The AI Bull Case Just Got Stronger… But We’re Changing What We Buy

What the hell is happening with AI infrastructs? They've gone basically nowhere this summer, and yet the fundamentals are way better than they were at all-time highs in June. So, is this a pause before the next leg higher, or is the best money in AI actually starting to move elsewhere? What's up, everybody? It's LG here and welcome to Milgrowth Stocks, the daily market show that swears we're going to sell on the next pump. Just give us one more big rip, please, I beg you. Today is September 2nd, 2026, recording on the first. Every week we sit down with our top AI analysts to ask them the big questions about the market. This week we're talking about everything stalling, if macro is playing a bigger factor than we think, and whether the frontier models are actually being disrupted or if their revenue is going to lead the next leg. And we actually recorded an extra segment at the end called Milkroad Pro After Hours where the guys spent 10 minutes debating their upcoming moves and looking at their portfolios. You can see that part and the moves they're actually going to make in Milk Road Pro at the link below for just a dollar. Guys, welcome back. Uh, great to have the squad together and keen to know what's going on because I feel like we're getting a little bit tired of just chopping sideways through this infrade, but I want to hear you guys how you're actually feeling about the market as our analyst. So, Melvin, let's start with you man. >> Yeah, I think uh semiconductor trade have been moving sideways for about a couple weeks now, for about a month, I would say. But I don't think the AI thesis is broken. Um AI demand remains extremely strong. We saw that with Nvidia earnings revenue. Uh their expected growth for next year is about 70%. Um hyperscalers are spending billions and billions of dollars. We all know that. I think that is the problem with the AI trade right now because much of that is you know largely priced in. Um but then you have the interest rates um uncertainty that's also holding the trade back. uh the 10-year Treasury yield is around 4.75% while the 30-year is about 5%. Basically, what happens is when yields rise, investors, you know, are willing to uh uh willing less willing to pay higher for semiconductor stocks that may not arrive for several years. um that can that they can earn a stronger you know yield a return from the safe government bonds and that's also been you know having some problems in the market as well and then other issue that we often don't talk about is the nonI related economy is also showing some sign of weakness um July retail sales declined for the first time in nine months while Walmart uh reported slowest sales uh growth in the last six years as well um So that may you know have some problem in in the macro um uh macro world as well. So other than that I remain extremely bullish. Fundamentals are fundamentals are greater than ever before. We saw token usage actually jump from 57 trillion trillion token usage four weeks ago to about 113 trillion about now. So the use of token usage have jumped massively just in the last four weeks. And this only validates our entire AI thesis. So I remain bullish on the on the trade. >> Wow. Vincent, how do you feel? >> Uh I think the core issue is terminal value in the AI infrastructure trade at the moment. So basically the earnings outlook investors are struggling with the longevity you can call it of the of of of the the earnings growth of AI infra companies right but I think this week there was some great data points especially from Nvidia with their 70% growth number much higher than what the market um expected you had Samsung coming out I think it was yesterday telling us that 70% of our revenue is locked in until 2031 uh which is something we've never seen before in the memory cycle. So these are all data points showing you that this market is is is is going longer than what investors are modeling at the moment. Uh I don't think that the AI infiltrate will will go down from here. I I also don't think it will just like boost from here or do kind of a V-shaped recovery, right? Um, I think this will last. I think you will have to, it's really about stock selection here. I think some stocks that that prove their earnings are lasting will perform well like memory for instance here. Um, but others will just go sideways. So, it's a it's a tough market environment for sure. >> K, how about you, man? >> It's kind of playing out as I've been saying for many weeks on here, which we were going to chop through the summer, um, which we have done, and it frustrates investors and then they start to question, is their thesis right or wrong? Uh, and I think the only thing that's happened in the last few weeks while we've chopped is that the thesis has gotten stronger. That's very clear. Um, I would say the good news is that while the thesis has gotten stronger in price hasn't really moved, we've now seen the potential for rates to go up to 60. What is it? Let me hear 65, no 60%. Uh, chance of rate hike, 61%, sorry. Chance of rate hike now in September. Oil is now sitting at 90 bucks. And yet the price of these stocks have not moved. they've not made lower lows. Um things have not gotten worse even though the macro backdrop is um expected to be much worse. That's actually a really good sign because if you've been holding Nibius or Bloom Energy or you know whatever you know Micron whatever you you've chosen we're actually up from those lows that it put in up a lot from the lows that originally put in last month. And