AI matures, it should accrue value across different industries. Something like healthcare right now is deeply fascinating to me. I'm never gonna be an expert in it. I just can't be. There's just too much to know. Um, but I do have a lot of people I trust in that industry that I talk to. Unfortunately, we can't have them on the podcast because uh they can't do podcast or media work. But I would issue a challenge to our readers. If you have anyone that you think is a great expert in the um AI space along healthcare, biotechnology, I would love to hear from them. Um, you know, it Austin, I rapid breakthroughs don't always lead to stock returns. But what we've seen recently in um, you know, there's a recent pancreatic cancer conference that when when it finished, people got up and applauded. I've I've talked to people in the space who say, ah, you you never see that. It's actually very legitimate. You look at the GLP ones and and what they're starting to do, you know, it's it's no longer just weight loss, it's curing fatty liver, it's reductions in cancer. There's so many, it's it feels like a snowball rolling down a hill of um, you know, people asking, oh, you've got these data centers that are polluting my lives and taking my water and uh, you know, making my electricity more expensive. How are you taking my job? How are you making life better for me? And it's like, well, there actually is a real endpoint for that in a new industry. And um, it's gonna feel, in my opinion, very real very soon. And uh I I wish we could get some more investments in that. It's just one example. So I I did want to bring that up. Um, the rate of change is going to go down. Um fortunately, many of these stocks are priced for the rate of change to dramatically go down. Um you know, some things the rate of change may go up, like a Broadcom for reasons we talked about today. But I do think as it goes, we will continue to diversify the portfolio. I don't expect us to permanently be invested as concentrated as in some like semiconductors as we are today.

SPEAKER_00 1:03:35

Um really great answer. And I'd love to just chime in on the second part here, um, Henrik, which is the portfolio in the podcast is the AI investor, um, not the AI infrastructure investor. So while you've put a lot of money into the infrastructure buildup, because that's just part of where the cycle is now. And even if that rate of change decelerates, which it absolutely will, I mean, we're now at the point of some of these hyperscalers doing 200 billion a year, um, you know, and they're up 10x in the last couple of years. So you can't 10x 200 billion. There's just not enough money, they can't take on enough debt. So the rate of change on the CapEx side will decrease, but this is not strictly um the AI infrastructure portfolio, it's the AI portfolio. And as these technologies develop, the nature of the opportunities will change and present themselves. So if you think about the mobile phone, right? If you wanted to invest in the mobile phone revolution, you know, maybe early on in some of the iPhone releases, you wanted to invest in the suppliers. You might have invested in corning, which ironically is now an AI stock, or you might have been investing in some of the chips that were inside, you know, the iPhone 4, iPhone 5, iPhone 6, whatever the case may be. But then as that industry matured, you had completely different investment opportunities that were not even possible beforehand. You had Uber, right? You had apps, you had DoorDash. That's still investing in the mobile trend. It's just in a different way. And as the technology matures, we are going to be presented with new opportunities. You know, I think someone you're going to be speaking on stage, I think, with our former boss, David Gardner, in a couple of weeks. And he is someone who has invested in e-commerce in this way. As the nature of e-commerce has changed, he made additional recommendations and as new companies and new opportunities presented themselves in ways that, you know, I think his first e-commerce recommendation was eBay, because that's all that existed at the time. And then as he was investing over those many years since in new companies and opportunities presented, he was there early as an optimistic investor, as a rule breaker investor. And damn, has he done well? Um, someone we both have learned a lot from. So, Henrik, I don't one small note on that too.

SPEAKER_02 1:05:33

You look at where the cycle is, and so much of it has been am I going to get to this narrow line with laser company before everyone else? That's that's not normally how things accrue wealth over the long term. It is getting to the trend first, getting an allocation to the most dominant companies and holding them. Historically, that's what's going to outperform. So I know everyone's in this window where it's like, oh, what's the bandwidth stock? What's this? Uh, sorry, uh, not bandwidth. With one, what am I saying? Uh bottleneck. What's the bottleneck stock? That's that's normally not how wealth is generated over extended periods. So I realize we're captives of this moment, but what is going to generate money, it it might be one great healthcare call in two years might generate more money than any other call in the portfolio. And we'll always be amenable to that.

