The AI Cataclysm JUST Started | AI Stock Analysis.

Hey, quick note. Today is Lauren's birthday and Lauren asked me to make a special bonus coupon for today only on the courses on building your wealth and the alpha membership at meetke.com. Also works over at househack.com reinvest. Check that out. I'd love to see you there. Special bonus coupon expires tonight at 11:59 p.m. In this video, we're going to discuss the bifurcation of artificial intelligence and what it means for your portfolio. We'll talk about some stocks to avoid and some stocks that could be big beneficiaries around this concept. But I really want to use this video to explain this concept of the bifurcation of artificial intelligence. And while that sounds complicated, it'll make a lot of sense towards the end. Unfortunately though, it does mean we have to start by talking a little bit about Zuck. Yes, Mark Zuckerberg. Because on well today Meta announced that they're going to release their more powerful than Meta Spark model Muse Glimmer as an open weight model. And this is pretty powerful because it's coming at the same time as Alibaba is releasing their Quen 3.8 27 billion parameter model this week. Zuck is now taking the point of view that the frontier or best quality LLMs shouldn't just be built for the highest paying companies or governments or institutions. It should favor individuals. This is kind of the Elon Musk mission with open AI. Elon Musk backed open AI so that artificial intelligence would be available for the people. And so it was not on my bingo board to see Zuck kind of take that lead, if you will. It's possible that's happening because Meta has been falling behind on their own version of Frontier AI. So, it's kind of easy to bag on the leaders and go, "Ah, well, Frontier Labs and the leaders, they're they're losers anyway. We we want to give this to the masses." You know, kind of convenient timing. But it comes at the same time as there are some serious concerns about the scurve issue that anthropic and open AI face and that is what sets up the bifurcation of AI and it changes the way that we invest. So in order to get this you have to know that there are really two ways AI funding gets spent. one is at the frontier compute layer. This is and then we'll talk about S-curves and that in just a moment. This is where the bifurcation starts. You've got on one hand your frontier compute, your Vera Rubin, so the Vera CPUs, the Reuben GPUs together as a GB300, a fullscale Blackwell cluster. You're spending millions of dollars per rack for these, somewhere around $4 million just for a rack on these. Uh there's talk that Jensen is trying to introduce these models with various or these these racks with different variations of memory. So that way we could get away from some of the most expensive AI uh memory clusters and get to clusters that could potentially perform just as well with less memory. And maybe we could use less memory because it's so expensive now and use some CPO technology which is basically fiber optics like what Broadcom or Marll produce to make Vera Rubin great without being an arm and a leg in terms of cost because the Frontier Labs only have a finite level of money. Anthropic OpenAI plus the Google or even the Microsofts, right? There's a limit to how much money these companies can realistically raise. And the Frontier Labs give Nvidia the highest margins possible. Okay? Somebody buying a 5090 consumer grade GPU isn't giving Nvidia the same margins as Elon Musk going, "Bro, I'm gonna buy 200,000 GB300s, please. Okay? Don't care what the price is. I'm a trillionaire." Right? Like so on that's one hand right frontier compute. Okay this is where like do you know what Microsoft just pulled off? Microsoft literally just changed their depreciation schedule from 15 years to 25 years. They did that because now all of their leases for offices and data centers are getting reclassified as an operating expense or a cost of goods sold rather than a capex. That's because right now a lot of executives know if you show high capex because you're spending so much money on these Nvidia chips, your stock could go down, man. You know, you're getting punished because the capex is too much. Google raises money. Google's been off alltime highs. They've been suffering because they're coming out with a big sucking. Oh, we need we're going to sell $40 billion worth of stock. We haven't sold any of those yet as of June 30th. So, we still got to sell that. Maybe they've been selling here recently. They also raised money from uh Birkshire Hathaway and another about $25 billion on top of that. So, you know, there's some real money raised going on. But the point is that's going into the main frontier compute side of this this two factored story. High margins, big systems, big racks, expensive. It's for the big boys with the big