yet the macro backdrop is perceived to be way worse. Whether that's consumer spending like Melvin mentioned, uh rates, etc. My guess is that rates don't end up getting hiked. Uh or if they do, it's one time and then we end up going the other way. It's kind of more of a fake out from the Fed to kind of, you know, show that he's not just following what Trump is saying. Um whether he actually does that or not, I don't know. But I don't think rates is going to be an issue. Um I do think oil will eventually continue to come down. Again, it's been choppy. It's looking like it's going to break out upwards. Um, which is obviously problematic, but again, the stocks have not put in worse lows as a result of that. And so, I think we're coiling. Um, and we're just going to have to wait through it. How long it takes, no idea. But you should use any uh dips in good companies that are part of this AI infrade uh or elsewhere to uh to buy and and um and then just wait. I think they're going to do well over time. Or you can go into other areas that we've been talking about like SAS, which has dominated throughout this entire time. I've been screaming that, but every time I do a post on SAS, no one cared. Like, no one cared whatsoever. I was like, "All right, fine. It's just me by myself." Meanwhile, Salesforce is up 65% since I bought it uh two months ago. Service Now is up 50% since I bought it two months ago. Uh Vincent's always trying to talk about crypto. Melvin's like, "No, no, don't even talk about this podcast. Up 30% in the last week." So there are other things you can buy outside of the infrastructure side which is why I like to say diversified but I know many of the listening here are on the infra side. I just think we got to give that a little bit of time. >> Just going to pause there for a second to point out that the market is showing signs of something kind of different happening and our analysts at Milkro Pro are all over it. They spent the last couple weeks making a lot of trades, getting out of some positions and then getting into a lot of new ones getting ready for the next wave of robotics space or even kind of picking some different AI winners. If you want to see what they have in their portfolios, what positions they're opening, it's just a dollar in Milkroad Pro at the link below. Maybe SAS is too broading, Kyle. Maybe it's too many things. There's too many SAS companies that are dead and will never come back, right? Whereas NeoCloud is very specific. So where SAS it's like who are the three companies you think about with SAS like Microsoft? Like it's not SAS like it is. But >> if you listen to Melvin show, he actually says most NeoClouds are going to go bankrupt and there's only a few that are going to make it. So I've been listening Melvin. I've been listening. [laughter] LG, you're the one who interviews me all the time. What do you mean? >> Yeah. What I'm saying is that I think I think these other category leaders have two or three names that come to mind, right? Even memory is the same thing where it's like, well, here's the three memory companies that matter and maybe there's a spattering of like dark horses you can bet on. But the rest, it's like that's the thing. Neocloud, same thing. Um, energy, you know, we talk about Bloom, there's a handful, whatever. What I'm saying is that for SAS it's like sales for you know what is what is the SAS company that makes maybe maybe that's what's going to happen though Kyle is you'll have a handful of companies that become the SAS leaders. Yeah. Anyway, memory there actually is only three companies, so it's that's a little different story. But anyway, I have a question. Vincent, one of your strong uh takes that you had, I don't know, a couple months ago was that the tokens will get commoditized, the models will get commoditized. Uh and as a result, this is going to hurt the frontier models. I also agreed with this at the time. Uh and and it has been true. The the tokens have gotten commoditized and now they've come much lower, right? If you look at the the prices of of the tokens of like OpenAI and and Enthropic, they're actually bringing their tokens down. So that was somewhat that was correct. What was not correct from that was that the AI frontier models would get cooked and their revenues would be impacted by this. Their revenues are actually accelerating like crazy. If you've seen Codeex, uh their amount of usage is just going through the roof. They've gone from like six million Codex users to like 25 in the last two months. Uh Grockpot Grockpot Grockbot is accelerating in its usage. Uh Antropic is growing right now. If you've seen their ARs, um all of them are growing and doing really well. Does this change your thesis? Uh is this bullish bearish? Like what how does this make you rethink what your what your original thesis was? Yeah. So I think we need a bit more nuance here. My view was always that it will be commoditized in the sense of the majority of tokens that will be used by you and I but companies as well will be from open-source models but the tokens that create the most economic value will always go to the frontier models. I think that was kind of the more nuance view though. Um I was always pushing the uh you're not completely off right. Um and I still hold that view. Um I just think it takes a much longer time until uh this teases plays out. Um because the models need to get better and then the open source models also need to get catch up and it it takes time, right? Um what I think though the the more important point here is it doesn't really matter, right? because actually