AI Slowdown Bullish? Full Breakdown & Transcript

[00:00] People are misinterpreting this massively, and I'm going to break my thesis down because I really believe that I'm going to be right about what I'm saying is not what the popular opinion is online right now.

[00:14] We're going to talk what Donald Trump just said. We're going to talk what's going on with the AI labs and this agreement and what this means for AI data center spending as well as the bubble.

[00:26] and basically I'm going to tell you why all of a sudden I am really effing more bullish than ever before. So if you're looking for a bear porn, this ain't it, but it can challenge your thesis

[00:40] and you can yell at me in the comments to tell me I'm wrong, which is fine. I love reading the comments and seeing, hey, somebody give me reasons as to why I'm wrong. Of course, then there are those people out there like, Kevin, this is clown go.

[00:52] And then I'm like, I know, but can you give me a little bit more on why you think I'm wrong? So. All right, let's get into it. So obviously we just heard that Tim Altman, Dario, and Elon Musk sort of all agree,

[01:06] including the chief of Gemini, all agree that, yes, maybe we should slow down on our AI research because the models are getting too powerful. Okay, this, and the head of Gemini argues, you know, maybe we can talk about the details,

[01:24] but directionally this is correct, was their argument. All right, so let's put some of the jade behind us. First of all, the jaded view, which I also still agree with, is that this is all just marketing. Okay, our models are so good.

[01:37] Oh, we accidentally left the door open and they escaped the sandbox and they're trying to kill people. Fantastic marketing. the reason okay that's jaded point of view number one

[01:49] the models are so good that you should use our product because it's so good it's going to kill people right I know that sounds weird and crazy to say but I think that's ironically marketing because it's sort of like

[02:01] wow these companies are going to take over the world they're going to destroy every other company which I also don't think it's true hence why I shoved that all into the marketing argument after we argue that it's just marketing there's another jaded point of view they all just want to

[02:13] slow down spending because the election is coming up and Donald Trump dismissed fears about how strong AI is and how it's going to kill everybody and that could end up hurting him in midterms. For now, if the

[02:26] labs say, hey, you know what, we'll self-correct and we'll slow down, that'll help Republicans get elected and that will prevent democratic regulation and restrictions on these

[02:42] labs. So, in the short term, you take a 50-day pause, 52-whatever-day pause, and you get a Republican, and maybe that helps you regulatorily. That's jaded point of view number two. That's

[02:55] not even my thesis, okay? That's the jade. I just want to catch you up on the jade. My belief is actually different, and it aligns with my peak fear theory that we talked about

[03:09] just a little bit ago. My thesis on, so just to organize this, we'll talk about the labs first, and then I'll talk about my sort of bull thesis for the entire market. Okay, so my thesis on the labs first.

[03:23] I actually think that the labs slowing down artificial intelligence from spending on new models or whatever is an IPO pump. Let me try to explain this.

[03:35] So why is that an IPO pump? Well, because if there's a coordinated slowdown in sending on the bleeding edge, then that means there's a coordinated slowdown on how quickly, basically, the bubble pops,

[03:51] which makes people nervous. That's point number one. But also, number two, it props up the appearance of profitability or potential future profitability. So watch this, and I'll show it to you in actual numbers in just a moment

[04:06] because we have this document here, a primer on AI lab and hyperscaler economics. That's very important. But I just want to start really basic here, okay? So the thesis is that opening eye and anthropic are going to IPO at some point.

[04:21] This idea that Sam Altman says we're not going to IPO until 2027, old news. He's been hinting at that for months because they're expected to IPO in October, which has also gotten delayed. And the whole idea here is Sam wants to see what happens here.

[04:35] If they do well, he'll IPO. If they bomb, he'll stay private. It's actually like optionality for Sam Altman, right? But anyway, the whole idea here is that these companies make revenue.