SPEAKER_00 1:06:27

Yeah. Uh I mean, it's we, as you said, we're captive to the moment, and these are the opportunities that present themselves. Boo-hoo. You know, what opportunities they've been. They've been fantastic. But over a long enough time scale, these companies, these opportunities might be the equivalent of Cisco and Comcast, right? In the internet build out, in that before there were, before there was Netflix as an investment opportunity, there were those companies, and you invested and you made money. And then, you know, new opportunities presented themselves. We're beating a dead horse here, but I I actually think that is a really important thing as listeners. The longer they stay with us on this show, they're going to have to witness some transitions like that as new opportunities present themselves. And your drum beat for the last few years has been bottlenecks and it's worked really well. And the temptation will be to always apply that playbook. But what worked then might not work in the future. And that's sort of forgive me, forgive me. That that's sort of how you end up like a Michael Burry looking for the next financial bubble for 20 years, right? Like, like trying to find like it worked really well one time, but that playbook doesn't always play forever. You need to be dynamic and accepting new opportunities.

SPEAKER_02 1:07:31

I didn't expect you to go for Michael Burry, but you did have that four ounce, uh, you did have that four-ounce cold brew before the show, Austin. That's turning the show.

SPEAKER_00 1:07:39

Made me a little spicy. Making enemies on the AI Investor Podcast. Um, Eric, last question from Danny Mack. Apology, apologies if you've already covered it. Will you be investing at the opening for the SpaceX IPO on Friday? Interested to know if you'll be making an immediate position or sitting on the sidelines for the time being. Eric, I will not be investing on IPO day personally, but that doesn't mean I'm not investing in SpaceX in the future. I don't see any benefit to needing to rush to the entrance on that day. Uh, the valuation here is enormous. The opportunity is even enormous. It is the size of the universe, so to speak. Um, incredible technology, reason rockets. Starlink is, I think, truly one of the most impressive products of the last few years. The float on this is also very low. And there's definitely some market dynamic aspects to how this company is going public that sort of is like no precedence. The size of it, the size of the float, the pent-up demand, a bottleneck, if you will, as investors are rushing to get into this. I'm not investing on day one. I'm probably not even investing on month one, but that is not a commentary on SpaceX, the company. It's amazing, it's fascinating. I would love to have a position in the future. I do not feel pressure to invest on IPO day. What about you?

SPEAKER_02 1:08:49

Yeah, and there's always when Cerebrus was IPO and we said, are we gonna feel pressure to invest in this? No. If if you get an IPO allocation, that's probably a good deal because it was going at 185, it was almost certainly going to pop. What happened? It it was priced at 185 and immediately popped on its first training day up to I'm looking at 386. But if you will have chased it and bought at those high levels, especially if you didn't have those initial allocations, well, it's at 235.80 as of close, as I'm looking at today. So I expect a similar thing to probably happen to SpaceX Austin. Um, there's some things about this IPO I I just don't like. Um you looked at a lot of rules were formed in 2000 around some of the divisions between what you call the buy side and the sell side on Wall Street. Um, some of the research that was put out and uh, you know, analysts essentially being directed to give good good research um that they didn't believe in. That's where uh you know Henry Blodgett, he got caught up in that. Um so it it just it seems like you you see some estimates from Goldman Sachs, and I don't know if these are officially published or part of the Road Show, but of uh revenue going up from uh, you know, numbers that seem almost impossible to achieve. Uh it just feels like it's it's kind of dressing it up for this. Um so I I would expect it to come back a little bit. It's a company that I deeply respect. I deeply, deeply respect it, but it also has problems. You know, they're they're they're gonna have a lot of compute rental um revenue coming, which is gonna make their growth rates look really strong. But you look at the terms of basically um XAI is only renting out because they have no use for the capacity themselves. Um, it's it's uh from a place of weakness.

