저자: MeetKevin · 외국인
요약 보기
마이크론 실적 발표 전 분석. 마이크론은 가격 상승(DRAM 약 65% QoQ)이 실적을 주도하나 공급 확대(싱가포르 HBM, 보이시 팹 2027년 가동)와 수요 둔화 위험으로 2027년 이후 가격 파워 약화 우려 제시. 밸류에이션은 2026년 EPS 59달러 기준 PER 약 19배로 저평가 주장, 장기 보유보다 1년 내 단기 대응 선호. 영상 말미 Wednesday 실적 이후 후속 방송 예고.
다듬은 글
All right, hey everyone. Welcome on back. We got to talk Micron because Micron earnings are coming up on Wednesday and frankly they could make or break the market. So, not everybody here might be invested in Micron or maybe you are and you've been absolutely killing it on the investment. Uh in this video we're going to be pretty neutral on the actual stock itself.
I'm just going to give you some facts here and we're going to use those facts to understand what's going on in the broader industry and how this could affect the broader, well, frankly hardware narrative given that hardware is clearly what's been running since uh we had our original beginning of April ceasefire, if you want to call it that.
Uh and then of course uh since then we've been looking for a broadening out. Uh we had these fits and starts of software a couple times. Those haven't really gone anywhere. Uh we had finance last week that did really well, specifically Robinhood. But consistently it's hardware stocks that keep winning, uh including Micron uh and uh SanDisk of course, but even outside of these memory plays, you've got companies like Marvell, which was one of our course member favorites.
Remember you can always join over at meetkevin.com. Use that coupon code Pope and you can see why on uh on uh Instagram or on uh on Twitter. But anyway, uh Marvell's done fantastically. I mean, we're at all-time highs now at 310 and I mean, we were even higher than that intraday uh last day the market was open.
But for now, let's focus on Micron and talk about some of these earnings expectations. What's going on there? So, uh first of all, we're expecting to beat the guidance that Micron gave by about 18%. The consensus estimate is only a beat of about 13%, but they'll probably beat, which is good. You want that.
You actually don't want them to miss on any component at all because then you might end up pulling off a Broadcom. Broadcom had pretty much across the board beats, but the timing of some of their custom chip revenue, Broadcom does a lot of ASICs as well like Marvell. The timing of some of their custom chip chip revenue kind of got delayed and so one of their their their next forecast quarter revenue was a slight miss compared to expectations and the stock still hasn't recovered.
It was down like 20% after earnings and you know now it's still down about 14% from their earnings. And so even though their earnings were pretty dang freaking good and the company's booming. Yeah, honestly for long-term buyers it created a dip opportunity for Broadcom. Yeah, even a slight miss can really tank some of these really big momentum movers if you will.
It doesn't take a lot to move these so to speak. So we're expecting a beat on top of expectations that are already expecting a beat. And the biggest driver that you have at Micron right now is really price realization and I think that's really important to understand what you're looking at an investment into Micron.
So I'm going to read you some notes here directly from their last quarterly report. They have an increase of year-over-year increase of DRAM supply of 40% and NAND supply of 30% or shipments I should say rather than supply. That's a pretty good year-over-year increase in shipments, right? But price is actually going up at a percentage of about 115% on DRAM and about 100% on NAND.
So in other words, you are growing price way faster than you are actually growing shipments, which is very common during a shortage environment. Now something to think about if you hear NAND and DRAM, I'm not going to go super into the weeds here. I'm going to give you a very very basic difference of the two.
The way I like to think of the two is I like to think of NAND maintains memory when the chips are turned off. So the power is turned off. When is that useful? A lot of agentic AI applications. I don't want to have to reload all of my agentic AI contacts every time the darn thing respawns. I should say respawn.
Restarts. I'm in like a video game mindset even though I'm here in like a real life MMO. A lot of NPCs out here. I shouldn't say that. But anyway, DRAM is really, really powerful all the time when it's on, but it basically wipes your memory when it's off. That's a very, very simplified explanation of the difference.
