Ai Hardware Hits PEAK on Meta's FLIP FLOP

[00:02] Welcome on back to another live stream. Today, we're asking the question, is Today, we're asking the question, is this the hardware top that is Meta announcing that they are selling excess capacity? Something that Mark Zuckerberg

[00:16] capacity? Something that Mark Zuckerberg had alluded to only happening if they over built. Is that what he said? Where are those interviews coming from? Where just convenient because they've gotten a lot of bad press over the last two or

[00:30] three weeks about token maxing and how much money they're blowing on tokens? true, but let's take a look at what's going on. First thing on the stock. The

[00:42] stock has obviously been in the freaking toilet for a while. However, the stock is up 8.8% adjusted today. I'm actually a fan of the valuation. I think it's relatively cheap where it is. We'll talk more about

[00:56] the valuation in a bit. But it's clear the longer-term downtrend or longer-term uptrend that the stock has had since 2022 has turned into a bit of a consolidation. You could definitely say above that 544

[01:09] level. Anytime you see this stock around 544, by the way, it's been a buy since October of 2024. So when you get those bounces, it's pretty nice. Right now, we're bouncing around that 607 kind of midpoint line,

[01:22] which we've bounced off of before. But the news that Meta is going to sell some of its excess compute comes just after it signed an extended contract >> [laughter] >> another extended contract with

[01:35] Coreweave, which is really interesting because then it begs the question, if because then it begs the question, if Meta is selling extra compute, or have they over built themselves? Did they over lease themselves by leasing too

[01:48] And does this mean they're going to be less of a customer for companies like Envis and Coreweave? And then potentially, what's worse is Meta somehow able to turn around and sell the same stuff that Enfabrica and

[02:02] CoreWeave are doing for a premium somewhere else. And so, we'll dive into a look at Enfabrica here. Enfabrica on the day down 17%. And CoreWeave on the day down 14%. Cerebras, unrelated, you know, their

[02:18] their their own sort of chip designer. They, as we shouted out before the They, as we shouted out before the Cerebras IPO, before the Cerebras IPO, we mentioned that this was a stock to avoid in a dangerous dangerous, uh in my

[02:33] opinion, a rip-off. And the stock has been straight down. Somebody who invested $100 in this company at uh 386, so all the way up at 386, if somebody invested $100 into the company then, uh

[02:48] they'd have about $57 left today. So, it's been a little ouchy right over Uh so, let's focus on Meta and what's going on with CoreWeave uh and Enfabrica

[03:00] and then, of course, Meta. So, the first thing that we should take a peek at is what Mark Zuckerberg said in relation to this Bloomberg article that broke this morning about hey, now they're selling hardware. So, Bloomberg accurately

[03:15] quoted the following. Uh we've got this up on the app.mevin page, which you can sign in to and see as well. Uh it's also in our Meet Kevin app. But, uh what you've got here is you've got a quote that Mark Zuckerberg, and this quote is

[03:31] actually from the annual shareholder meeting. It's not in one of the earnings shareholder meeting. And I listened to the whole annual shareholder meeting, and the quote is that Mark Zuckerberg, uh CEO of Meta, had signaled to

[03:44] investors that he would be open to selling excess computing infrastructure or AI service where customers would pay for AI usage, measured in tokens, uh and he said in response to that, "Quote, it's definitely on the table."

[03:58] Zuckerberg said during the call with shareholders in May. "Almost every week, there are different companies that come to us from the outside asking us to both stand up an API service or asking if we have the compute that they could buy

[04:12] from us at a premium to what they bought at. We haven't done that yet because we think we have a use for the compute. But obviously, if we get to the point where we feel we have overbuilt, then this is

[04:26] an option that we have. And now, before I go further here, a lot of people have been stopping at that quote. And I think that's a little bit of a failure because the quote itself says, "Hey, yeah, if we

[04:40] we've overbuilt, then we'll just sell API services." Okay, so is this then a red flag of the whole hardware top? I mean, Amazon, Microsoft, Alphabet, Meta, and Oracle, you know, the big five basically

[04:55] spenders of CapEx here are guiding that they're going to spend $715 billion in 2026. That's up 70% from the $410 billion that was spent in 2025, which was already a lot. You know, that's almost a double.