it's really bullish that uh for the broader AI infiltrate that entropic open eyes of frontier models are showing those bottomline numbers. Actually, there's some great data points from the Dylan Patel with Drokes Patel interview um this week where he's saying that uh Entropic can generate 50 million revenue per megawatt against a cost basis of between 10 to 15 and eventually 25 to to 40 million um of that. And what is the what's what's the important part here is that remember when the when open AI came out uh saying that we only have 18% quarter over quarter growth and everybody was asking okay if you're not growing uh or or if you're only growing that much who is paying for the buildout well if entropic and eventually openai is able to get more or basically a positive margin out of each megawatt then each compute pays for the next compute, right? So, I think this is the the the the more bullish new ones here. And I don't I I never really cared if the Frontier Labs will win. I just care um that the AI trade the AI infrade will win. And it helps that they're not losing. >> For those uh that didn't see, so Enthropic was only 18% quarter over quarter, but then if you look at the month of August, or was it the month of July, I guess, um it just went absolutely through the roof and they grew 20% month over month. Uh so it really started to pick up. Enthropic was profitable in Q2 and actually in that podcast Dylan um Patel mentioned that he thinks OpenAI will be profitable in Q3. Um so the tides are really turning right. And then also uh I think it was yesterday OpenAI has just reached a billion AR on their um annualized revenue on their ad business too. So now it's not even just subscription API fees. They're starting to monetize in different ways, which again is super super bullish for the infrastructure trade because they're finding more ways to monetize outside of just usage. So yeah, I think again looks really great. Melvin, what were you thinking? >> Yeah, I think that Dylan Patel's podcast was really really eye opening. Um because what what Vincent mentioned like Win Vincent mentioned that they're operating one mega uh megawatt of data capacity costed them around 10 to 15 million. Now all these uh labs can generate up to 50 million in revenue from that same capacity. Um this is going to allow them to fund you know their own growth in you know instead of largely depending on venture capital IPOs and whatnot. Um this may be the reason why they're delaying IPOs. I'm not sure when those IPOs going to come out but hopefully anthropic maybe this year. Um and what he also said is that if you look at like new hardware, you know, like Nvidia Blackwell GPU and custom AI chips, um they can deliver up to five times more performance per watt. Uh this means that controlling a large share of new power capacity gives the labs um an even larger share of actual computing performance. And he also believes that this is this is so interesting that compute could reach to 70 to 80% by 2028. Um and potentially given them control over the most usable computing power. He believes anthropic and open AI will um gain that um and the biggest like I think the compute is the biggest moat right now. I think I think whoever controls that wins this AI race and open AI and anthropic is in the forefront of that and SpaceX as well. I I think they're they're they're in a huge huge battle as of right now. >> Yeah. I think the biggest risk in all of this is that OpenAI and Entropic because they're getting profitable, they need to lean less on external capital, right? Um and also Dylan is is is said that in the in the pod, next year they're projected to buy up to 40 to 50% of the new compute that is coming online. Right? So basically half of the new compute that is being built and energized next year is going to open and entropic. What happens if the growth of openi or entropic were to slow down for instance because regulation slows them down needs to check their models before they get uh published. that I think is a major risk that people uh need to understand and that's why I remain bullish on the kind of open-source argument. We need open-source um um compute demand to diversify the the the customer base, but it's getting tough when the frontier models get um um margin positive. What's interesting is the reason they're going to get 50 up to 70% of the compute as you said Melvin and Vincent was because they'll pay more for it because they can make more money on it, right? So what you can expect if that happens is the margins of all the the compute sellers should go way up, right? Like right now SpaceX talked about selling a gigawatt for was it 30 to 50 billion which like most are not doing that, right? Like most Neil clouds are like I don't know 10 billion um and starting to get a little bit higher. But I think what SpaceX is saying is well I'm pretty sure where this is going is we'll always be able to do 50 30 to 50 and in that world that means the infra layer needs to go much higher right because I don't think the market is fully pricing in 30 to 50 gawatt uh uh 3050 billion per gigawatt right uh it means the Neoclouds need to go higher it means bloom energy needs to go higher it means the data center companies need to go higher but like you're saying at some point we might start to price that in and then if they start to stall out and we got to go back to the open source. They're probably not paying 50 billion per gigawatt. It'll still sell, but it's going to be 20 or 10 or I don't know what the number will be. And that's when the the margins are going to have to come back