And of course they've got four core business units. You've got the core memory business unit. That's for like hyperscale cloud. Obviously, that's where we're getting big spend right now at data centers for Micron. Then of course you've got sort of the original data center business. Think storage, think the original components inside of Dell racks and storing information.
Google Drive, Amazon Cloud, right? Like that sort of core original data center stuff. And then you've got mobile, automotive, and industrial and consumer sectors. So think mobile phones, think, you know, memory chips that are going into your dishwasher or your you know, your refrigerator, your washing machine, or into your cars, you know, your seat settings, like basic things like that.
Really basic chips. Those sectors by the way, the last two, automotive, industrial, consumer, and these are bundled into two. I'm listing them out into a little bit more. And mobile, those are actually declining, which is very interesting. So even though prices are going up in all of the sectors, shipments are down in mobile and enterprise, those other categories.
So it's really just the AI narrative that's driving chip prices right now. And so anything that would soften the hope that this exponential growth of the chart of Micron is going to keep going is going to be a really big risk factor for the stock, especially with how high it's run in the near term. Now, the thing is, even though at $1,100, it seems like it's run up an insane amount, because of the Wall Street forecast for this company, it's actually cheap.
Like UBS prices it in as having a potential double. I don't actually think they're wrong, which is crazy to say, but right now, the Cape I have for them is 1.03, which means they could easily justify a valuation that is probably closer to 2 and 1/2 times this, somewhere closer to $2,500 a share, but there's going to be a lot of volatility between here and there, both to the upside and to the downside is my expectation.
Now, that's pretty incredible, and I'll give you the components as to what makes up that, so you could understand why that, you know, we we would come to such a valuation. It has to do with an expected earnings per share at the end of this year of $59 per share, which originally, when we look at that and we divide their current price by 59, we're like, okay, wait a minute.
Oh, that's actually not that high. That's only 19.2 times earnings. Yeah, like these people are printing money. Even though that stick is, you know, the chart is hockey sticked so much, because they are printing money, they are printing so much they are only selling for 19.2 times earnings. On top of that, their growth is expected to be about 18% per year over the next 4 years.
So, that gets you pretty close to about a one price to earnings growth ratio. Now, if you look at like trailing, the numbers are going to be different, but when you look at these forward projections, yeah, the projections actually suggest the stock is cheap. And this is why the most sensitive thing you could look at would be anything that would hurt those projections.
Now, I just want to shout something out really quickly that's a totally free feature. It's called our Alpha Wire service. If you haven't actually checked it out yet. Uh we just released it last week and it's pretty amazing. Basically what it is, it's uh think about it like a um uh an anchor over at CNBC or Bloomberg.
You know how they're like, "Oh, this is just coming across the wire. This is just coming across the tape." You can get that for free now inside of the Meet Kevin app. Uh so, if you download the Meet Kevin app in the Apple or Android app store, you can utilize this. If you sign up there, you can also then eventually use the desktop version if you want.
Uh you know, you just have to sign up in the app first. Just app.meetkevin.com is the desktop version, but you need to make that account first uh inside the app and you can use the wire service on desktop, on Android, on Apple. And it's totally free for now. Uh but it's kind of cool because uh you know, you get these little breaking updates like Trump heads to Camp David as Iran talks falter or whatever, right?
And those will come through. This is brand new. It's super in beta. We just released this in the Meet Kevin app. But I think it's really cool because you know, I also can pin uh wire alerts that come through and then I'll put like sort of the Meet Kevin commentary under it and that's really clearly outlined.
So, check that out. It's right under the Alpha wire tab and you'll see inside the Meet Kevin app we actually have uh a little terminal button where you could switch between housing and stock. It's because we've got like the stock AI tools now that we've custom built. We've got the real estate AI tool, uh the valuation AI service coming out at the end of the month, which is really exciting.
We're really just like cranking software right now. And we we are jumping up and down about what we're doing and and we're providing this for free for now. So, we're very excited about it. You can use it at least for the next few many months uh uh totally for free. So, I I think get your use out of it.
