This Could Trigger the Next Leg of the AI Bull Market

AI stocks are cooling off and those prices from a few months ago are looking pretty good right now. But our analysts are saying that this is a natural cool down before the next leg up. And Nvidia would probably agree with them as they just announced a fresh $500 billion raised to turn GPUs into a financable asset class. So maybe it's time to be greedy as all this gets revised way higher. What's up everybody? It's LG Ducat here and welcome to Milk Road AI, the daily AI show that's missing one part of the quartet and we hope he didn't land in Kansas. Today is August 11th, 2026. Every week we sit down with our top AI researchers to talk about the market and whether we are still in this very juicy bull phase. And today, despite missing Melbourne, we're diving into the Nvidia news and the implications that another 500 billion has on the short and long term. Plus, we're going to look at the potential anthropic IPO, how much each hyperscaler, memory company, and Neocloud is making for every dollar spent on models, and which hyperscaler is actually the best to bet on right now. If you want to see the actual moves that these guys are making, all of that is in Milk Road Pro. Our guys actually bought three different companies today, and it's only a dollar to see what those are. Check it out at the link below. And a reminder that our podcast today is free, and it wouldn't be possible without our partners at saver.money, the stablecoin payments platform built for Asia. Keep an ear out for more information about them later in the show. Boys, we're back. No Melvin this week. He's stuck literally in a tornado. Uh so there any all the Melvin fans, uh send your your hopes and prayers to him. Uh sounds like a scene straight out of Bill Paxton's uh twister. Anyways guys, [laughter] welcome back. >> This is second tornado in like a month. Chicago, I [laughter] guess. >> Melvin always has tornadoes, man. Anyways, uh that's what that's what living in the Midwest does to you in the summer. Uh guys, it's going to be the three of us today, but still a really u a packed episode because there's a lot that's happened since our last roll up last week. Uh let's hear how you feel about the markets today. Vincent, how are you? >> Yeah, I'm good. I'm uh on vacation actually in Spain. So, I'm uh little bit dressed differently than usual. Uh but the markets are definitely not on vacation. Uh it was once again a pretty Yeah, not rough but exciting week. I mean, we had the memory craze over the weekend, right? We had Yuken, I think one of the biggest X accounts on focusing on memory, basically turning bearish, selling all his memory, and then the whole kind of social media went crazy, the whole bearishness on X. And then, yeah, the CEO of Micron coming out yesterday, I think it was saying, "Oh, by the way, we can only service even less than half of what people want us to to to sell memory to." So, that's gone once again. Um and then uh we had Nvidia coming out I think yesterday it was right announcing 500 billion uh with a massive banks basically that uh yeah companies can now use to to buy GPUs. Uh so that was really bullish. So and then Entropic also coming out with its IPO. So a lot happening remaining bullish. Um, stocks wise it was more or less flat, right? Even though we had that crazy kind of noise on on next. >> Kyle, how you doing, man? >> Yeah, I remain uh I remain super bullish. I think earnings continue to crash uh while at the same time jobs remain flat. I think inflation will come in cool tomorrow. Uh I guess if you're listening to this, it's probably already Wednesday. I don't know when this comes out, but anyway, Wednesday's CPI numbers will will be cool. I think we're now at like a 50/50 chance that the Fed hikes rates in September, which that was much higher, you know, a week ago, two weeks ago. Um, so I feel like we're still sitting in this kind of goldilocks, you know, market regime right now in the macro side of things. Um, and and a lot of that I think is just because AI is really starting to make its way into the P&L of companies, uh, as we're seeing in these earnings here. And so I think, um, I think the bull market continues. Again, as I've been saying though, I'm very patient here in the summer. Probably we got to wait until September for things to really, you know, fireworks to come happen again. So, I think we're just still in this sort of um consolidation pattern uh at the moment, especially in the infrastructure side of of stocks. Um I mean, if you look at SAS and others like that, they've been ripping. Um but on the infra side, so memory, neoclouds, chips, whatever, it's not really going anywhere. Uh you know, they're still down 20 30%, some even more. Um and that's fine. I think they're just kind of chopping, but I think we will reach new highs. Um I don't know, probably by September, October. And um I think these are all great buying opportunities in my opinion. Everything looks good. Kyle's got to get into options. If Melvin was here, he'd be like, "Kyle, you got to do options, man." Because if you're you're predicting certain numbers a certain time, that's that's your kind of thing. Uh but speaking of numbers, guys, one of the big numbers that I'm curious about is this $500 billion deal uh announced uh by Nvidia, I think yesterday. Was that today? This week? Um, and I would love to I mean for people maybe who haven't heard about this uh maybe Kyle you can walk us through like what that is exactly because it's basically a deal with like every major uh fund manager, every major bank uh and something that I think a lot of people are really excited about. >> Yeah, they brought in some of the So Nvidia basically brought in some of the biggest credit firms in the world was like Blackstone and I think Black Rockck and what was the list of names? Trying to find it here. Um >> Apollo. >> Yeah. Brokefield >> Goldman. Exactly. And so, you know, think of like your biggest banks in the world essentially and funds in the world. And the idea here is that um they're basically mobilizing $500 billion worth of capital from these credit companies to um to help finance and fund the infrastructure buildout, the AI infrastructure buildout. So, um a lot of this