[05:10] And now you've got Bank of America and Evercore ISI suggesting that we might get all the way up to $1 trillion in 2027. Now obviously, the $1 trillion uh uh you know, survey suggests a growth rate that's slower than 70% going from

[05:26] rate that's slower than 70% going from 410 to 715, but still, 1,000 / 715 is 40% growth on a huge number. the guidance that we're getting from companies is that CapEx spending has is

[05:39] companies is that CapEx spending has is is really nowhere near peak. And when we companies, we're not seeing any kind of peaking out on capex intentions yet. That would be a really big red flag, right? If Meta turned around and then

[05:52] started guiding lower capex spend, that would be the red flag. Now, obviously some people are wondering, well, is this the beginning, right? All it's really going to take is one of these guys, like a Meta, to go, all right, that's it.

[06:07] We've peaked out on capex spend. We're now going to start turning the dial the other way, and we're instead of growing our capex spend, we're going to start shrinking our capex spend. As soon as one of them does it, their stock

[06:20] skyrockets, and then all of a sudden all the others get the idea, you know what? the others get the idea, you know what? We should also be a little more careful with our money. Let's turn down the spend. Then they all do it. Then what's

[06:32] the next domino to go? Well, then it's the Nvidia, the AMD, and the entire hardware stack, and the memory stack that tanks. The problem is when those puppies tank, you have leveraged ETFs at now the

[06:46] highest levels that we've essentially ever seen, well, not even essentially, that we have ever seen. They're at the highest levels of leveraged ETF uses we have ever seen right now, and what's worse is they keep coming out with more

[07:00] leveraged ETFs. It's like, here's the chart on it on screen. But it's like chart on it on screen. But it's like basically a daily basis that we come out with new leveraged ETFs. You can see this chart over at app.meKevin.com. Just

[07:13] the uh uh Apple or Android app store. If you don't have a login yet, again, just download the app, Apple or Android app store. The easiest way is just sign up with Gmail or or Apple login or

[07:28] remember. Uh you can also make an account if you want. But anyway, um the leveraged ETF asset under management sector has been booming. Uh it's, you know, up 5x basically since uh the end of 2020, and it's skyrocketing. And

[07:42] what's happening is almost on a daily basis you're getting more and more leveraged ETFs that are being created around new products. Quantinium, which was a recent IPO, a recent spin-off, literally already has its own 2x ETF.

[07:56] All that crap is great when the market's going up. But let's just say there's a reason why the SEC prevented uh ETF filers from making 5x ETFs because all those 5x ETFs had they actually gotten approved would have

[08:11] already gotten liquidated and sent to zero. That's not oh, we're closing it down and giving everybody their money and their profits back. That's it's at zero. That would have all happened already if we had 5x ETFs just when the

[08:24] Iran war was going on. Cuz it only takes a 20% straight decline to make that happen, right? Which obviously happens. So, you know, and that's why people are looking at this because okay, well, if the hardware

[08:38] stock sector goes down, then hardware companies, which they're good at this, you know, even the CEO of Marvell will tell you this. Hardware companies, semiconductor companies are good at adjusting their workforce

[08:51] Why would he say that? Well, he said that in an interview like a year ago. He said that because semiconductor companies are used to the boom and bust. Remember, I made like a 10x on my Nvidia

[09:04] shares. I sold my Nvidia shares probably at an average of about 195, which is had some volatility. It's gone down to like 165. It's gone up a little bit like But um the point is

[09:17] >> [laughter] >> partly because I'm like yeah, bit over here. There's some other opportunities to go make money on. But the point is the whole stack rolls over fast. Uh and

[09:32] lay people off when they're in a recession. So, that's why people are worried about is this meta issue a signal? Now, again, I want to go back to the second part of

[09:44] the phrase that most people are leaving out right now, going back to Mark Zuckerberg quote. The Mark Zuckerberg quote is the It is that and that is partially what gives us

[09:57] confidence in investing uh in in in investing in building this Okay. So, in other words, Mark Zuckerberg is actually telling us, "Hey, we don't necessarily think

[10:12] that we need to sell API services because we've overbuilt. We're saying we have the option to always sell extra compute and because of because we are able to sell compute, we can just keep building."