money. And this is why Jensen cozies up to Elon, baby. Okay. Then you have the second side of the market. The second side, this is the bifurcation, right? Split the cataclysm for those World of Warcraft folks. Okay. Then you got the inference and the fine-tuning side. So big models think massive pre-training, some post-training, massive expensive data sets and um uh and and obviously therefore they need these massive servers and data centers. Then you get the enterprise AI side. This is kind of like where, in my opinion, our company sits along with, frankly, 90% of real world AI. Totally fine to use openweight models. You're not trying to create the next greatest LLM. You're trying to use the cheapest LLM possible to make the most freaking money possible. And this is why 5090s from Nvidia have doubled in price. This is why the RTX 6000 Blackwell chip has doubled in price. Reinvest owns both of those. But the reality is we can get 90% of Frontier well quality and and frankly you know we're maybe three months behind Frontier LLMs in and when I say we I mean like all enterprise AI that's running on local AI systems. You're maybe 90% of the way to frontier quality. And I use that on a time basis. Maybe we're 3 months behind is what our dev team thinks in terms of how far enterprise AI is behind the frontier labs. So those big spenders are here and you know most enterprise and businesses are like maybe three months behind because of the benefits of openw weight models which if you're three months behind and you know this AI boom's been going on for about 36 months maybe a little bit more closer to 40 months you're really only time-wise 10% behind you're 90% caught up that's my point so the businesses using Blackwell RTX 6000s which run for about $18,000 or 5090s which you know run for somewhere around 4,900 bucks now unless you buy them in bulk. People are able to get a massive leap towards frontier quality on consumer grade chips or like business grade chips. Not frontier level chips, business grade chips, RTX 6000. You're not spending $4 million for a rack. You're spending 30 grand on an AI workstation and then you buy like five of them. Okay, big deal, right? You could do a lot of compute with that. You could, that's what we're doing with reinvest. We could do a lot of damage and a lot and make a lot of progress with these workstations. And so then the question is, okay, so how then does this affect the entire market? How does this affect where you want to invest? And so this is where I got to thinking to myself, okay, maybe you could invest in a company like Corsair because if people are buying workstations, well, great. people buy workstations, they're going to need PSUs, they're going to need motherboards. Sure, they'll need memory uh and and some of these GPUs, but those GPUs are lower margin for AMD uh and Nvidia on those consumer grade products. So then you think, okay, well, you know, Corsair kind of packages all of this together, but I hate to say it, when I look at the financials for Corsair, they just had banger earnings. But when you look at the financials for Corsair, you're kind of like, uh, they're not the greatest. Look at them. I'll throw them up on screen right here. I've got sales down 2%. Now, in fairness, their gross profit is up 21%. So, they've got some growing pricing power here. It's a name to pay attention to. It's bouncing off a low, right? I'm not here to bag on it. I'm actually saying it's bouncing off a bottom right? uh and and they are going to sell AI workstations, which they do caution won't be part of their revenue until like late 2027, but a lot of people are already selling these stations. They've only got an 8% net margin. So, they're really trading for a two peg 13.7 PE ratio uh or sorry, a 15.2 PE ratio. Earnings are expected to be 90 cents and you've got a growth rate of about 7 to 10% of the company. So you're trading for two peg for an 8% net margin. It's not great. The balance sheet is okayish and uh you know they do a free cash flow of about $98 million. It's a smaller company you know $ 1.5 billion company buying back a little bit of the issuance that they're doing for stock comp and paying off some debt but these are little nominal numbers. And so I'm like ah these layers are commoditized. So you really have two ways to invest. One way is the consumer towards enterprise level, but that's really commoditized. The other way is memory and edge level GPUs like the high margin Nvidia chips or AMD chips. But the problem with that in my opinion is those guys are heavily reliant on their valuation or the valuation of companies like Anthropic. And based on I I don't know how accurate this is, but this chart at stockanalysis.com shows Anthropic stock valuation has actually come down from a peak of nearly $1.4 trillion down to $826 billion. I don't know how accurate that