obviously comes around GPUs, but it's everything, right? I don't think it's just actually you know what Vincent maybe you tell me I'm not sure is it just for the GPUs is it for the whole thing data centers power all that or is it just for the financing of GPUs do you know >> yeah so it's basically for everything >> to to to build the entire AI infrastructure not only the chips but you need to build the data centers on Nvidia chips and then the rest of >> yes so the the way Jensen he wrote a a a good blog yesterday and basically he said um what they're trying to do here is turn um AI factories is what he calls it which is basically a data center with GPUs in it uh but these these Nvidia AI factories they're trying to basically come up with a new asset class and um creating a more like straightforward fungeible way to finance these. So um the the problem in the market for especially smaller companies like the neoclouds like Nibius and Coriv etc who don't have balance sheets like the hyperscalers is it's kind of hard for them to finance these huge data center buildouts with all these GPUs you know it costs like I don't know 30 40 50 billion per gigawatt to build all this stuff out and then you have yearly costs of like energy and that hard for them to get financing around that. And so what Nvidia is trying to do here is basically create some more kind of standardization. You could think of this similar to like um airplanes. They're often leased out by by the the airline companies. Um and you know, however many years ago it was, they sort of created a a more standardized way of financing uh and leasing out these these um airplanes. Uh same with mortgages would be maybe the easier way for most of us to think about it. At one point, we created mortgages and said, "Okay, it's going to be a lot easier to get a mortgage on a house because this is a thing that's going to last for a long time. It's going to hold its value. It's fungeable. There's a bunch of people that would want to buy it. So, banks can be um a little more sure in their financing details and and and and standardize it." And so what Jensen kind of said was look these GPUs as long as they're Nvidia GPUs. Uh and the reason he says that is there is software around these GPUs that make it a little bit more funible and have resale value on secondary markets because you can take Nvidia GPUs and you can change the software around it so that it works for training or it works for inference or it works for other things. So um you know you have the Michael Burries of the world that say that GPUs die after two to three years because you get new GPUs that come out and make those obsolete. what Jensen Wong is saying and what he's trying to to to say to these credit firms is look sure we get new models that come out for GPUs and so maybe the best like these these three or four-year-old models aren't needed for like frontier training anymore but they can move to inference right >> and so Jensen Wong actually said I think that these things will last for 10 years meaning you can get value from them for 10 years and so this makes the asset a lot more um a better financing uh deal for a lot of these these credit companies and So, um, yeah, I guess Nvidia GPUs are becoming an asset class and they're already starting with $500 billion and Larry Frink was on u on CNBC yesterday and talked about how he thinks this is going to need to go to trillions in the coming years. So, um, it's um, it's pretty cool to to watch this play out, but this is super bullish for anyone who's looking to do financing inside of the AI side of things. 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 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. So, what does this what does this actually mean for the market then? So what is what this means that there's just more money committed from these funds into this asset class. So what is what's the market's reaction going to be or what's the the the side effect from this going to be? >> Yeah, I think we need to unpack this because the investment implications of this and who is going to benefit is is actually not that straightforward. What is straightforward I think is that the broader AI infrastructure trade is benefiting because what this essentially is is that kind of a sixth hyperscaler like balance sheet is entering the market right they're talking about 500 billion now not in 2026 or 27 but over time if you compare that 2026 cup of the five big hyper scalers is 750 billion combined right so that's a massive amount of money that will be spent on on Nvidia GPUs but Then obviously also power, memory, networking, all that kind of stuff. If you go one layer deeper on the infra side, you recognize that Brookfield is part of that kind of consortium. Now Brookfield um has a framework deal in place with Bloom Energy of about 25 billion. And so me personally, I think Bloom Energy is actually a really good trade based off this because when you're spending that much money, you need on data centers, you need a lot of power, right? And Brookfield has this framework in place so it's easy for them to access uh the the Bloom Energy uh boxes. So that's kind of number one and where I'm really bullish on. Number two is obviously Nvidia themsel because what they did with this is they lowered the cost of capital for people who want to want to uh buy and build out buy GPUs and build out data centers, right? Um because those companies can now access the capital externally from those lenders um and deploy the Nvidia GPUs, right? So they have a source of capital when they want to deploy chips. They have to use the Nvidia chips, but they do not have that source of capital if they want to deploy TPUs or the AMD chips or or whatever, right? So that's really bullish Nvidia because Jensen was basically making sure that people are using his chips when building data centers, right? What that also means is that it's another step of Jensen in terms of removing risks of hyperscalers kind of leading the market because all those hyperscalers have their own chips right Google has TPUs uh Amazon has tranium chips and so on um and now he made sure that there is a broader target group that he can target that are buying uh GPUs. Now the last point on Nvidia and