저자: 미상 · 외국인
요약 보기
스탠리 드러켄밀러의 13F 공시 분석. 그는 4분기에 샌디스크(SNDK)를 전량 매도하고 아마존(AMZ)을 신규 매수했다. 샌디스크는 AI 데이터센터 수요로 메모리 가격이 1년간 3배 뛰어 분기 매출 61%, 조정 EPS 404% 증가했지만, 메모리는 상품이라 가격결정력이 없고 75배 PER은 사이클 고점 리스크가 크다는 논리. 아마존은 물리적 AI(물류·로봇), AWS 가속(24%), 자체 칩 성장으로 구조적 해자를 가지며 PER 약 30배가 매력적이라는 주장.
다듬은 글
Welcome. I am uh I'm so incredibly glad you decided to join us for today's deep dive. Yeah, it is great to be here. We've got a really good one today. We really do. We have a genuinely um fascinating mission on the docket. Stanley Dreen Miller's latest 13F filing just dropped. Always a big day when that happens,
right? and his fourth quarter portfolio moves. They offer this absolute masterclass in reading the current artificial intelligence landscape because right now there is just so much noise out there about AI. Exactly. Every single company is slapping the AI label on their earnings reports hoping for a bump.
They are. So, we're going to look at a stack of comprehensive financial analysis reports alongside this new filing to figure out how, you know, one of the greatest risk managers in history actually separates structural long-term value from short-term hardware hype. Peeking over the shoulder of someone with his specific track record is always revealing.
Oh, absolutely. It forces you to question your own assumptions about uh about where a market cycle actually is versus where the retail crowd thinks it is. And for those of you listening who might just know the name Ducken Miller as like a fixture in financial headlines, let's establish why his portfolio demands this kind of attention.
Right? The background is important here. The sources highlight his tenure running the hedge fund Dukane Capital. This was between 1981 and 2010. The completely different era of the market really. Yeah. And during that near three decade stretch, he averaged a 30% annual return. Just to put that in perspective for you, the broader market's historical average is somewhere around 10%.
Right? So tripling the market average for 30 years is frankly statistically monumental. But the part that actually defies gravity, um the part that's just wild is the consistency. Oh, the no down years. Exactly. Our notes point out that he achieved that 30% average without a single downyear. not one negative year in three decades,
which is exactly why his current moves are so heavily scrutinized, even though he closed his fund to outside investors back in 2010. He only manages his own family's wealth now, right? Right. Through the Dukane family office, but because of the sheer size of those assets, his office is still required to publicly disclose their holdings every quarter through those 13F forms.
So regular investors can still follow his portfolio. Exactly. And his latest disclosures show a massive textbook pivot. A really aggressive one. Very. He completely liquidated a wildly successful position in one specific sector of the AI boom and then aggressively deployed that capital into a historic tech giant in another sector.
Okay, let's unpack this. We have to start with the asset he sold because, you know, on the surface, walking away from this company looks completely counterintuitive. It definitely looks strange if you only look at the recent charts, right? He liquidated his entire position, every single share in SanDisk, ticker symbol SNDK.
Yep. And I was looking at the notes on this company since they were spun off from Western Digital last year. And the growth looks like a misprint. It's staggering. It really is. They're up what, 1470% since the spin-off. Roughly 147%. Yes. What is driving their stock right now? Why are they exploding?
The defining catalyst is the AI infrastructure boom. It triggered an absolutely unprecedented supply squeeze. Okay, so SanDisk designed storage products based on nan flash technology. Yeah, we're essentially talking about memory chips here. Got it. Half of their business is embedded drives for edge devices.
So that's the storage in your laptop, the systems in modern cars, high-end gaming consoles, the everyday consumer tech. Exactly. Mhm. But the other half which is the real profit driver right now consists of solid state drives for massive data centers. Ah the AI data centers. So with this massive rush to build out AI infrastructure, everyone suddenly needs a mountain of memory chips and there just aren't enough to go around.
That is the core of it right there. When demand goes parabolic and supply is constrained, the underlying commodity prices just explode. And the data shows that, right? Yeah. Yeah, the data shows nan flash memory prices have literally tripled in the past year alone. Tripled. So if the price of your core product triples, your top line must be going crazy.