is, but if that's true, there needs to be a fire under the butts of Daario and Sam Alman to IPO as soon as freaking possible. That way they can raise as much money as possible, keep blowing money on that really expensive artificial intelligence hardware from Nvidia and AMD and keep the hardware cycle going. Otherwise, you're going to end up hitting I mean you're definitely going to hit a limit. My opinion, okay, when I say definitely, I want to be clear that's an opinion. I think artificial general intelligence is fugazi. you are going to hit a limit with artificial intelligence at the way we know it today. Uh this is a typical S-curve dynamic. Artificial general intelligence in my opinion is a lie. It requires the ability to reason in in scenarios you haven't seen before. Not relying on patterns. That's something that AI today is really bad at. AI today is really good at patterns and pattern matching and and linking patterns. In fairness, you have to adapt knowledge from one domain and apply it to another. AI today is not great at doing that. You have to recognize when you're on the wrong course. You have to selfrecognize when you're hallucinating. AI is not good at that. And you need goal management to determine when should you stop, when should you ask for help, when should you restrategize. AI is not good at that today. And you have to recognize when there are practicalities that limit your AI suggestions, if you will. AI kind of fails at all of those levels. And and that's fine. That's not saying it's bad. It's it's potential. But we're trying to solve those problems. And what we're doing is we're, as Michael Sailor would say, asmmptoically approaching the limit of technology. We're approaching the line. This is like Bitcoin issuance, right? This part right here. We are asmmptoically approaching zero. We're asmmptoically approaching the end of progress to AI. And this is why these openweight models are catching up. So damn, Kevin, that is a lot. Too many big words. Simplify this crap for me, Kevin. Just tell me what the bottom line is. Okay, the bottom line, actually, I have to tell you one more thing. One more thing before the bottom line. U in order for Nvidia to keep all this crap going, this spending going, it's no surprise, and I looked at Nvidia's financials this week, and it's no surprise that Nvidia is exploding their investing activity. Most people don't look at this on the cash flow statement for Nvidia, but Nvidia is blowing money into the purchases of non-marketable securities and marketable debt and equity securities. $26.5 billion in one quarter. That is more than they did in all of 2025, including these investments into Lancium, uh, NBIS, Marll, Cororeweave Intel Lummentum Coherent Synopsis, Nokia. Why? Because it's simple. Nvidia invests in these companies. Those companies buy chips, which is high margin revenue for Nvidia. Investors in Nvidia rejoice, and then there's more cash available to keep the cycle going. It's circular financing at its best. Open AIOropic bottom line might end up being a tough a peak in terms of when they IPO for what they end up doing for AI hardware. Okay, that's like your peak push potentially unless Microsoft and Meta and Google somehow keep this going maybe. Okay, that's the frontier level side for the enterprise side. I mean, yeah, there are companies that put stuff together. I mean, you could argue AMD's got its Halo, Nvidia's got the Spark. These are lower margin products. I don't we don't find them particularly useful. Apple's got the mini, right? You could use those as AI workstations, but lower margins that makes none of these really particular desire, really particularly desirable on a workstation basis. And you know what's more desirable? There's the bottom line out of all of it. It's the companies. And this is why I put my own hat on. And I'm like, well, like as as openweight models get cheaper and cheaper and we could buy this hardware and do more and more damage with hardware that's getting us close to frontier level at a cheaper price, we're basically taking advantage of this commodity. And so we're winning in my opinion as a prop tech startup, you know, a real estate uh technology company. So that makes me really excited. Now, I get that that feels a little anticlimactic because it's sort of like, hey, we've been looking at all these great hardware names, and I still agree. I still agree that if the second wave of hardware investment comes the way I expect it will, it's going to come from the dumping of SpaceX money, the dumping of anthropic and open AI money. I still believe that second wave comes as a medium-term trade, right? That's like a six-month opportunity. Still, Nvidia, Marll, uh, LSCC, some of these other names, I think they can still do quite well over that six-month period of time. But my point is, we're approaching the