then Kyla I'll let you jump in is they also brought on or they also loaded their balance sheet with some risk because they're backstopping the risk of the the the chips not being used or sitt sitting idle there not not being utilized right I think that the number is 25% of the value they're covering if that would be the case right that was what Kyle was talking about earlier um and that to me is the core reason why Nvidia did a pop um because they're loading that risk onto their their balance sheet. >> Yeah, the market still thinks that this is just circular financing and I don't think they're seeing the bigger picture of it and I think it's probably just because they don't yet understand the long-term value of GPUs, right? I think they just assume it's done at three to even 5 years. And I think as years go on, they realize that it's it's continuing to be more and more. I mean, even if you look at the I don't know if I have the chart handy, but there's a chart that came out um I forget who did this research, but the A100's um are, you know, up in 2026 in terms of their rental prices. Um which they should be down because these are six-year-old chips, but they're actually up, I think, like 25%. Um which is pretty wild. So, actually, I have the chart right here. Um so, you can see they're all going up uh over the year. Um which is quite interesting. So, that was kind of cool. But I think um what was really interesting about the about Nvidia and you know if you're just kind of investing in the infrastructure trade in general is about a week ago we had SpaceX's earnings that came out. They said they're looking to get to you know 8 to 10 gawatts uh I guess it's 5 to 10 gawatts in 2027 up from they'll be at two by the end of this year. So that's a a massive demand and the market expects is Morgan Stanley they expect that SpaceX will spend $160 billion next year to reach that. That's also another 70 billion this year. So about a week ago we did not have that in the numbers. And so if you look at the rest of the hyperscalers not counting SpaceX right we we are looking at basically for 2027 we're looking at about a trillion. You add SpaceX, we're now uh 1.16. Okay. Now, we have Nvidia adding another 500 billion on top of it. Now, of course, as you said, that's not all going to go in 2027. We don't know how fast that will get spent, but if you listen to Larry Frink, he was talking about how we need to raise this and deploy it fast, and we're going to need a lot more, which has been the trend for for for many years here. So, I I I don't doubt they're going to spend a large sum of that in 2027, if they can. if not by the end of 2026, they'll spend a bunch of it. So, um you know, I talk about a lot how the analyst has just been so wrong on on the the capex numbers. And you can look at on here on the right side, it looks at the forecast in January. They thought that the numbers for 2027 was going to be 500 billion. Already the estimates without including SpaceX are at 994 billion. Now, you include SpaceX, now you include Nvidia. And it's just it's wild. I I tweeted this out yesterday morning and I said I was talking about how SpaceX is kind of the new hyperscaler in town and I even said I was like I wonder who's going to be the next one. And literally like 7 hours later Nvidia's like oh it's going to be me and all the biggest credit firms in the world. And I was like holy this is crazy. So it's hard to not This is why I say you know in the beginning when we were just talking why I say I think we're just consolidating here but the infrastructure side is going to go much higher. Um because there's just so much demand when you look at the earnings of companies and how much they're using tokens right now. And then if you look at the amount of spend that's got to come from SpaceX trying to get to 8 gawatts to 10 gawatts next year and everything that all these hyperscalers are trying to do, there's just there's no way that we're not, you know, spending way more here. So I I just it's hard to not be bullish. >> How do you think about the the Neoclouds actually? Because I mean Melvin is not here. He's the hyper bull, right? Uh but yeah, I want to hear I don't even know if it's in your portfolio or not. Uh >> no, it's it's not in my portfolio. My one fear of them was one, so at least on Nibius's um case, it's the the multiple of it is huge. I think it's like a 65 PE on a company that has, you know, a lot more debt and capex spending than it does revenue. And look, it has the ability to grow that revenue of course, but um when you look at the way SpaceX is doing this, they can build these data centers way faster. They're charging like 30 to 50 billion per gigawatt whereas Nibius and Cory are charging like 9 billion per gigawatt. Like it's a whole different scale. Um so I just like I don't know there's just a lot of risk within those businesses. Now I do think this helps. This is actually very very helpful for them. So this makes me more bullish I would say. um because I think there's less risk because they're probably going to get better rates on anything they need to raise from this 500 billion that's coming through Nvidia. So I would say this is um this is actually very bullish for for Neoclouds and would make me a lot more interested. Michael Bur interestingly just put a big short on Nibius uh Monday. Yeah. Or last week. Now to be fair it was at like 211 bucks and it is down to like I don't know 90 or something like that. So, like he's he's in the green at the moment, but they've got their earnings tomorrow, Wednesday, and I think it's going to be lights out earnings. And um I would imagine NBS pops from it, but who knows? Let's see. But I do think the um the coming months here looks really strong for for NeoClouds. >> Isn't it the the opposite for good earnings these days? >> Great earnings means your your stock dumps the next day 10%. >> Yeah. Yeah. [laughter] >> Yeah. There is another chart that is interesting for those that have not seen. I'm just trying to find it here. >> Maybe while you're looking it, let me give my uh opinion on the Neoclouds because I agree that the 500 billion from Nvidia are really bullish because the biggest risk to me that NeoClouds had is kind of how they financed this build, right? They