down. >> Yeah. But the the bullish argument here is that Frontier Labs will always reinvest their free cash flow into training, right? because they need the next best model in order to remain frontier to have the gap um to to the open source models and to competition as well. And so this should keep demand alive and avoid this slowdown, right? So the only real issue I see here and and it's a real issue we saw it right um uh is is is regulation is the White House saying we need to control this. we need to check um the the the models before they come out which they did with Fable 5 and also OpenAI's model. Um yeah, that's it's the biggest but you're totally right. Um and we also saw it in the Nvidia earnings, right? The memory inflation part hit them so hard that the projected margins from 75% down to 72%. Um, so I think it's it's it's not all of the AI infra trades or AI infra companies will benefit from it, but the the the real pions picks and shovel trades like a memory like uh Bloom Manergy for instance will definitely have pricing power from this. >> Yeah, I feel like there's definitely a more bare cases against um you know these frontier models than bold cases that you know that we see time and time again. Like for example, do you guys see a world where hyperscalers like Google, Amazon and you know Microsoft um you know these companies may not want to remain simple infrastructure providers while like you know outside labs like open AI and anthropic capture most of the value. Do you think there's a world where they could prioritize its own services like we see SpaceX sign that clause in their in their in the contract they signed with Anthropic. Um can they limit like would that be a bare case? Do you guys see down the future the compute gets so constrained that people are starting to back out of deals to um you know further you know um prioritize their own >> but is that why is that a bare I guess you're saying it's a bare case for openthropic because they can't get the compute which means they can't make any more revenue. Yeah I see what you're saying there. Um, it's a bare case for them, but that's still super bullish for the rest of the infrastructure build out, right? Because either way, >> the compute's getting used. Uh, and I think that's actually one of my most that that's why I like SpaceX so much is the optionality, right? It's like they can either rent out that compute or they can use that compute. And I love that. Uh, Meta, I don't I haven't bought them yet my portfolio, but I do think that they're kind of building towards that same thing. Um, and so I think the the optionality play is is super interesting to me for the comput side. >> I don't see that happening Melvin to be honest because what we were saying earlier is that Entropic and Openi both go uh profitable this year, right? Entropic may be profitable already. >> Of course, >> if that's true, then they are able to pay the most for incremental new compute that comes to the market, right? they they just outbid everybody else in terms of how much they pay per megawatt because they know they make so much revenue revenue out of it that um yeah which basically just consolidate the market. Um, so I don't think that's a real uh argument there unless their regulation or any other use case come or any other issue comes out with them not further increasing revenue. Um, and that would slow the entire thing down. But that would not only be bearish for Frontier Labs themselves, but also the broader infra. >> Where are you guys looking outside of the infiltrate? Like obviously the infiltrate is has been one thing I would say about the infiltrate is like it's still very much crowded. Uh there's still a lot of people in it. Um and yes it's down you know a lot of these these apps are down 30% from alltime highs, 40% whatever. But you can see that retail still is is in it. They've been buying the dip on it. They're still talking about it. When we post about memory on X still our best performing content which means retail still is still there and they're still they're still loving it. Um, and so I I do think it still has its next leg higher like like you said, Vincent. I don't think it's going, you know, hundreds or thousands of percent like it did over the last year, but I do think it it does rerate higher once again. Um, but, you know, I do think it's important to have other things outside of the infrastructure in your portfolio. I'm curious, are you guys buying into that or looking into any of that? So I think you need to be a bit more nuanced because the infiltrate has several layers to it and I and and my view is that there are certain parts of the AI infiltrate that will shine in the future right beyond memory stuff like power semiconductors stuff like materials that benefits from 800 volt DC and the new where Ruben and and and the chip architectures beyond. So I think there's still there's still alpha to be made in this trade while at the same time I agree the broader AI info trade I think will outperform the S&P but not provide those crazy returns. I think and and I I highlight this highlighted this so many times, the the conversion of AI agents and digital assets is a really hot topic now because not only not only because crypto starts to performing well, which in my view has nothing to do with AI agents just yet, but with the broader tokenization uh narrative, but this is something really important uh to look at. My deep dive this week will also be on on on figure technologies also a way to play or express this this kind of intersection and then again AI adoption um I think in pharma