It is. In their January quarter, revenue surged 61%. Wow. They hit $3 billion, which was overwhelmingly driven by those data center sales. But the operational leverage is what really caught my eye in the report. The earnings jump. Yeah. If you look at their core profitability, their nonjapad adjusted earnings, those quadrupled
404%, right? Up 404% landing at $6.20 per diluted share. It's a massive windfall. It is. But I also noticed in the reports that SanDisk has this joint venture with Kioia, a Japanese manufacturer. Yes, that's a crucial detail. They share the costs of memory design and fabrication equipment, which you know on paper sounds like a really smart way to generate cost savings and secure the supply chain.
It does. But does that partnership play a role in how a macro investor like Draen Miller views the company? It plays a massive role and it actually hints at the underlying vulnerability of the entire business. Oh, so well building a modern semiconductor fabrication plant costs tens of billions of dollars,
right? It's not cheap. Not at all. The fact that SanDisk has to partner with KioI just to afford the fabrication equipment shows you how brutally capital intensive this space really is. They are sharing the burden because neither of them has the pricing power to do it alone. Wait, hold on. If earnings just quadrupled and analysts are projecting massive growth, why are we talking about vulnerabilities?
Because you have to look ahead. If I'm holding a stock that's up almost 1500% and the AI boom is showing no signs of slowing down in 2026, my instinct is to hold on for dear life. Sure, that's the natural retail instinct, right? So why on earth would someone dump the entire position? What's fascinating here is that you have to look past the current quarter's phenomenal numbers and look at the structural reality of the product itself.
poker. Our source is quote William Kerwin, an analyst at Morning Star, and he states it very bluntly. What does he say? He says, quote, "We don't believe SanDisk holds an economic mode. We view flash memory chips as commodities that don't command any pricing power." A commodity. Exactly. So, a memory chip is just a memory chip, and SanDisk is just surfing a macro wave.
They aren't offering something wildly proprietary that no one else can make. You nailed it. And to add some context to that, SanDisk is actually only the fifth largest supplier of NAN flash globally. Oh wow, I didn't realize they were fifth. Yeah. And when you sell a commodity, your entire profit margin is dictated by the broader market cycle.
And the memory chip industry is notoriously, brutally cyclical. So it's the classic bullwhip effect. Precisely. It reminds me of the great toilet paper shortage of 2020. That is a great analogy. Everyone panics. Demand spikes. The shelves empty out and prices go nuts, right? Then every factory on Earth goes into overdrive producing toilet paper.
Suddenly the supply chain catches up. People stop hoarding and you end up with a massive glut of inventory that nobody wants to buy. That is the exact mechanism just applied to highly complex silicon. So right now we are in the panic hoarding phase of the cycle. Yes, the AI created a shortage, prices tripled and SanDisk's earnings quadrupled.
But the immutable law of cyclical hardware is that today's supply shortage undeniably turns into tomorrow's glut. Always all the manufacturers, not just SanDisk, but the four companies larger than them too, they are all rushing to build more capacity to capture these high prices. And the moment that new capacity floods the market,
prices will crash just as fast as they rose. Let's talk about how the market is currently pricing that risk because the analysis notes SanDisk is trading at 75 times adjusted earnings which is very high. That sounds incredibly steep for a hardware manufacturer selling a commodity. It is exceptionally steep.
Now the bullish counterargument is that Wall Street forecasts their adjusted earnings will grow at 155% annually through the fiscal year ending in June 2028. That's a huge projection. It is. And if that perfect linear growth actually materializes, you could try to justify the 75 multiple. But you just said the cycle is going to turn.
So that linear growth is an illusion. That is the ultimate trap of a cyclical peak. The market prices the stock assuming the current supply constraints and high prices will last forever. But the analysis warns that the 75 times valuation is highly likely to collapse the second it becomes clear to the broader market that the memory chip cycle has passed its peak.
And when that realization hits, the multiple compression will be violent, extremely violent. And the thing is, no one rings a bell at the top of the market. You can't know the exact Tuesday when the supply glut is going to hit the data center. You cannot. Therefore, holding a commodity stock at 75 times earnings is relying entirely on hope.