Uh why not? Give it a shot. Give us some feedback. Let us know what you think. But anyway, let's focus some more on Micron. So, those are the EPS projections and the growth projections. Any kind of red flag here that could tank the stock would be any kind of slowing in average selling prices, any kind of slowing in order growth, obviously miss targets.
But again, to me this is a COVID era toilet paper seller. They're selling toilet paper when everybody wants toilet paper, there's not enough toilet paper. This is not trying to be an insult to the company. The company knows this themselves. They literally warned in their filings that most of their cash flow right now is coming from price realization, not because they're delivering more product.
They realize that. Like if you literally look at their quarter-over-quarter numbers, you'll see how extreme it is. So, I'll go over here to quarter-over-quarter. I wish I could share the screen screen in my studio and just show it to you. So, if if if you're a course member, you could see it. It's under the stock tab.
All of these are All this data is under the stock tab. But, what I want you to think about is this very next quarter. Oh, I don't even know if I screenshotted it. Let me screenshot it and make sure it goes in there. But anyway, it's from the quarterly report. Ah, yeah, here it is. Okay, great. Oh, okay, perfect.
Yep, right here. So, we've got sales of DRAM products increased 74%, but price increased about 65%. That means that's quarter-over-quarter. So, in a 3-month span, they literally hiked prices about 65% in a span of 3 months. This is an insane shortage. Part of that is what we saw after Claude co- co-work really went crazy.
I shouldn't say Claude co-work. Whatever, you know what I mean. Claude code, there we go. After Claude code really went crazy in Q1, prices for everything, HBMs, you know, high bandwidth memory, but also H100s from Nvidia had this massive surge in Q1. But, those prices have already started to settle.
And that's one of the big concerns we're worried about here for Micron is we're already tracking H100 prices, and they're still elevated, but they've come way off the peak that we saw in the first quarter. And so, some of this price growth that they're talking about here might not last. But, I mean, listen to this.
They raised prices on DRAM, remember the one that works really well all the time when it's powered on, but it loses its memory when it's turned off. Prices rose 65% and their actual shipments only grew about 7 to 9-ish percent. So, that's still good. You know, shipments are still growing somewhere around, you know, 32 to 36%.
But, most of their money is being driven by price because they can. It's sort of like, "Wow, okay, let's try raising the price. Wow, people are still paying it. Let's try raising the price. Wow, people are still paying it, right?" That's what they're doing and they're milking that right now. Now, that does not last forever.
We know that. In fact, if, you know, PP is too strong for too long, you know, some say 4 hours, you need to go visit the doc. Okay, anyway. Uh the same is happening actually on NAND products. NAND products up 82% but 75% of that was because of price. So, yeah, quantity is going up, but most of the growth is frankly price.
Okay, yeah, so I did have this correct uh all um already updated on the the stock uh terminal side. Okay, good. So, uh let's now keep going on to the the the broader risk factors here. So, any kind of cost reductions at companies or increases in supply are the big risk factors for a company like this.
So, some potential headwinds are going to include the fact that there are efforts now at Meta, Microsoft, and companies like Walmart to decrease this token maxing culture we've gotten into. Meta last week told employees that it would soon limit AI use after seeing an exponential increase in token maxing.
I mean, in fairness, they were a little loony toony. They went for like, "Hey, let's have leaderboards and see how many tokens people are using and uh you know, reward people based on token maxing." And so, people were basically looping agents doing essentially very little in the way of productive work, uh but maximizing their token usage.
That's pretty dumb. So, Meta's killing that. Walmart is killing that. Microsoft is now considering moving from opening eye uh and Claude, mostly like a ChatGPT-4.0 level for Microsoft co-work. and they're considering moving to DeepSeek V4, which is like a huge insult because now it means we're going from American LLM providers to sort of like a copycat and mega efficiency version DeepSeek, which you know, we highly expect DeepSeek basically trained their versions by just training off of the outputs of OpenAI, so kind of bypassing all of the training that OpenAI did, which is kind of brilliant but also very evil.