[04:48] They spend money on costs of goods sold, which is their inference compute, okay? This is inference the same kind of stuff that Dell is selling you the enterprise compute You know one of the reasons Dell doing so well is because firms want AI in a lot of of them Not saying all of them but a lot of them want AI in You know we have a crap ton of AI in I mean I think we got like 50 different

[05:11] machines running. You see our electricity bill. We actually just got some Tesla solar panels because of it. More on that in a different video. Anyway, revenue minus COGS gets you GP, which is gross profit. They're shoving training into R&D. Old news. We already know

[05:28] that. The problem is, a lot of analysts are looking at this R&D, and they're like, dude, you guys are going to have to keep spending on the bleeding edge, and so your net income is going to be at a loss forever. But if everybody in a coordinated manner, like, it's sort of

[05:41] like you're in a horse race, okay, horse one, horse two, horse three. If all of them together decide, yeah, yeah, yeah, let's slow down a little bit, it's kind of like the, and China's not going to slow down, but the U.S. one, it's kind of like the whole oligopoly together,

[05:55] like OPEC almost, is kind of like, yeah, yeah, yeah, we're going to reduce production. That's going to pump up our R&D, which actually pumps up our net income. It makes us profitable faster or more profitable faster, right?

[06:12] Now, if only one of them, like if these two, if Gemini and OpenAI, let's say, slow down, and if Robert's like, F it, boys, we're going to keep spending, then they're going to be ahead of everybody else. And then that leaves them in the dust.

[06:25] which is why you kind of need them all to coordinate their slowdown. Now, obviously, there's going to be some cheating and lying and all that. We expect that. But my point is, if they all say they're slowing down, it's going to make analysts, in my opinion, prop down the R&D expenses

[06:41] and make the companies actually appear more profitable. That actually extends the very bubble and is ironically bullish.

[06:54] more on that in just a moment. We're going to come back to this AI primer on hyperscalaries in a second as well, but I want you to see my tweet. Okay, so I lost my checkmark, by the way, because I just updated my profile. I hope you like my new profile picture of my PP.

[07:09] You can see my PP right here. Put a nice little banner here. I had the flying picture, but I thought this was kind of cool. Data7, stock and real estate analyst, CEO of Reinvest, a nine-figure firm, two million on YouTube, featured on blah, blah, blah. I don't know. Give me feedback on it.

[07:23] Let me know what you think about it. I think it's kind of cool. Okay. So, this is what I just tweeted. The stock market is about to skyrocket. We hit peak fear. Hear me out. The Saudi pipeline, this has to do with the Houthis.

[07:36] We already knew about that. That's what led Brent to hit 107. That was old news. But, like, most people didn't hear it on, you know, CNN or Fox or whatever until, like, lingering the day on Friday.

[07:48] And for people, like, what? It was already priced in, like, late Thursday. Anyway, that's already been priced in. Three to four rate hikes have been priced in. Everybody's so freaking worried about the rate hike. Bro, the market's already pricing in three to four.

[08:02] So if we get fewer than that, or we get no rate hike, that's boorish. And if we get those rate hikes, face case priced in, baby! Okay? Plus, we got the nearly 5% on the 10-year, which in the long term,

[08:16] I mean, just zoom back the last five years, this is always where it peaks. Maybe this time is different, but this is always where the tenure peaks. No deal coming until after the election.

[08:29] Okay, that was leaked by J.D. Vance last week. And Donald Trump, both of them, the White House, they're like, yeah, if you look it up, look up the news articles. They said, yeah, we might not have a deal until after the election. And then Donald Trump is like, oh, Americans won't mind

[08:43] because we have to stop Iran from getting a nuke, and they understand that. Okay, well, they do mind, but both should be true. Also, there were leaks from the White House last week We covered all of this on the channel, but you know this. That's why you're subscribed here.

[08:56] If you get too many videos, by the way, a tip that you could do is download the Meet Kevin app. It's free. Customize your notifications to which playlist you want. Like, last week, in fairness, okay, I posted, like, three pilot videos.

[09:08] And I don't know. I think, like, 80% of you don't give a flying pun intended F about the pilot video. So just uncheck pilot notification. You can do that on the Meet Kevin app. You can't do that on YouTube. Anyway, might be at war the entire Trump term. All that was leaked by Trump last week, we covered that.

[09:25] Okay, that's really bearish. Oh my gosh, we might be at war, another forever war for the next two years? This sucks! Now sure, we're not like way off all-time highs, right? Because the last video I made, this was peak fear, people were like,

[09:37] But Kevin! But Kevin, we're banging off all-time highs! How could you say you're blessed when the S&P 500 is, like, very down, bro?