they used private credit, they used kind of short duration depth versus uh buying into something that is really long-term, right? and and and they got squeezed from from that kind of financing. Now they have access to a massive pool of capital that structures the financing according to the the life cycle or the the life duration of of the GPUs that they can buy, right? So it lowers the risk that they're adding on the balance sheets and the financing risk. So to me and and interestingly the broader infrastructure trade and including the neoclouds did not really pop off the news. So the market is not seeing this the way I'm seeing that or at least I think I do. Um so that makes me really yeah interested at least on the NEO clouds from here onwards. Yeah, I can't tell why. I mean, the market's super red today and I I I can't tell why the market is just continuing to fade this infrastructure buildout. I mean, it's not that they've continued to fade it. Obviously, these stocks have done extremely well over the last two years, so it's hard to say they're fading it. Um, but and that's why again, I just think we're in a bit of a consolidation phase before they start to rip higher, but we're obviously seeing the earnings are incredible. The fundamentals keep getting better with things like Nvidia, all the capex raising. Um, and so I just, you know, I just don't really see it. But here's a chart that I wanted to show which is quite interesting. Um I don't know if Vincent you saw this one but it breaks down if you were to spend uh a dollar into the the model. So OpenAI or Enthropic um where does that dollar go like to what companies? And um so it starts at if you spend a dollar first they need to be subsidized because they're not profitable companies. So you spend a dollar someone gives an extra 17.3 cents to add to that dollar to make it actually work. A large portion of it goes to hyperscalers. They end up making 30 cents per dollar 17, right? So that's a large chunk from the hyperscalers. Neoclads down here only make2. So um not a large chunk of the pie by any means. Now um if you look at uh or sorry that's um 2 cents, not 2 cents, right? No, that is 0.2 cents. Um if you look at um SpaceX, it would be about five times that because they they're charging a lot more. So they would actually have better returns than the average NeoClouds, but they're the only ones that are that are able to charge that. And then you come down here and you're looking at like memory gets 5.8 cents, chips get 1.1 cent, foundaries 2.8, uh commercial chips, so this is your Nvidia is getting, you know, 13. So just interesting to see that. But you can just see that Neoclouds, they have a lot of financing and a lot of risk in their business and they don't make a ton from the trade, at least not right now. Now, that could change and who knows, especially if GPUs last longer than everyone expects, then, you know, they're going to make more money over time, but um they they're definitely squeezed out a lot of the the profit here, but this is very bullish hyperscalers, I would say. I think this is a great point and and maybe you can leave the chart open to make a sorry to make a connection to to Google and and and the hyperscalers in general because the way I'm thinking about them is they or at least Google has free options in terms of deploying compute right they can deploy it to sell more cloud capacity where we can see um it's really profitable right and we also saw that uh I mean 82% year-on-year growth from from Google this quarter was really bullish. The second opportunity is they can deploy to Gemini to to make a better LLM model. And then the third option is to deploy it to internal products to make YouTube ads better, to make s Google search better, etc. Right? And the question to me is how should they deploy compute to make the market reward them or stop compress compressing their multiple right now? The obvious answer is people are saying, "Oh, they're spending so much money, the capex, and and they have negative free cash flow." I think the deeper thinking is that if they were to deploy more capital of that compute into the cloud only, which makes money today as we can see on that chart, right? the frontier models that Gemini is not making any money but but is consuming a lot of compute then probably they would do better because cloud the the cloud business would [laughter] would >> it's not it's not that it's not making any money they made a billion from Apple it's nothing in comparison right [laughter] >> but there's something but yes there are opportunities to make money from it but I I agree with you go on sir >> no that was basically I wanted to hear your view on that if if you agree that if if Google, for instance, would stop pursuing the LLM layer with Gemini and use all of its compute in the cloud business and would because the the the return is already visible in their balance sheet in their earnings report, right? >> As I just outlined, >> do you think they will perform better? >> The only one that's outlined their exact kind of share in between was SpaceX. So Elon on the call said 90% is going to go to reselling compute. 10% is going to go uh into Grock and Cursor. Um Meta didn't explain their breakdown. They just said, you know, we could sell more compute. We're deciding not to uh because we want to put internally. Google has obviously not said either. Um there has been that restructuring at DeepMine at Google. So, there's speculation that they're actually going to start putting even more of their resources into the Google Cloud side um rather than into uh into their AI side. I don't know. Um I don't know what they're going to do, but I think what I think is interesting is they have optionality, right? I think the key thing is you just want to own as much compute as possible because once you have that breakthrough and you go oh if we just put more compute behind this it will compress our you know ad cost or it will do this you know increase our our multiple here or increase our profits here then we can just go and shift it over right like I think that's the key thing is you want that optionality and that's the really hard thing because it's so hard to get your hands on compute at the moment and so as