But when you switch models, you know, cost plummet. If you use Sonnet, you literally save about 95% of the cost that Opus costs over a claw. You know, so you use claw Claude Opus, you know, it costs you 20X what Sonnet costs. And if you use Kimi, which is a product from Moonshot AI, which is another Chinese startup, that costs you another 1/20th.
So another 95% cost reduction. So cost reduction is is really a big key here. The more that we see, the less we end up seeing in the way of pricing power at chip manufacturers. But that's all a downstream effect, right? So that's why we're so focused on their guidance because we really think they're going to be able to guide strong for a probably at least another year.
Bloomberg Intelligence thinks they're going to have pricing power through the end of the decade. I kind of doubt that. They say they'll be limited supply between now and likely 2029 to 2030. And they expect higher pricing throughout 2027. I agree. Throughout 2027, like getting into next year, I agree with that.
All the way to the end of the decade, I do not agree with that. And I'll explain why I don't agree with that, and it has to do with the supply expansion in just a moment. Uh but efficiency in models slowly, even though it's model use is exploding, token use on a second derivative we already know is declining, right?
So token use, if you look at the Google IO event, we can see we went from 2024 to 2025, 50X increase in token use. Now we're at about a 6.9X increase in token use, and that'll probably keep declining in rate of growth. So, as we then get away from token maxing, we'll see more of those declines, and eventually these this pricing power will fade over our chip companies.
That's a warning for all chip companies, we know that. But for right now, things are looking really good, and the company is trying to grow their capacity to because they're a big manufacturer of their own chips. While they use uh third parties, they do manufacture a lot right now. They're paying down debt.
Their balance sheet is really good. They've got plenty of cash to pay off all of their debt right now, uh and they're actively paying down debt, which I think is a very smart move, and I think they know that. Uh and even though they're paying down debt, they're also still blowing money on CapEx, expected to be 25 billion for the year, but they've also already spent a over 11 billion for the first quarter.
So, you know, I think that they spent a lot already. But, you know, almost half of their year's budget is already spent in the first quarter, but yeah, maybe that's just timing, right? But anyway, part of the reason they're expanding supply so much, and this is why I don't think they'll make it to the end of the decade with this sort of pricing power, is they're getting like 5 and 1/2 billion dollars of subsidies from New York State, and on top of that, they get CHIPS Act 35% subsidies from the Biden-era CHIPS Act.
Now, Donald Trump has totally criticized the CHIP Act CHIPS Act, but in the background, they are actively running an office to manage who gets dolled out tax credits through the CHIPS Act. So, publicly, Trump bashes anything other presidents do, but in the background, he's using it in a classic Trumpian style to pick winners.
We we kind of know this is how Trump operates. It's not a dish, it's just literally calling it like it is, you know? Uh it's it's almost sort of like saying, "Hey, can we get through a Meet Kevin video without Here's a coupon over at Meet Kevin." Now, in fairness, the price goes up over time. We've increased the price again, and at the end of this month, we're going to have another big price increase because we're coming out with even more stuff.
Uh and and we are spending a lot on R&D, so we're doing a lot to really give back to the community. And uh that actually enables for those of you don't want to pay a dime, you using this AlphaWire service. I highly recommend you check it out. Uh the AlphaWire service is pretty badass. Uh and let us know in the comments what you think about it.
If you don't like it, delete the app and uninstall it. But honestly, uh like I I wouldn't shill something free uh like this if I didn't think it was great and you would be like, "Oh man, Kevin, yes. This is actually good. Thank you." Like really. Uh okay, so uh here's the problem with the expansion of supply.
I'll go through this pretty quickly here. Singapore high bandwidth memory expansion uh coming online in calendar 2027. This is basically them saying they're meaningfully expanding their packaging for chips chip assembly uh for high bandwidth memory in 27. So we're going to have a big increase in in their Singapore plant for HBM.
That increases supply. I'll talk Oh wow, I just zoom I didn't realize I zoomed out all the way. Oh, that's interesting. I can I can actually control the zoom from my phone. That's kind of cool. Or my watch. That's kind of cool. But the downside of that is if you accidentally bump into it, you make a mess of your your angle.