[10:24] In some regard, they might be following SpaceX's strategy. X, we know, went into massive debt to build out Colossus and then Colossus 2 and build out these data centers, picking up all these H100s from Nvidia.

[10:39] In fairness, that's not the greatest technology anymore. You know, you've got Blackwills, now you've got uh Rubin the Rubin architecture coming up, Vera Rubin The Nvidia chips just keep getting better and better and better and

[10:51] eventually those H100s will just not be as desirable as the latest and greatest inefficient. But, you can still rent those out for a premium. And SpaceX is doing exactly that by

[11:05] calling up contracts at companies, whether it's Google or Anthropic or whether it's Google or Anthropic or whatever, and selling compute capacity. I think Facebook is realizing, "Huh, if SpaceX

[11:19] uh is able to sell compute and then get analysts to say, 'Oh, wow, we're going to price in 25 or 50 billion dollars of revenue for SpaceX have that much business otherwise, revenue otherwise.

[11:33] this potential future revenue. Well, then we may as well sell a little bit as well because it'll stop the stock from falling. And ironically, if we can maybe unironically, if we can stop the stock from falling, then we can raise money,

[11:49] whether it's debt or bonds or whatever, and turn around and and turn around and spend more on AI because, quote, we could always rent it out. Now, I'm not the biggest believer in that. I

[12:03] don't really love the idea of owning all the infrastructure. I like Meta because they make a lot of money. The big thing about Meta to me is they make a ton of that. I mean, when we look at the

[12:18] fundamentals of for this company, they're really good. I have their last financials up. Uh let me put those on screen right here. Look at this. This is screen right here. Look at this. This is revenue at Meta before selling any API

[12:31] access to their infrastructure. Before selling any of it, this is a company selling any of it, this is a company that generated $56 billion in revenue. And they literally grew that from $42 billion last year. I'm like, "Oh,

[12:45] billion last year. I'm like, "Oh, holy smokes, dude. How How do you grow holy smokes, dude. How How do you grow $42 billion of revenue to 56? Damn. You got people stuck on Instagram." And the reality is, the reason Facebook

[13:00] does so well, I call it Facebook still. I feel like a boomer. I don't call it to boomers. I'm just saying like I'm stuck in what I'm used to. I still call X Twitter, right? Uh but um what's remarkable here is they're

[13:15] getting more efficient at selling and placing advertisements on Instagram, on Facebook, Facebook Marketplace. People are using this like crazy. Uh and and people find good deals on it. You know, my grandmother, who's like a

[13:29] yard sale hunter, she loves using Facebook Marketplace. I know Graham Stephan talks all the time about using Facebook Marketplace. And we know he's >> [laughter] >> Uh, I mean, good for him, right? He

[13:41] But the point is, this company milks money. And this is before API revenue. So, look at this, their cost of revenue is about 34%, which means they've got margins of about 65.3% on a gross level. And if I look at

[13:58] their, uh, net margins, I'm just going to go all the way down to income, uh, and net income. And this isn't even considering depreciation, right? Like that hardware depreciation. I'm just going to assume

[14:13] fine. In real estate, we usually add that depreciation back in because, I mean, come on, if you're in real estate, you know it's bullcrap. Uh, but they're bringing 47.5% to the bottom line. This company prints money. This is printing

[14:27] money. To bring 47% to the bottom line, to increase your bottom line by whatever it was, 30 or your income from operations by 35%. You increased your operations by 35%. You increased your bottom line here by 16644,

[14:40] 60.8%. These These are phenomenal numbers. I mean, their marketing and have to pitch their product because they know their product is so good, people just go to Facebook and Instagram or the other threads or whatever other, uh,