long as they can keep building this stuff out which is what makes me a little bit bullish on SpaceX because they have that optionality and they can build it faster than anyone. Um, I think that is that that is the key. So, I don't know what percentage should be where right now. Obviously, they need a big percent to go to Google Cloud because that thing is growing at just an accelerated rate and um and they need those revenues. So, why wouldn't you take advantage of that? >> I mean, Cinder on the call said they're prioritizing clearly their their LLM model, the Gemini basically, right? And everything comes next. And I think the market is not rewarding that because they proved cloud is growing and they should allocate the capital there because the returns are clear long-term. Where I completely agree with you is on that optionality layer which is also why I like Google the most out of the hyperscalers because yes Microsoft has a model layer as well if you can call that. [laughter] I mean for any of you guys I think Milkro we don't have it but me and my big company right we use the Microsoft AI products still to this date I mean it's not good to say the least [snorts] right [laughter] so so they're not competing on the on the model layer and Amazon they have no model layer um Meta yes but they did not anything there right >> so to me Google is is the most interest interesting one out of the pack because they have this option ality and can choose once it's clear where whether the model layer is going to win or or providing the most value or the cloud business or put deploying the compute to to to existing internet products. um they had they have the option to choose where to put the compute, right? >> Well, and I think Gemini, they'll continue to focus on Gemini because they can start to integrate that into um Android phones and make that their kind of like a Gentic model. Um and maybe they charge some sort of subscription on that or they take off of Google Pay like I don't know how they'll monetize it necessarily, but I do think they're going to end up using Gemini uh for that if not an aggregation of it. Meta I think is open sourcing their models, are they not? Isn't that what came out? >> Yeah, they did. >> Yeah. I think this week or last week they launched a new model open way. >> Yeah. I mean what what will be interesting to see is what's going on with the frontier models anthropic and open AAI. We have not seen their revenue numbers in a while. Uh the last we saw they were accelerating faster than any company has ever accelerated in history. Uh I think Enthropic went from what was it they started the year at their AR? >> N billion. >> Yeah. 9 billion to like something like 70 something billion about halfway through the year or a little over halfway through the year and expectations to go over to 100 billion. Um we had talked I think last week or two weeks ago that they might delay their IPO. It sounds like it's actually still coming in either September, October. So we're going to get some numbers eventually and as far as I've seen the OpenAI numbers are through the roof over the last like month or two. Um I guess they've had a ton of codec subscribers. um supposedly they're doing extremely well out of nowhere. So- which is I think counterintuitive to what most people think because we're having so much adoption of open source models. Everyone thinks that entropic and open I see it in my comments on Twitter every day that these companies are going to go bankrupt and they're screwed and it just I don't know. I'm waiting to see their numbers but it looks like my assumption is it's they're continuing to accelerate and grow. Um and that would be extremely extremely bullish for the market if true. So, we'll hopefully hopefully they actually IPO because I I just want to see those numbers. >> You know, one thing we've talked about a lot on this show is that crypto is quickly becoming a huge part of the global payments infrastructure. And nowhere is that more obvious than in Asia. But if you're actually running a remittance company or a payment business, you know that the hard part isn't moving the stable coins. It's dealing with local banking partners, compliance, liquidity, and all of the operational headaches that come with sending money into places like India and Southeast Asia. That's why today's partner is Saber. They give payment companies stable coinpowered infrastructure to collect and make payouts across Asia without having to build all of that complexity themselves. They've already processed more than $3 billion in transactions across 40 different countries. So, this isn't just a concept. These guys are actually doing this for real. If you're building payment infrastructure or expanding into Asia, make sure you check out Saber. Money. >> What is Sorry, can you recap? What is what's the latest that they said today that they would do like that that there's hints that they're going to IPO or that they're just kind of frothing up investors to IPO? What is the actual signal that's come out for this? >> Yeah. So, there's no no official confirmation of this, but sources say that they're pointing towards September and October, right? >> Okay. >> Um, where I agree with Kylie, >> I mean, if Entropic will go public and they show in the earnings report that, hey, we're growing, we have margins, we're we're turning into a profitable business, right? then this will kind of end this entire uh AI ROI return on invest debate and would be really bullish for the broader infra buildout. Now obviously the other side of the dollar is if they're not posting those numbers that would be really bad for the AI infra buildout because then people would the bears will come out asking really questions hey are we really spending 500 billion bucks on something that is not proving to be growing or or earning money right >> would they IPO with those numbers being bad >> probably not >> that's what I mean though right is that it must they must it must be favorable numbers it must look good >> I would assume So, yeah, I would assume so. So, >> otherwise, is that is also something is is OpenAI going to race them to that because OpenAI is also supposed to IPO later this year. So, is that going to be turn into a little race