[09:51] I was like, all right, well, I mean, QQQ's down, let's see, I think our top was, like, 747, right? So QQQ's about 715. So 715 divided by 747 on QQQ So that down like 4 4 rather Who cares Okay the S 500 has peaked at 779

[10:12] if I just use SPY. So 764 divided by 779. So we're down 2%. Yeah, also big deal, right? Kevin, how could you be bullish? We're so close. RFP, which is another one,

[10:24] we closed, call it 215, 215 divided by 223.44. even that peaked out, down now 3.8%. But the point is, like, those are small numbers, right?

[10:36] It's not like we're down 20%. My thesis, though, is that the S&P 500 and the NASDAQ, I know that looks a little whiny, but we like this one a lot for trade. But anyway, and there's no coupon code.

[10:49] Can you believe it? Like, my pitch has been, like, no coupon code, download the free Kevin app. I don't know, it feels like something's missing. Like, what's wrong with Kevin? Like, I don't know. my belief is that we should actually be like 20 or 30% higher,

[11:05] that we should be breaking through 800 by the end of the year, which is about right here, and potentially go euphoric, and this bubble doesn't stop until we get to like 1,000 on the NASDAQ 100.

[11:17] So I think there's an easy 20 or 30% in this, which if I take 7.15 and I take 30% on top of that, it's 9.29. That's not unreasonable. Now why? I get it. They're like, but carbon's a bubble, a bubble. But see, that's the irony about this, boys and girls. That's the irony about this, and nobody sees it yet. The irony about this is this, these companies are the bubble, okay?

[11:47] And if they spend less, then it implies, oh, my God, but that's going to slow down the data center buildup. The data center buildup is already slowed down by wafer and memory manufacturing capabilities and energy.

[12:01] It's already bottlenecked. So it ain't going to make a difference in terms of, oh, we're going to slow down all of the spending here. The spending is still happening at max capacity. You heard it from NVIDIA directly, but, dude, we'd be at a double if it weren't for the constraints.

[12:15] So they're constrained. So, like, literally, if we're at 100% and we're, like, we're going to slow down to the tune of 30%, we would literally be at where constraints are for NVIDIA. Them slowing down the future spend, letting analysts say R&D costs will be down, means these companies go bankrupt later.

[12:35] Or not at all, right? But my point is, if there's a fear about an AI bubble, these people opening AI and dropping Google, slowing down spending to the level of constraints, basically, also helps get some of the, like, price inflation of memory down.

[12:50] But that's actually, ironically, bullish for semiconductors, AMD, NVIDIA, the data centers, the CoreWeave, the SpaceX, whatever, because it means you can now extrapolate growth for these companies for longer.

[13:10] Because if you think OpenAI is going to go bankrupt next year because their numbers are so bad and their IPO is going to plummet or they're going to plummet after they IPO, and then analysts are saying, well, no, actually, if they slow down their AI spend a little bit,

[13:23] they're actually going to stay alive for at least two or three years longer before they need more fundraising or they need to get more software or whatever. That then extends estimates for how long they can support spending at NVIDIA and the companies that don't want, whatever.

[13:38] Like, literally, as an analyst, I go look at these earnings forecasts for companies and the memory companies, the chip companies, all the forecasters are signaling a slowdown in 2028, 29, 2030.

[13:50] Like the growth rate should slow down. This will slow down spending. It just extends the bubble. So is it a bubble? Of course it's a freaking bubble. But it's the party that you can still make money from.

[14:02] I think. Okay. Sure, we're not all way off on all-time highs. I get it. But we've been held back. We should be way higher on this bubble of cap-backs. Major financing and epic earnings. So we're probably 20% to 30% too low on the stock market right now,

[14:17] you know, index levels. So individual stocks should be a lot. Like, why is NVIDIA not a $300 stock? It should be. It's a .7 peg right now. It's so ridiculous. Anyway, is it a bubble?

[14:30] F-yeah. And it's going to pop one day after the canary dies. But the canary is still singing, okay? The canary, one of the canaries that we generally identify actually anthropic profitability.