and finance is is really interesting still >> yeah where I'm looking at is uh robotics you know we saw like over the last like two weeks they had a whole um robotics Olympics or whatever it's called we saw like like I don't know if you guys seen the videos how crazy it was. We have robots that are beating um Usain Bull's um you know um record and there's so many other you know so many other things that they've demonstrated and and I think we I don't know if it will happen this year but into 2027 2028 I think we're going to have a huge huge robotics run up. um a lot of the robot robotic plays are kind of priced in but as we get closer to you know them getting deployed all around the world I think we'll have a huge moment and I'm like starting to look into that a lot more now and I think that is the next you know uh wave >> I know there are people listening to this right now who work at AI companies so this one is specifically for you if you're trying to get your company in front of people who are completely obsessed with AI and actually investing in this stuff well you found them Milk Road reaches more than 500,000 investors across our newsletters, podcast, YouTube, and X. And Q4 is filling up. So, don't be the brand that waited too long. Go to milkroad.com/sponsor. >> I do agree with that. You're about to see like robo taxis is really starting to c like catch on. Um whether it's, you know, kind of Whimo, but their growth has been kind of minimal. Um but you're hearing a lot about it with Uber, you know, are they are they going to be disrupted or are they going to be the leader because Robbo taxis and then obviously Tesla's really pushing forward here and they're having their cyber cab launch on September 3rd, so this Thursday. So, I do think that's going to spike a lot of interest as they launch, I'm assuming they're going to launch their cyber cab uh on Thursday within Texas. Um, and so I think that's going to make some noise and that's going to force people to start looking. And then also they've got um Optimus which their next version is supposed to be released in the next I don't know one or two months. Um, according to JCAL from all in podcast, it's on another level versus what you saw at those robots Olympics too. So, um, you know, I think right now the robotics trade still is a little quiet because most of that's happening in China, but once Tesla launches Cyber Cab and then Optimus, I think, you know, the US is going to start to get real loud about it. Um, we actually just did a um robot podcast, when was it LG earlier this week with um with Robbo Strategy? >> That's right. It came out just two days ago for when you guys are listening to this with Scott Walter who's kind of like their their their head of research. This guy is honestly just a quick plug for that show. First of all, this guy's hilarious and super fun to talk to. Uh the kind of guy who literally showed me a robot that you swallow [laughter] and like analyzes your gut like magic school bus style. Um and it definitely to me like that's our second podcast that we've done with an external guest in this past month who's a robot robotics expert. Uh this one doing incredibly well. So you know people listening, thank you for that. But it's definitely something I'm glad you brought this up, Melvin, that it's like I I think I think as things slow down, it's like people will just kind of want to hear about what else is going on while you're waiting for all this stuff to play out, right, >> Kyle? Do you think the Tesla event on Thursday will be important or will be news? I >> I don't think it's going to be significant to their revenue, but I do think it's important just in the sense that they're going to launch Cyber Cabs in Texas. To what scale? I don't know. They definitely have a lot of cars that are out there. They've registered a bunch over the last like week. they've expanded their uh areas that like their geo areas that they can um they can work in. So, I would imagine they're going to put hundreds in Austin and I think it's a stepping stone that you need, right? They're going to have to prove that this works, that it's way cheaper, that um it's way safer. And if they can do that in Austin and expand it onto highways and get these cyber cabs running from Austin to Dallas to, you know, whatever, Houston, um I think that's going to be a a really cool kind of feat for them. If they can really just get cyber cabs everywhere in Texas, then they can go and just do that. Okay, now I'm gonna do that in California, Nevada, and Florida, etc. Um, so my guess is they're just about to abs and you know Elon even retweeted someone said they're going to flood um cyber cabs in Austin and Elon just responded and said yes. So I think it's a good next step but do I think it's going to be relevant to their revenues? No, not at all. >> Yeah. Speaking of Elon, he came out this week saying that in 2027 we're short of 15 gawatt of energy and because of that 15 gawatt of AI compute can may not be able to to turn on that year and and even more importantly SpaceX uh is building its own facility to to build turbine blades uh because that's the actual shortage in the market uh because there are only three companies out there building those those those gas turbines in the end, right? Um and I think while the AI infiltrate is slowing down, energy, which is in the essence the core bottleneck of it all, maybe even bigger than than memory, right, is still something really important to look at. Also beyond bloom energy, right? I think Bloom we covered a lot, but there there are other names. Wisteristra is something