or is that something where they'll probably just wait it out? >> I mean, they've talked about waiting it out, it seems. So, I I don't know. I don't know what they're going to do. It's Yeah, it's so hard to say. We don't know. So, we'll have to see what they do here. But, um >> I think you know, their numbers are growing. So, it's a it's a good time to IPO. I mean, if I were open AI, really depending on my own numbers, right? I have no clue how their numbers are. But if I'm open and I think my numbers are better than Antropic, I would probably wait it out, let Antropic go public, and then come to the market and say, "Guys, we're better. We're growing faster. We have more, right?" >> Yeah, I don't think that they're better numbers than Entropic, though. Entropic was smashing them, at least earlier this year. >> Agreed. So, let's talk memory for a sec because I know many listeners care about this. There's some really good um estimates that came out from UBS. I'm just going to share a chart here and walk you guys through. So, you actually mentioned I think at the beginning of the show that what was the guy's name? Yukon. >> Yukon. >> What's Yukan? So, he's a big memory bull. He came out said he sold all of his memory. Um and I don't know if he was long-term bearish on memory or if he was more short-term. I think it was short-term bearish. >> Shortm. Um and so but what's interesting is UBS just came out with some estimates. Um and and the big fear around memory is that it's it's cyclical and um it's the margins that are currently like close to 90. I think they're like 89% or something right now are going to end up getting eaten, you know, by 2027 or something like that and and we'll go down from there. And that's the kind of cyclicality of memory. What UBS puts out is this chart here which is quite interesting. So we're looking at two things here. We're looking at DRIM. Remember there's memory is is not just memory. There's different types of memory. So there's there's DRAM uh and then there's HBM. DRM is the the big core product of of Micron today. Um it's generally the thing that's more cyclical. And um what UBS expects is that the margins are actually going to continue to increase all the way until mid 2028, but the margins will actually stay higher than they are today in this moment all the way until mid uh 2029. um which is a pretty crazy. So we got two and a half years essentially uh or I guess is that two years? No, two and a half years of margins higher than they are today for for DRM. Then they eventually go down but at the same time which is the the product that's now growing a lot more which is HBM because this is what's needed in data centers for GPUs things like what Nvidia has. Um those margins are just going to continue to rise all the way until the end of 2029 and we don't even know when they're going to come back down. Um, and that's the very interesting kind of takeaway from it all is, you know, currently those margins are sitting at 65% or 68% according to this. Um, and we're going up to basically 80% by 2029. Um, which is just an incredible opportunity for any of the memory companies, uh, like Micron. So quite bullish. Um, what are your thoughts on this? I think you saw some of these numbers, right, Vincent? >> Yeah. So the the whole debate was kicked off last week by the SanDisk earnings and they reported crazy earnings, good margin growth, but the the guidance in terms of gross margin was not what the market expected. It only grew by I think one or 2%. Right? If you look into the the history of what the memory market is, what usually happened was there was demand for it. They increased capacity. They oversupplied the market. Gross margins came down. Now what you need to understand is the type of contracts in the memory space completely changed. Right? It used to be really a consumer dominated market, right? Apple for instance was one of the key customers and they were essentially doing with the memory suppliers whatever they wanted, right? They they they killed agreements. They were not buying stuff. Um they they were negotiating really hard. Now all that's switched. The memory uh suppliers are coming to names like Apple telling them, "Hey, we're signing long-term agreements with you with floor and and and and kind of top prices, right? To kill that cyclicality." >> Micron's coming to Apple and saying, "I'm the daddy now. I'm the daddy now." [laughter] >> Basically, and we did not remember what you did to us a couple of years ago, right? [laughter] Exactly. >> But what you need to understand is the priority of those memory names is they want to kill this cyclicality in the business and obviously also those swings in the stock prices. Now with those long-terms agreements, they're doing that. But what they're also doing with that is they're not squeezing every inch of margin that they have at the top end. So they're they're saying, "Okay, we take a little lower gross margins for those long-term agreements because we can rid of get rid of the cyclicality." And that's what we're starting to see in those uh guidances of companies like SanDisk and the market is reading it differently because in the past it was different when gross margins peaked. Right? So that's my view >> remaining and what what's interesting for the microns highness of the world is so they've already sold out a bunch of their memory for all of 2027. So you can basically predict and see exactly what their revenues are going to be. We know what the margins are somewhat. We don't know exacts, but we we've got pretty good ideas. So, like for the next year, the earnings are not going to be much of a surprise for the Micron, the SK highness of the world. What will be the surprise, what I'm most excited about, why I think they'll have their next leg up is so the free cash flows of of these companies is about to be massive heading into 2027, right? As they start to sell the supply at these very large high margins, right? That's only started ramping up really this year. Next year we'll have a full year of just like high margins, you know, most of their supply being completely sold out, if not all of it and and probably growing those contracts into further years, 2028, 2029. So, what are they going to do with that free