Hopefully you didn't see uh you know, up my pants here. What pants? Uh anyway, so uh it works out. Anyway, um in September of 2022, they started their Boise, Idaho fab. Uh and they actually broke ground in 2023. Guess when that comes online? Mid-2027. Took about four years to develop. So that means you're going to have the Boise, Idaho fab come online at the same time as their Singapore fab is expanding uh production.
That's a lot more supply. And mind you, this is just Micron, right? Like the leader of memory production is actually Samsung. And then you've got SK Hynix. Samsung sits about 38% market share, SK Hynix at 29%, Micron's actually only at about 22%. They're something to keep in mind. They're one of the smaller.
Uh now, if you look at uh their additional fab development, they have a second fab in Idaho that they're starting construction on this year, like right now, and they expect that to be online by the end of 2028. So, between the middle of 2027 and the 18 months thereafter, so middle of '27 to the end of '28, you're going to have like three fabs just from Micron coming online to start trying to realize this this, you know, price increase.
Now, I think that there's an indication they're starting to try to rush these plants. You know, their first one took about 4 years. This next one's 2 and 1/2 years of development. Maybe it's smaller, but to me, like if I were them, I'd get it. Like, "Hey, prices are really hot right now. Let's try to build these as soon as we can so we could print." Like, make hay when the sun is shining.
That's why they're paying off debt. They're really smart. At the same time they're paying off debt, they're using cash flow to build the machine. Very, very smart. Now, in the near term, Apple did warn uh that memory prices are going to uh increase the cost of Apple products, and that's going to be unavoidable.
I agree. For the next year, that's probably unavoidable. That could lead sales to fall at Apple, which in my opinion will drop the stock, which actually then becomes a buying opportunity for Apple. I think Apple is roughly fairly valued right now. The upside isn't that great. I'd love for them to get cheaper.
Honestly, I would love an opportunity to get into Apple before they actually come out with an AI product. I'd like to be in it before they figure it out. Maybe they will never figure it out. I suppose that's a risk, but I'm not in it right now. But anyway, so you've got these expansions of these facilities uh to keep in mind.
Their G&A isn't really moving, which is, you know, just indicates they're really able to scale what they have right now. They do warn uh that uh cash from operations is highly dependent on selling prices, which are extremely volatile. We know that. That's why they're on the moon cycle right now because everything's just skyrocketing.
But the durableness of those price increases will almost certainly fade. It's just a matter of like, does supply ramp first to reduce that pricing, or does demand slow down first or worse, both. When both supply increases and demand falls, this stock will absolutely tank really fast. It's going to suck.
Like I would not want to hold this stock for the next 10 years. I would want to hold this stock potentially for the next year, especially if it tanks after earnings. They've got 53,000 employees, which is actually pretty impressive, more than I thought. And so the bottom line here is that the hardware end isn't really close yet.
You could get some volatility after earnings just like we saw with Broadcom, but frankly it's probably going to be a buy the dip between now and next year when more chip supply comes online as long as hyperscalers don't you-turn on their capex yet, which there's really no sign of yet. But I think 2027 is probably that year where we start seeing some of the rollover.
So in other words, still have time to go. The company's got a great valuation right now based on those Wall Street projections, which could change. Balance sheet is really strong. This company is a goldmine. It is a really durable company as well. Like I don't think there's a chance that they could go bankrupt.
I know famous last words, but their balance sheet is so freaking strong and they're paying off debt knowing the crash is coming and they're just trying to milk as much money as they can between now and then and build the most most robust infrastructure and grab the most market share they can. Honestly, they're smart.
I really respect the executives at Micron. They are doing exactly what they should be doing. They are very disciplined, very strong company. I have very little bad to say about this company. And I'm actually very excited about it. So Wednesday will be a very big day. It is the day I'm coming back. So we'll we'll cover the earnings afterwards. But anyway, if you like this, bye.
Comments 0