[14:56] pretty good at video now. They're trying to take, you know, some of the Google YouTube playbook and, I mean, I'm not streaming on Facebook right now or YouTube because I feel like that's my

[15:09] home, but, you know, maybe I got to wake up and get some more videos up on on on But, uh, really this is This is quite phenomenal. Their EPS is exploding, and as a result of that, their, uh, valuation is actually quite cheap. We'll

[15:22] and then I want to break down is this a company that is actually burning company that is actually burning you know, capital because of token usage and therefore they're freaking out and turning around needing to sell capex. So

[15:39] we'll explain that in just a moment. Let's just do a valuation on them to see where they sit. Now how I like to do valuations is I like to look at what the Wall Street forecasts are for growth and then I compare to my own sort of

[15:53] and then I compare to my own sort of forecasts for growth. So for meta probably because of the interest release expenses they actually have minus 4% on growth for next year and then the year after that Wall Street is pricing in

[16:05] after that Wall Street is pricing in 21.2% 14.4% plus 16.8%. If I add that 21.2% 14.4% plus 16.8%. If I add that all together, I get 48.3

[16:20] then if I look at their their growth this year is expected to be 36%. If I this year is expected to be 36%. If I look at their EPS, I've got them at 3995 and the stock's at about 610. So 610 divided by this equals

[16:35] divided by this equals 15. So they're selling for about 15.27 times earnings. Divide that by call it 12.1 times growth. We've got a PEG ratio 12.1 times growth. We've got a PEG ratio here of about 1.26.

[16:48] Honestly, I think this growth estimate is low. They should probably be closer to about I I'd say at least 15%. it. And if I run this at

[17:03] And if I run this at 3995 today's EPS and I run them at 15% growth times 2.69, I've got this at about a $1600 stock at a fair price. Now the reason it's gotten wrecked is

[17:17] heavily due to their CapEx spend. People think they're spending like drunken sailors. You've got a lot of employees that are being compensated by stock comp remained high while at the same time they're no longer doing stock buybacks.

[17:31] they're no longer doing stock buybacks. Usually share compensation, right? Look at this. Uh right here. Share compensation, $6 Uh right here. Share compensation, $6 billion of stock comp in 3 months. Now

[17:43] They got $32 billion of cash from operations over here. They're burning about 19 on CapEx of that. So a little more than half of that. Depreciation is about $6 billion as well. That's what I would, you know, ordinarily add back in

[17:56] in a in a real estate company. But that's a lot of stock comp. And the this. Repurchases of common stock, nothing. See that? Last year this time,

[18:08] Now that'll come back when they slow down their CapEx, which I'm looking forward to. But their repurchases right now, nothing. So in other words, you're doling out a lot of um

[18:21] um stock to existing employees or new employees or whatever, and they might be selling that stock so they could go buy a boat or a plane or a car, I don't know, whatever people want to buy. And

[18:34] they're not doing any buybacks to offset that selling pressure. That's part of the problem that you get. Okay, well, that's not great. So what is our stock AI has sort of an automated version of what

[18:49] Kevin thinks for this company. And it doesn't have my updated 15% in there. It just basically has the actual fundamentals of math just algorithmically. So we actually made this. This is all part of the

[19:04] Meet Kevin Alpha Reinvest sort of platform, if you want to call it that. But what's really kind of cool about our stock AI product is you can type in a ticker over here. Since we're talking about Meta, let's type in Meta.

[19:18] And I can grab the Meta details, and we can get a fair value estimate for the company. You can get my notes on the company when we've manually done notes, but you also get uh the valuation matrices, and uh and

[19:32] then of course red flags, green flags, what's changed quarter over quarter. There's a lot of cool stuff in this. But anyway, let's stick for a moment to But anyway, let's stick for a moment to uh Mark. And what his commentary is

[19:44] uh this spending. And do we actually think this is sort of like uh you know, a give up? Is he basically saying that's it, we're done, uh the hard work compounding is over?