cash flow? Well, right now, Micron cannot buy back their shares or give dividends. Um, that's part of I forget what it was for that they were doing, but anyway, they they can't do it until December. So, in December, I think it's like December 6th or something like that, all of a sudden they're allowed to do this again. They can they can um buy back their shares or give dividends. Uh SKH Highix just mentioned last week that they're now starting to think about um giving back to shareholders. So, I think what's actually going to happen here is over the next year, we're going to start to see a lot of money coming back to shareholders because they have so much free cash flow. Um and I think that's going to be a driver of the stock. Now whether we have to wait for December for that to happen, SK Hindix can obviously frontr run Micron because they don't have those restrictions, but probably I think the market front runs them actually buying their own shares. Um and so I do think that's going to be a cause for Micron to get back up above a thousand and and and probably much higher as the market realizes two things. One, you can get money by holding the shares. two, it's not actually as cyclical as everyone thought and they're continuing to hold these margins um and expand out into 2028, 2029. So, I think it's a bullish outlook. >> Yeah. And there were two more data points that we can mention in terms of the the bullish outlook. So, first, Apple came out, I think it was last week or this week, saying that the iPhone 18 Pro costs are expected to surge by nearly 40%. 99% of that is memory costs, right? and they're rethinking their gross margin strategy. That tells you about who is the daddy in this negotiation, right? Um, and then number two, which is even more important for the AI uh kind of listeners and buildouts, Nvidia is slashing memory content by 50% in its rare CPUs in in in in the new chip architecture that is basically coming out. And this is not because they they found some efficient way of of of of making the the chip work with 50% less memory. It's because they cannot get their hands on enough memory and they're actually giving away some performance of those chips just to get them out. Um that's just some some two more qualitative data points on why this memory buildout is not is not slowing down even though the the kind of sentiment on X and I think this is an important point these days because the the way stocks trade is really about narrative and sentiment and that's also why I don't think memory stocks will go up like this again because it takes some time to build the narrative again and we need something to to to to accelerate that. Now, we had the 500 billion of of Nvidia announced today and it did not move the market really, right? So, we need something to kick and spark off the sentiment. Again, >> Vincent, I had a question on my live show that I did yesterday and someone said, you know, every day you post a bunch of charts and data points that are super bullish about the infrastructure build out. Why aren't why aren't the prices moving? What is your answer to that question? >> Yeah. So I think one part one one thing is that those stocks moved really hard uh this year and performed really well especially memory especially stocks like Bloom Energy and they're just taking a brever which is completely normal. I mean memory did what 6 7 8x or even more just this year. Um so it just takes time to recover. Um and also people are taking money off the off of off off of off off the table and reallocating it elsewhere. Um, yeah. And then number two is what what I was just explaining. It's it's sentiment, right? All the hard data points are to me at least picturing this this bullish future, right? And the the 500 billion of Nvidia was just the last data point supporting this teases. So to me, it's really all about uh sentiment these days in stock markets. You had a great post I think a couple of weeks ago or so where you're saying, "Wow, stocks mark stock market is really trading like crypto these days." It really feels like it's really narrative driven and to me that's actually really strong argument on why yeah the names are not up more these days. >> Would you argue Kyle too that some of these or either of you would you agree that that the numbers that you're sharing are publicly available? Would that not would that not make it priced in or at least that whether it's like actual price actually priced in for profits or not these projections aren't secrets, right? So, but even Yeah. So, would that not be part of it? >> The the data is definitely not a secret. Um I think the market doesn't fully believe it. Um which is why two things keep happening. >> Analysts on Wall Street keep having to raise their estimates for capex every few weeks. Like literally every few weeks now. Um, and also they have to raise their estimates for earnings uh every quarter because they keep they keep being wrong, right? They're they're underestimating earnings. They're underestimating capex and every quarter they've got to like rethink that. There's a good chart. Actually, I can I can pull it up here. I think I'm still sharing screen. One sec while I pull it up. Um but generally if you look at this chart here um look at the revision changes that they've had to make in 2026 almost 15% so far in 2026. They've already had to revise upwards of 13% for 2027. Um and so you know generally it's the opposite. Generally they're actually have to revise down because they're too optimistic. And I think analysts are not optimistic enough. And I think it's because it's it's like the market doesn't quite or the analysts anyway are not quite believing in the impacts of AI yet. And I don't know why that is because if you listen to these earnings calls like it's there. Um maybe it's because of the like um this like shadow like we don't know what's happening with Enthropic and OpenAI and that's really the like the the key piece of it all and so like we need them to go public for the market to finally believe it. Like I don't know if that's it. Um, but every other data point you look at is extremely bullish, but the analysts just continue to underestimate it all. Um, and I don't know if that's because like they don't run companies, so