limit of the end of that big spend hardware cycle, and then we're going to move to the lower margin hardware cycle. A move from the high margin hardware cycle to the lower margin hardware cycle, the more enterprise optimized cycle is not bullish hardware because it's commoditized hardware. So let's let's draw that because you know I saw this shift to CPUs back when AMD was like 196 bucks and we talked about it in the alpha membership. We're like hey reinvest is buying a lot of CPUs. This could be an early trade, right? And people regularly say, "Kevin, man, you come early all the time." Uh, but my thinking is if the dollars from the AI boom have gone into high margin first, then and those are the beneficiaries of Nvidia, the memory stocks, Micron, whatever, all the high margin hardware. Then naturally as we go into low margin, people are going to be looking and going, "Oh, okay. Is that going to be like a Corsair, you know, your PSU suppliers, your assemblers, your enter, you know, your basically your little computer system put together. That's a new word. Uh I think the answer to that is no because there's still low margin plays. And so the real pricing power is actually in the survivor uh software companies. That's where the money is. And so that's where the effort has to go is picking those. Now, you know, I I don't always like to just do the oh get the full list over at me.com. You get that? Okay. But I'll just give an example. We already know Palanteer is one of those beneficiaries. Another example would be like an axon, but there's a lot of volatility in these. For example, in my opinion, there was no rationale for Palunteer to really bleed at all. Yet that my fundamentals could say, "Hey, it's actually got a good valuation around this 130 range." Sure. Would I like to buy it as a backup the truck at 89? Yes. But our technical analysis, as you can see, was converging towards this 106 line anyway. And that's why we bought more exposure to Palanteer before we ever got uh through 106 because there's no guarantee you're going to get to your target, right? The point is to me there was no fundamental reason to see a bleed out here when it is actually one of the major beneficiaries. Another example of a major beneficiary is a company like Axon. Now the problem is it's software. So you get a lot of volatility. We literally just broke into this down channel and what a surprise, we literally rejected the top of that down channel. I mean, it's almost a perfect down channel, right? So with this said, the market isn't of this mindset yet. Even though we can get this massive boost in Palunteer stock in the span of few days, some of that is going to be trade momentum. I mean, look at this. We're massively overbought on the day. If we go to the week, we literally just went from almost oversold to overbought in the span of a week, right? So that this is early and that's my point with this is this transition from high margin to low margin hardware isn't beneficial for hardware at all. Instead, it's beneficial to software, but that takes time. And Q3, Q4, I hope, which is now to the end of the year, is the beneficiary of that play. Uh, and to me the fundamentals point in exactly that direction and it's exactly what we're seeing at my startup. And again, we're a propt tech startup, right? So like we have about, if you include developments and the real estate we have at market value, about $100 million of no bank debt real estate. And with that foundation, we are also building our own artificial intelligence suites for what we're eventually going to call as a as a terminal. And that's going to include the housing artificial intelligence and the stock artificial intelligence and the algorithms that we're building to basically take Kevin's mindset and scale it out to any company or most importantly any house to determine, hey, is there a potential good deal here? Obviously, you still got to go look under the hood yourself, but it's a great head start. And so that's what we're really excited about and that's what we're seeing as we're building this company uh piece by piece every day. We just build build and we're really excited. Uh and so I feel like it's my duty to share that with you. Uh and so if you want to join us and get a little bit more behind the scenes. Hey, join us over at meetke.com. We'd love to have you. Use that special Lauren's birthday coupon code. Little bonus for the day and we'll see you in the next video. Thanks so much for being here. Goodbye and good luck. Why not advertise these things that you told us here? I feel like nobody else knows about this. >> We'll we'll try a little advertising and see how it goes. >> Congratulations, man. You have done so much. People love you. People look up to you. >> Kevin Pra there, financial analyst and YouTuber. Meet Kevin. Always great to get your take.