[19:57] My opinion? No. And I'm going to show you why. Here. Mark Zuck says that and he was asked right before this question. He This was This is from the earnings call. He was specifically asked

[20:11] about 2027 CAPEX going up. Meta says it's too early to say, but Meta argues a loop argument. And they basically imply lumpy training. In other words, I'll read it to you. We have our next set of more advanced models in training right

[20:25] now, and that will, I think, just continue. I mean, that's a loop. I don't anytime soon. We're going to have teams that just consistently, that are consistently focused on training more intelligent and more capable models in

[20:38] things that came out of the earnings call is they talk about how profitable artificial intelligence is for them internally, which is exactly the everybody's talking about on, you know, Twitter right now, which is, "Oh, you

[20:53] know, they spend a ton of money on tokens, maybe $50,000 per employee, or they spend, you know, whatever, X number of tokens." we know token usage is getting more efficient over time. Not only is it

[21:07] get more efficient. But, I thought this was very useful. Their CFO actually says, "And now there are very compelling internal uses for their own artificial own models, but I suspect their advertising buildout. I like Meta as an

[21:23] cheap advertiser that makes a lot of money, and I kind of think they're sort frankly, the stock is going down because they sort of spend a lot on CapEx, but they don't really have a leading or bleeding edge model to show for it,

[21:37] So, on what's also very interesting what's also very interesting is some people call this a mid-cycle move. Now, this is where sort of the theory comes in a little bit. Early

[21:50] cycle, basically, everybody rushes to blow money and spend money. We'll pay want it. We'll pay whatever price for the GPUs, we just want it. But, over time, companies realize, "Huh, you know, we have to get a little bit

[22:04] more efficient. We And this happens mid-cycle. Let's sell some of the capacity we're not using." Like what Meta is here with the API sales. That's exactly how Amazon started back in 2000.

[22:18] During, you know, before the dot-com bubble, Amazon started by saying, "Hey, we're going to sell the unused data center capacity we And we're going to sell it." That's how AWS started in 2000. Kind of incredible.

[22:33] here to just sort of catch up with what other AI companies are already doing anyway. To me, a peak signal in hardware is uh, you know, when companies start guiding for cuts. We're not seeing that.

[22:47] Uh, after that, you would see CPU guy or CPU and GPU guidance shrinking. We're not seeing that. We would see backlog start falling. Google Cloud has over $460 billion of backlog. Microsoft over $80 billion of a backlog.

[23:00] start of the monetization cycle and a little bit of a stock pump, in fairness. Now, I don't blame anybody pumping. I like pumping, too. I mean, my favorite thing to pump right now, or the two things I pump regularly, the Meet Kevin

[23:14] membership, where you get all nine courses, every trade alert, my top 15 or over the next 10 years, every private live stream, the Alpha reports, people say could be tax membership. Remember, you've got coupon code Pope that expires

[23:27] What a lot of people are doing is they are bundling that up with the uh real are bundling that up with the uh real estate uh homes AI service, uh which you can get at househack.com. And I actually just did a demo video of that on the

[23:41] channel. It's the Great Housing Market Reset video, and in it I go through not only a housing market reset update, but I also give a big uh demo of the product, how it works, how you could use it, and a house hack update, which is

[23:55] lot of people have been asking for a house hack update, so there's one. You can also get the house hack updates inside the Meet Kevin app, uh which you can have on your app on your um you know, Apple or Android phone or the

[24:09] literally click over here, investor updates, and then we have a whole investor update section for house hack and reinvest, which is kind of exciting. But anyway, going back to Meta for a moment, uh I do think that there's a

[24:22] chance this move from Meta could reduce some panic buying, you know, this sort of like buying GPUs at any price. Uh I also think you know, Zuck is pretty clear here that they have growth levers in this. Mark Zuckerberg thinks that

[24:35] agents will require massive amounts of compute. In his annual shareholder letter, he said that, in his earnings calls, he talks about it. He sees four major opportunities, which he always talks about, make the apps and

[24:49] with agents, business compute with agents, and then hardware, like the Uh they expect that they'll be able to start charging premium subscriptions for that kind of stuff in the future. I'm

[25:03] glasses. I've got like a bunch of them all over the place. This is one of them. clean them up a little bit. They're a little dirty. They still got some snow Uh, but apparently three times as many people are now using the hardware

[25:18] glasses uh, year-over-year. A lot of pilots. I remember I was one of the first pilots using the Meta glasses. And I thought it was so freaking cool uh, flying with those suckers. Because it's great. It's like, "Oh, I'm coming in for

[25:30] Boop boop. All right. The Now I'm recording my landing. All right. >> [laughter] >> It's awesome. Um, You got to have your hand on the throttle. Come on, everybody knows that.