Yeah, that makes sense. In that context, liquidating the position isn't premature. It is the discipline of knowing when you have won the trade, taking your massive profits off the table and leaving the riskiest, most unpredictable tail end of the cycle for someone else to gamble on. That is a masterclass in risk management.
He rode the supply shock and now he's getting out before the factories catch up. So if you are sitting on a mountain of capital after cashing out of a cyclical hardware run, where do you put it? You don't just sit on cash in a bull market. No, you look for the exact opposite of a cyclical commodity,
which is you look for a structural monopoly. And that brings us to the second half of this 13F revelation. In the fourth quarter, Ducken Miller opened a massive new position in Amazon. Ticker symbol AMZ. A classic name, right? We all know the historical performance here. It's up something like 223,000% since its IPO in 1997.
Just staggering numbers. But the sources break down Amazon's modern dominance into three distinct pillars, right? You have the e-commerce marketplace obviously, which is the largest in North America and Western Europe. You have their advertising business, which is surprisingly the third largest adtech operation globally.
People often forget about the ad revenue. It's massive. And then you have Amazon Web Services or AWS, which is the largest public cloud for infrastructure and platform services in the world. Three massive businesses, all generating entirely different types of value. Yeah. But the key differentiator here isn't just the pillars themselves.
It's how they are treating the current technological shift. And here's where it gets really interesting. Our sources detail how Amazon is moving past AI as just a cloud service and they're integrating it directly into the physical world. Yes, they have developed hundreds of generative AI tools specifically to make their retail and logistics networks more efficient.
This is a concept analysts are calling physical AI. Physical AI. Yeah, it is incredibly difficult to execute, but the payoff is massive. The examples in the reports are wild. They are using AI to optimize inventory placement. So, the specific product you might buy is already sitting in a warehouse 10 miles from your house before you even click purchase.
Predictive logistics. Exactly. They're optimizing complex warehouse workflows and last mile delivery routes. And they were deploying AI frameworks that allow their massive fleet of fulfillment center robots to navigate the floors and process commands using natural language, which is incredible tech.
I just love the idea of a warehouse manager casually asking a robot to go reorganize the shoe aisle. It paints a fun picture for sure, but the financial implications of those natural language robotics are staggering. How big of an impact are we talking? Well, Morgan Stanley Analysts, led by Brian Noak, recently put out a note explicitly listing Amazon as one of the companies best positioned to benefit from this physical AI deployment in the coming years.
Let me push back on that for a second, though. Sure. Applying cutting edge artificial intelligence to thousands of physical warehouses and delivery fleets has to be unbelievably capital intensive. Oh, it is. Does all that spending actually translate to the bottom line or is it just an expensive science project?
It requires an enormous upfront investment, but the long-term margin expansion is exactly what is drawing in investors like Duck and Miller. Okay, walk me through that. According to that Morgan Stanley note, despite the heavy capital expenditures, they expect Amazon's operating margin to expand significantly over the next few years.
By how much? They're projecting operating margins to jump a full two percentage points. Just 2%. I mean, in the context of the 400% earnings jumps we just saw with SanDisk, 2% sounds almost negligible. You'd think that, but you have to apply that margin expansion to Amazon's scale. Fair point. We are talking about expanding from an estimated 11.1% operating margin in 2025 all the way up to 17% by 2028.
Wow. For a company generating hundreds of billions of dollars in revenue, that kind of total margin expansion translates to an astronomical amount of free cash flow. That makes total sense. They aren't just selling a chip at a temporary markup. They are permanently lowering the cost of running the largest logistics network on Earth.
Okay, that context flips the script entirely. That is the literal definition of a structural advantage. Exactly. But we can't really talk about Amazon's future without talking about the third pillar, AWS. No. AWS is the anchor for everything else. It is the foundational layer for a massive portion of the global internet.
And that scale alone gives them a tremendous almost unfair advantage in the AI race. Our sources pull a great quote from Amazon CEO Andy Jasse that sums up that advantage perfectly. Oh, the data gravity quote. Yeah. He points out, quote, AWS is where the preponderance of company's data and workloads reside and part of why most companies want to run AI on AWS.