SPEAKER_00 1:10:41

Uh their contracts definitely definitely snatching victory from the jaws of defeat there, just as an interesting. Has there ever been anybody who's been able to pivot as quickly or as impressively as Elon? I mean, he was going after, you know, Grok was supposed to be a frontier model, it's not, and then but he's gonna make you know money rain from the heavens because of all the commute he's been able compute he's been able to build in the meantime. Incredible, incredible, yeah.

SPEAKER_02 1:11:06

And I I think space compute, it's it's something perhaps in the future, but they're having to rush the narrative to to meet the IPO timeline. Um, you know, Austin, the the last thing I I would say on it, uh, it seems like do you remember Facebook?

SPEAKER_00 1:11:25

Uh their IPO. Oh, that is a perfect analogy for this. And yes, I do. Most hyped IPO, biggest on earth at the time, I believe. Yeah. And the skepticism was around whether or not they would be able to monetize mobile. And the shares traded down. I forget how much, but they had six months well under IPO price. It could be way longer than that. And I forget the magic. Yeah. I mean, I mean, it was an extended period. It was not a short, like they traded way down below IPO for a period on based on a skepticism around um being able to monetize mobile. And then, of course, you know, they they figured everything out and became one of the best investments of all time. That is a great analogy for this.

SPEAKER_02 1:12:02

I think it's gonna be pretty similar where um Facebook had a gap between promises and what they could deliver. And in that gap, they IPO'd at $38. They trade significantly lower. Um, I think SpaceX is a wonderful company that I deeply admire. I think they're uh rushing a little bit to IPO um on some promises that are gonna be difficult to deliver. So yeah, it's it's a company. Anyone who's listening, they can obviously the the whole point of this podcast is to make your own personal decisions, but I will be seeing this one out. And um I I do worry that the higher probability is they will have some um some difficult periods uh in between across the next year to two uh before they maybe find their footing.

SPEAKER_00 1:12:47

Fantastic analogy. I really like that one. Um, just for the people who want to keep us honest here, it looks like FaceX, uh, sorry, Meta or Facebook at the time traded below their IPO price for a little over a year. And then interestingly, man, they got down to $90 a share. Actually, fun fact, sad fact, they got down to $90 a share in 2022. I think I bottom ticked that dip within like three cents, and I have the receipts to prove it. Um, and then and then wrote it up like 450% and sold. Of course, you know, proving once again that you gotta just hold on to these things and not sell them because I sold too early. But I won't make that mistake. When we when we buy the SpaceX dip, when they when they get down below and I bottom tick again, I will I will hold on to that one. Impressive company receipts are being kept right now. Receipts are being kept. That's okay. Our listeners can keep us honest here. Um, I'll see if I can find a screenshot for my my meta bottom tick. Um, Eric, that was a marathon episode brought to you by whatever. Are you Starbucks Cold Brew? Stoke Cold Brew? Are you Stoke? We're kind of Rexy fans. People might not know that about us, so maybe we gotta go Stoke now.

SPEAKER_02 1:13:55

I I would do Stoke. I'm I'm originally from Seattle, but I think I'll make the transition to Stoke. We'll we'll do Stoke.

SPEAKER_00 1:14:01

Seattle's Seattle's losing Starbucks. They went to Nashville. So in in anger, you have to switch brands now. You gotta switch loyalty. Uh lashing out. There we go. This is all about us lashing out at our hometowns in this episode. That's all this is. We're just airing the grievances.

SPEAKER_02 1:14:15

I got a lot of problems with you people.

SPEAKER_00 1:14:17

Eric, thank you so much for the marathon episode. I look forward to the next one. I'll make sure to have a tall glass of Stoke poured for the next one. I expect you to do the same. Listeners, thanks for tuning in. And a final call for any of the champions who stayed with us through this entire episode. Please, please, please leave us comments, share this with anyone you think would benefit from it. We do not do ads on this show. The only thing we ask is for your comments and your feedback. Email us, comment on Spotify, comment on Facebook, rate us on your podcatcher of choice, and we will see you next time. Thank you so much. The AI Investor Podcast is for educational purposes only and should not be considered investment advice.