[25:45] yoke. If I only do this, it doesn't look right. yoke. Anyway, okay. So, uh, there's there's really a flywheel here that Mark

[25:57] Zuckerberg is trying to build, you know, better hardware, scaling the compute, monetizing it. I don't think they're really monetizing it yet because I mean, we don't really see these personal agent premiums or plans or

[26:11] never get to it. That's just Mark's vision right now, and maybe it'll die a but anyway, that's sort of the idea of all this. So, let's just see what some uh, are saying in the chat here. Thanks so much for being here. Um,

[26:26] remember to use coupon code Pope. I I still love that I got a nod from the Pope. Like I feel giddy inside for that. Uh, let's see here. Uh, somebody here numbers that you don't believe will continue to perform as well or change

[26:39] That's the thing. That's the cool thing about growth rates is, you know, Wall Street can look at it in an on average estimate a 12% growth rate. I can look at it and estimate a 15% growth rate. And you know, you can look at it and say

[26:54] I'm only going to estimate a 5% growth rate. Uh somebody here says, "I just use historical PE." I think that's a really big mistake because, you know, when you use historical PE, like look at the historical PE on like Micron for example

[27:08] from a year or two ago. It would look like you're paying a 2,000 PE ratio for the company. You can't use that, you know? Like you have to like you have to look at where the company is in the growth cycle and then go, "We're not

[27:22] That's in the past. That already going to happen in the future." Now, obviously, you know, if a company has then they're like, "Next year we're going to grow 30%." It should raise an

[27:37] eyebrow, you know, some sus, right? Uh but uh but anyway, let's see what else you got here. Marvell, Axon, AppLovin have been printing tendies. Hell yeah. prices. Hyperscalers are creating a monopoly on compute. Nobody else can

[27:50] that's a fair point as well. Hardware's Nah. Yeah, I I mean, I think what you're saying is very interesting. I think they do want a monopoly on compute. And it's

[28:03] more like an oligopoly, oligopoly. Somebody else uh here asked, "What does this mean for Marvell? Shrink the position?" No, man. personalized financial advice, but if you want to get jazzed and snazzed on

[28:19] Marvell, okay? If you want that. Maybe you don't want that. But if you want to get jazzed and snazzed in Marvell, and it's going to make you want more of it, I highly recommend you watch this video

[28:32] to the end. It's this one right here. Elon loses it, leaked SpaceX's new AI device, okay? Watch that video, and at the end, you're actually going to see talk about Where is it? I just saw it on the thumbnail. Marvell is in here. Like

[28:48] videos of Jensen? There, Jensen, Nvidia, and Marvell. Yeah. Look, this is like inception. You're watching, me watching me. >> [laughter]

[29:00] watch that. If you want to get jazzed on Marvell, that kind of makes me like, "Oh for it." Of course, they're going to get wrecked when hardware plummets. Dude, the entire economy is going to get destroyed

[29:13] if the hardware market plummets. We all know that. Everything is cooked and we're going into a recession if hardware rolls over. Uh but, that's why you just Like, I don't want to be here and say, "Oh, it's going to happen tomorrow." cuz

[29:26] tomorrow. The indicators aren't here. Uh and you know, you you want to invest in Train America, but you also don't want to get blown out, you know? You don't on your debt and invest cautiously. That's uh that's my take. All right,

[29:41] being here. We will see you in the next video. Goodbye and good luck out there. code Pope. And when you check out, you can bundle up and get the Reinvest AI. can bundle up and get the Reinvest AI. Goodbye and good luck.