[14:43] I talked about that in my, oh, I did a video on this. It had Jensen in it, and it was like, this is the top of the AI bubble, right? There's Marx, oh, come on, there's Marx top of AI bubble,

[14:57] anthropic. You know what? This one I think is worth watching, and we're going to reference some of those pieces in just a moment It this one is it this one It might be this one right here The collapse of the AI bubble starts with anthropic warning That like 100k views or whatever Yeah yeah yeah yeah yeah So that a 35 minute video and we really go deep on some of the weeds there That a really good

[15:22] one. That's the canary in the coal mine already covered the canary. The point is this slowdown extends the canary. It's like they've watched my video and they're like, Kevin, you're right.

[15:34] will slow down spending in a coordinated manner to align with capacity constraints anyway. Weirdly foolish. Okay, bottom line.

[15:46] Oh, yeah, and I've got to play the Trump video, too. I think Trump is about to majorly taco and declare victory before the election. He set up that this war is going to last forever, which is the perfect time for him to literally insider trade against us.

[15:58] That's why I wrote down here, you can bet his trading desk is gearing up for the trade ahead of us. I don't have a title in there. Well, I write this without any eyes. But anyway, Trump is about majorly talking about victory before the election. He'll declare a victory, he'll promise the $5,000,

[16:11] and then he'll have the highest chance Republicans control. Now, let's listen to Trump, and then I'll show you that institutional piece. What is that weird freaking pixel box? I don't know, whatever. Do what you want, but if you want to be poor, have no jobs,

[16:26] and watch other countries do great, or other states do great, you know, watch Virginia's been very proactive, they'll go other places if you want to make a lot of money if you want to make a lot more money than you've ever made before if you want to see your house go up your education get better

[16:38] I would suggest now use the word good put them in the right areas put them but you don't want to kill the golden goose because that's what you'd be doing you'd be killing the golden goose this is the oil of the next 50 years

[16:50] in many ways that's interesting data centers are the oil of the next 50 years I I mean it's a bubble don't get me wrong but does that mean it's going to pop within the next two years

[17:04] it depends if the canary keeps singing so far the canary is still singing so i'm still long-term really and i want to be clear about that because sometimes you're a cabinet it just

[17:17] flips off it now look i do change my mind when i get new data in fairness but you should you should do that too. But have I changed my mind on real estate? No, no. Whenever it pops

[17:30] up, I don't know what it is. But honestly, I kind of want to see the stock market bubble keep going because A, I want to see all you make money as much as freaking possible and as fast as freaking possible. If you want to diversify, I think you go buy some real

[17:44] estate. If you don't want to buy real estate yourself, I get it. Maybe one day you can invest in my company if you want to go buy some other person's real estate. I'm not trying to show you myself. But I want to see big money, obviously. I still maintain that in

[17:58] the long term, we're going to be fighting deflation. Get through the war, AI deflation, the jobs market's going to be concerning. Don't really have those concerns right now. The canary isn't dying right now. It's going to die in the future, but that death isn't

[18:13] right now. So let me now briefly look up those sheets. So these sheets right here are very interesting because, and watch the other video, the collapse, again, it's the collapse of the AI bubble starts with Anthropic, okay? That's just one canary in the coal mine. But

[18:28] we talked about these sheets right here, and we talked about this a little bit on the whiteboard, but I wanted to zoom into this a little bit more. The way we believe that Company A, this is from Barclays, Company A is probably Anthropic, Company B is probably OpenAI, and we believe

[18:42] that these projections here are obviously estimates, but what you can see what they're doing is they're trying to project AI gross profits after training,

[18:57] and they try to argue that they bring 55% and 38% to gross profit after training costs. And so they're only allocating 10% of the total R&D training costs to the final model,

[19:10] because they're arguing that 90% of the training is just for future models, new models. This right here, this R&D training cost, was the scariest part for me because if I subtract that off of gross profit, these companies are upside down.

[19:24] If these companies are upside down, then they can't keep spending on neoclouds and data centers and chips and more gigawatts, gigawatts, baby! But if they, in unison, slow down this line right here, they literally pump their own IP together.

[19:45] Google pumped this stock, opening iPump, and Floppy pumped. They all pump each other. It's a freaking circle pump. I see a lot of advertising. These things in each other's hair. I see like nobody else knows about this.

[19:58] We'll try a little advertising and see how it goes. Congratulations, man. You have done so much. People love you. people looked up to you Kevin Passass by Max Lannis and you too got me Kevin always great to get your take