that you can look at that is not up a lot actually down this year I think 6%. Um you can look at liquid cooling also in that sense some battery names. So I think there's alpha to be made. >> Elon actually came out and said the same thing this morning. He actually had a um interview with G or G20 summit and he further said 15 gawatt they're all going to be short and that's why they see you know he was saying that that's why Google Anthropic and are signing massive deals with them. Uh but yeah, this is going to be a reoccurring thing um through like going into next year. Yeah, >> just buy SpaceX and Tesla and let Elon figure out the whole bottleneck situation. >> Literally, this guy's going to hold this 100%. Uh the other thing I think you look at which takes a lot of this out is the application layer. This is why I got so bullish on on SAS because you know um certain SAS companies like like Salesforce for example, we'll just use them. they were down to like a 8p or something ridiculous like that, right? Meanwhile, they're growing like 10 to 20% year-over-year. The business is still booming and still doing well and yet they're priced to like absolutely no growth. And the interesting thing about when you look at the application layer is even if we have these bottlenecks and we can't get GPUs energized or you know whatever happens on that side doesn't actually matter because we have enough GPUs today to provide the tokens that these companies need to reinvent their businesses right and so the tokens are already cheap enough from them and only going cheaper and so whether the models get commoditized and it's an issue for open AAI or cloud or whatever doesn't matter that's actually a great thing for the application If you know we're short a few GPUs and that really impacts like I don't know Nibius or OpenAI or Microsoft or whoever couldn't get those data centers online that's okay because there's still enough that something like Salesforce or other companies that are using Salesforce can still get the tokens that they need. So they have a lot less of risk and especially when they were priced down at like 8p um which they were in was it June or July whatever it was now um and yet when you look at their earnings so Salesforce earnings was it last week I think uh yeah last Wednesday um what did they show that their business is actually accelerating and booming as a result of AI. uh for two reasons. One, we are not going to go build all of our own software just because we have the ability to uh especially very important things like uh a CRM or a database. These are one difficult things to build, but two if they already exist, why go and try to rebuild it? This becomes like the source of truth for your company and both humans and agents can go and use it and agents actually become super users of Salesforce or of CRM of databases, etc. So, um it's actually a perfect product for agents. Meanwhile, everyone thought agents would just go rebuild it. They don't. They just love using it. And then Salesforce also has Slack, which is a place where anyone who works in a company using AI. Um, it's the place where humans and agents work together and they communicate with each other. So, Salesforce ends up having two incredible products for this agent revolution. And, you know, it doesn't really have the risk of these bottlenecks. Uh, and so to me, it was kind of an easy an easy buy. Um because you know the downside there was basically no downside left cuz they also have billions of dollars where they can buy back their own stock. Um and there was potential that hey this is actually going to work out and you know what we saw was the seats which everyone thought was going to uh shrink. It actually grew is accelerating now. Um and uh and so they're just their business is actually booming. Same with Service Now. Palunteer had a killer killer earnings a few weeks ago. Uh all the same. So you're seeing some of the software companies um that were just beaten down um they don't have this risk that we're talking about the bottlenecks um but they have the benefit of just everyone onboarding and using AI. So I think that was um a huge opportunity and and I think the market's starting to understand it now like Salesforce is up 65% from the bottom. Um but I think it's still got some some legs left in it. One question on this. What do you think from here onward maybe until end of year and third out is the more attractive trade? Buying into those SAS names or AI adoption names maybe right that are up um a lot over the last past few months and weeks or AI infiltrate that is consolidating at the moment. >> This is this is what keeps me up at night trying to figure this out. So if you asked me a month ago, the answer was easy. uh or two months ago the answer was easy, right? Uh now it's kind of like, you know, the the ones that were overvalued have come down, the ones that were undervalued have come up and it's like they're now meeting in kind of the middle. Now I don't know, which is why I just say it's the everything bull market. Buy it all. [laughter] >> Yeah. Uh or go look in crypto because there's a lot of undervalued gems over there. >> I think this everything bull market argument is not that bad. I think it's really about portfolio construction these days, right? We want to make sure you have exposure to not only the AI infiltrate but to other things as well. At least that's that's my view, right? Some some in the AI infiltrate memory etc. But some in AI adoption, some in SAS maybe uh like I do for instance with Palunteer. Um but then also some in in in this whole digital