they're not seeing the impacts of AI inside of their own company or they're not using AI enough. Uh, I don't know what it is, but they continue to underestimate everything. And I think that's why this is happening. >> I I think this ROI part is really important here. What what we need to kick sentiment off again is something like Tesla scaling robo taxis or Eli Lily coming out saying, "Hey, we have this AI created drug that is yeah, solving every every disease there is, right? We need or Antropic going public saying, hey, we're growing like crazy, right?" Palunteer to me was this kind of earnings that that proved AI can be valuable and can be adopted in a in a way that is is is increasing efficiency, decreasing costs, whatever, right? But we need more proof of AI really creating value. And I think that's what the market is missing and that's why they're not giving those multiples or or at least at the moment not not not sending those stocks up higher because obviously if we're not seeing that return then why do we need to invest in that infrastructure? I think that is the core of it. >> Yeah. At some point the market will come around to I don't know what it's going to need but you're right it probably needs that new use case that new thing where they're like oh okay we missed that right. Um I don't know what that's going to be but I I can see what you're saying. We need that >> new products and I think you've been saying that for a long time Kyle, right? This is this has been the drum you've been beating for months is like the application layer, right? And you're seeing it reflected in some earnings, but those are things that it's like the broader public and like hype machine. It's hard for it to latch on to that >> Eli Liy's using AI to make new drugs or you know whatever Salesforce using AI or that kind of stuff. It's like that's kind of hidden where something like even you think about what happened the start of this year. Cloud was it was so public. It was so public what it could do is replacing people's jobs. SAS apocalypse like that was that was such a huge like PR push. So with a you know the market's going to need another >> like campaign basically [laughter] >> consumer agents consumer agents >> some need >> right >> this is also where the biggest opportunity is right my bet for instance is hood >> most investors today in hood are not understanding what AI agents are going to do to their volume of transactions on the platform now fled was talking about it on the earnings call actually investors still don't care right I was um texting with another investor pretty well known actually uh on X-ray asking hey are you modeling this and he's saying no there's too little information don't know how to model it right so the AI adoption side of things and we saw it with Palunteer right the crazy earnings they showed value boom 30% the stock was up I think Lily Tesla Robin Hood are all names that once we have the quarter where we see the inflection AI are really impacting the bottom line then those stocks will pop and that's >> the trade that people are not in at the moment. >> Yeah, that's a good point. I can see that consumer AI might be might be it right. This the the AI Siri if that comes and the world's like holy this is it. That's your next chatbt type moment. Um that could be one of them robo taxis one we've talked about a ton could be it. Yeah, I agree with you. we need we just need that to kick in which I mean here's the problem is if that takes too long then that could be you know a hold on the market for a while um so we'll have to see but I in my opinion I think the agentic world is still like there's so many things that are happening with agents it's not one big thing right it's just so many things in the background that are happening with agents that are so valuable for companies which is why I think earnings are going up so much um and I just think that the market is not understanding that yet. Um, and we're still very early in that. And so I just think as every company starts to integrate agents in their own way, um, it's a huge huge huge push for for earnings, which doesn't drive up and for the economy, which doesn't drive up, um, like inflation. And so I think we're going to this world where earnings continue to go up without inflation following with it, which doesn't generally happen, right? Usually when earnings goes up, it means companies are hiring a bunch more people. Inflation ends up going up because we're consuming a lot more because wages go up. And then the Fed needs to raise rates and I think we're not having that happen right now where earnings are going up but unemployment or sorry employment is staying the exact same and actually jobs was down recently. You don't often get those two happening at the same time which tells me could we go into this like earnings bull market without the fear without inflation going up with it. Um which means rates can stay low or even go lower. That's a hell of a bull market uh opportunity um for a variety of different companies. Why I keep calling it the everything bull market. That's what I think we're antering into. Um, and I don't know if that needs that one catalyst. It's just every company's catalyst, right? Because every company's using it and seeing benefits. So, maybe we'll get there. I don't know. But that's my that's my kind of base case at the moment. >> Maybe we'll start again when Anthropic gets their their PR campaign going for their IPO >> because if they are going to IPO, they're going to need one hell of a campaign on top of their on top of the money they're making. Uh, gentlemen, I think we can wrap there. Great show. Didn't even need Melvin. Uh, hopefully he's alive. Uh so uh hopefully everybody in Chicago is doing okay or in that area. Otherwise you guys great to chat. Thank you for all the thoughts today uh and we'll see you next week. [music] Want to stay ahead of the biggest technological shift in history? Subscribe now to get insights straight from the sharpest minds in tech and finance. Quickly you'll note this show is for educational purposes only. Nothing here is financial advice. Investing always carries risk. Never invest more than you can afford to lose. Thanks for tuning in. See you in the next one.