It really all comes down to that concept of data gravity. Explain that for us. If you are a Fortune 500 company and your pabytes of customer data have been sitting in Amazon's cloud for the last decade, migrating all of that to a competitor just to run a new AI model is an expensive logistical nightmare.
You'd have to move everything, right? You're highly incentivized to just use the AI tools that Amazon provides natively right there where your data already lives. So they have a totally captive audience. They do. And the reports show AWS is aggressively leaning into that. They are rolling out new AI products across their entire computing stack.
But they're also making huge strategic partnerships, right? Yes. Very strategic. They position themselves as the primary cloud provider to Anthropic, which is widely considered one of the top foundational AI model developers right up there alongside OpenAI. And by locking in Anthropic, they ensure that the most cutting edge models are deeply optimized for Amazon's specific infrastructure.
And we are already seeing this show up in the quarterly numbers. Oh, absolutely. The analysis notes that in the fourth quarter, AWS revenue growth accelerated to 24%. That's huge for a business of that size. It is their fastest pace of growth in 13 quarters. But what really caught my eye was the hardware side of AWS,
the custom silicon. Exactly. A huge driver of that 24% growth was tripledigit sales growth in their custom chips, which takes their moat a step further. How so? Well, think about why custom silicon matters when everyone else is just buying off-the-shelf GPUs from Nvidia, right? Designing their own specialized chips for training and inference allows AWS to offer better pricing to their cloud customers while simultaneously keeping more of the profit margin for themselves.
Oh, that's incredibly smart. It is another layer of insulation against the broader hardware cycle. So bringing this back to the portfolio strategy, we saw Duck and Miller exit SanDisk because 75 times earnings was a massive risk for a commodity cycle that is bound to peak, right? How does Wall Street value this expanding moat at Amazon?
The valuation comparison really drives the thesis home. Wall Street estimates suggest that Amazon's earnings will increase at a steady 19% annually through 2028. Okay. A steady 19%. Yes. And based on that predictable structural growth, Amazon's current valuation sits at roughly 30 times earnings. 30 times earnings for the dominant player in e-commerce, cloud computing, and retail advertising
versus 75 times earnings for a memory chip maker that has no pricing power. When you look at it through that lens, 30 times earnings looks incredibly reasonable. It is a very attractive entry point for patient capital. You are not betting on how long a supply shortage will last. you are investing in the steady compounding of efficiency and scale over years and decades.
As we wrap up today's deep dive, I think that is the core lesson for you to take away from this. I completely agree. This analysis isn't meant to be a simple tip sheet on what one family office bought or sold in the fourth quarter. It is a fundamental lesson in distinguishing between a short-term commodity cycle and a long-term technological moat.
Right? Riding the wave of a flash memory shortage requires perfect timing before the bubble pops. But investing in a deeply integrated cloud and physical robotics ecosystem relies on structural permanent shifts in how business operates. It is the discipline of knowing what game you are actually playing.
Beautifully said. But before we sign off, exploring this dynamic between cyclical hardware and structural AI raises an important question that I think is really worth exploring further on your own. I love a lingering question. Let's hear it. We spent a lot of time today discussing how the memory chip industry and really all hardware is governed by predictable boom and bust cycles.
Right. Supply shortages lead to overbuilding which leads to a glut. Exactly. And historically those cycles play out over several years. But with artificial intelligence accelerating the pace of technological consumption and evolution faster than ever before in human history. Yeah. Will these traditional hardware cycles become significantly shorter and more violent?
Oh wow. Because the physical demands of the data centers are shifting so rapidly the chips become obsolete faster. Exactly. If AI completely changes the timeline of hardware obsolescence, meaning a massive data center needs to be entirely refitted with new networking gear and memory every two years instead of every five or six.
That changes everything. It does. How will that fundamentally change the way we value the physical infrastructure of the future? If the boom and bust happens twice as fast, it changes the entire risk equation for investors, for the hardware manufacturer themselves, and for the tech giants trying to build the future.
That is a fascinating thought to end on. The rules of the hardware cycle might be rewriting themselves right before our eyes. And catching the top of the next wave is going to require tighter risk management than ever. It certainly will. Thank you so much for joining us for this deep dive. Keep asking the big questions. Keep looking past the hype.
Comments 0