asset tokenization Melvin uh story. [laughter] >> Well, because the thing is too is like these things move really really fast, right? like if you weren't in biotech then you missed a massive trade because of the news that happened with Madna right um and these are the things and then if you weren't in crypto then you missed a big move in everything in crypto you know 30% in a week um it's still early for that so like I'm not overly concerned but these things move quick and especially if it's because of AI like reinventing something or like revolutionizing something like um mRNA uh vaccines and cancer treatment then the things just move and so you want to be allocated in some way some sort of exposure um because these moves happen and they happen really fast >> and so much has to do with sentiment. These things are really moving in waves, right? And you you just don't know which will be the next wave. The AI infrade wave will definitely come at at at this point in in in the year again, right? Maybe in September when we don't see a rate hike or maybe with the next uh earnings where Nvidia comes out once again proving we're growing much more than than what the market expects like like this like they did this quarter. You just don't know. And that's why I like the kind of more diversified approach. >> I feel like the market needs anthropic and or open AI to IPO so that they can start to like hear their thoughts on their revenues and get some guidance and all that kind of stuff cuz that's like the big bugaboo that's spooking the market at the moment. You know what I mean? Uh, and so we just need one of them or like maybe SpaceX will start talking about Groc um because they didn't really give it much love in the last earnings um because it honestly wasn't very important product but maybe now it is. But the market really needs to know what's going on with those models to make some decisions on where and for goes next. >> Can I ask one last question specific to memory? Samsung came out this week telling us that hey 70% of of our total capacity sold out until 2031 with long-term agreements, right? what the market or what the bare case on memory is is that it's the peak of the cycle and we're going now into the trough because we we're overbuilding capacity, right? The the Samsung news of this week, but also Micron signing 16 different long-term agreements is showing us that the market structure in memory is completely different versus pre- AI agents, right? So why is the market not rewarding that? Is the sentiment just so bad at the moment around AI infrastructure that investors do not care or is there something I'm missing here? >> I think Koreans got to leverage up. I think that's what that's [laughter] going to move the market. No, really? No, I'm I'm being serious, unfortunately. Uh but no, I I do think that like I think memory is sort of becoming like that Nvidia trade and that's the main problem and I and I I I don't think like memory is such a huge part of this and memory is becoming like the strategic infrastructure trade like like big tech is treating HPM like we see that's the reason why we see um Samsung and other players locking in 70 70% of capacity right almost like power and land like Jensen provides or like Jensen said like [snorts] you cannot have a chip you know Nvidia chip any chip without memory and without it it's freaking useless. So at some point in time I think this memory chair will start to go up. I think we have micron earnings sometime this month. Um so I expect that to further prove that this trade is a lie >> here. Here's why we can't get back up to alltime highs. This is exactly what what Melvin just said. Uh but in a chart the the leverage is gone and that leverage right here, this big gap that you see is what is the gap in the in the price of these stocks right now. Uh so either we need more fundamental investors to come in and buy this stuff or Korea's got to lever back up and then we'll rip to alltime highs. Uh but I don't know if they're going to I think a lot of them lost a lot of money unfortunately. [laughter] >> Take out the loans, guys. It's not a bad bad idea. >> They'll be back. They'll be back. Um guys, I I want to wrap this up by asking your moves for the week, but that's actually something that we're we're going to be putting behind the curtain. So, for anybody still listening at this point, um and you want to know what these guys are actually going to buy or think about buying or their positions or how they're going to manage from here, uh you'll have to sign up for Milk Road Pro at the link below for just a dollar. Uh we're going to record an extra like 10 minutes just to talk about your specific moves uh that you want to be making and also answer a couple of Milk Road Pro member questions. So, thanks everybody for listening and again, if you want that that preview, if you want to know what they're doing, get into Pro at the link below. Thanks for listening to Milk Road. If you enjoyed the show, make sure you like and subscribe. [music] And if you're struggling to find winners in the market, that's exactly what Milkro Pro is built for. [music] Our analysts have called some of the biggest winners early. And Pro lets you see what they're buying next, every trade they make, and the research behind every position. [music] Check out Milk Road Pro at the link below. Everything you hear on Milkroad is forformational purposes only. These are our personal opinions, not financial advice, and we may own some of the investments we talk about. Always do your own research and make the decisions that are right for you.