저자: 미상 · 외국인
요약 보기
AI 붐으로 메모리 반도체(마이크론 등) 주가가 과열됐으며, Burry의 논리에 따르면 ROIC 평균 4% 대비 자본비용 10% 이상으로 자본을 파괴하는 구조다. 메모리 산업은 42년간 34회 30% 이상 폭락한 경기순환 업종이고, 현재 PSR 12배로 역사적 평균 2.2배를 크게 웃돈다. S&P500 성장분의 50%가 반도체에 의존해 시스템 리스크도 크다. 대응으로 경기순환 하드웨어/반도체 매도 또는 역베팅, 안정적 고마진 소프트웨어 기업으로 자본 이동을 제시한다.
다듬은 글
Okay, think about this for a second. In the financial world, the most fascinating story is almost never the one that everyone keeps repeating in the news. Normally, the real story is where an analytical mind dares to go against the crowd. We all know Michael Burry, right? The guy who analyzed the real estate market in 2007 and found pure rottenness where the rest of the world only saw solid gold and easy money.
Well, today the markets are experiencing an absolutely unprecedented euphoria. All thanks to the artificial intelligence revolution. Memory manufacturers like Micron have been reporting triple- digit returns in the first half of 2026 alone. It's basically a total party on Wall Street, but right now, at the peak of this blind enthusiasm, Burry reappears.
And mind you, he's not doing it to applaud technological innovation. To understand why Burry is shorting the semiconductor industry, we need to look at the raw data. He claims that the collapse of this bubble is imminent. And in this analysis we're going to break down exactly why it challenges the dominant narrative that everyone else is buying into.
Now, to fully understand this puzzle, we need to briefly introduce the geopolitical factor that is key here. We're not just talking about a new technological toy or a consumer advance. We are talking about a fierce race for artificial intelligence supremacy. This is literally a matter of national security and global dominance.
This massive tension between superpowers has forced governments to shield their own supply chains, which has sent market optimism soaring to truly stratospheric levels. It is an extremely complex global scenario with a lot of money and political power involved. And well, before I continue to unpack all the layers of this thesis, I want to remind you that if you find this type of in-depth, data-driven explanation useful, I invite you to subscribe and leave a like.
That simple support is the fuel that allows us to continue bringing research of this level. So thank you very much for that. Okay, let 's get right into it. Today's menu is as follows. First we will see the geopolitical euphoria. From there we will move on to the capital destroyer, then cyclical logic, we will review Sealdini's biases , evaluate the market engine and close with investment strategies.
Section one, geopolitical euphoria. Let's assess exactly where we stand today with the demand for memory chips. To illustrate the madness we are experiencing, look at the words of Timok, the CEO of Apple. We're talking about someone with decades of experience managing the largest and most complex supply chains on the planet.
And seeing the current prices and the brutal shortage of memory chips, he called it, without hesitation, a once-in-a- century flood. Physically, manufacturing high-bandwidth memory, the famous HBM, is incredibly difficult. It requires stacking silicon sheets with millimeter precision, which generates many factory defects and therefore a very limited supply.
This encapsulates today's entire consensus. The massive demand for AI data centers has absolutely devoured all production capacity. In the short term, the scenario seems invulnerable, and all this frenetic optimism is based on future projections that are honestly astonishing. This number is particularly noteworthy.
50. That's right. Market consensus expects that by 2027, 50% of all spending on artificial intelligence infrastructure will be allocated exclusively and solely to memory. Half. We're talking about hundreds of billions of dollars flowing to just a few companies. Such a deluge of money creates an immense blind spot.
Many investors are assuming, just like in 1999 with internet infrastructure, that spending by large technology corporations will simply never stop. Section two, the capital destroyer. This is where we begin to see the tremendous contrast between the narrative and the cold, hard numbers. Here comes the first big reality check and it's the key to understanding why Micron is a long-term value destroyer.
Michael Burry doesn't say that AI was making an invention, no. He goes against the business model and the central problem is the low ROI versus 10% equity. Over more than four decades, the company's median return on invested capital has been a dismal 4%. But at the same time, the actual cost for the company to obtain financing in the market always exceeds 10%.
To give you a clear idea, imagine taking out a huge mortgage loan at 10% interest to buy a building that only yields 4% annual rent. They are systematically burning 6% of their money. It is because of this chronic inability to overcome its own financing costs that Berry's thesis calls this company the capital destroyer.
And if we dig a little deeper into those 42 years of history, things get even darker. This is where Borry 's financial logic truly shines . Look at this table. The company has had negative returns on equity in one out of every three quarters since it existed. But the data that really leaves you speechless is the free cash flow.
Look, free cash flow is the oxygen of the company; it's the real cash left over after paying for operations and investments. Well, this flow has been completely negative in 48% of the quarters. For half its life, this company has been burning cash like crazy, and that leaves you with the huge question: how on earth can a corporation survive with these numbers and on top of that become the superstar of the current market?
Section 3, cyclic logic. Let's move on to understand why this industry behaves in such extreme ways. This number says it all. To understand how cyclical memory companies work, you have to understand that they are not stable software technology companies, they are industrial giants subject to brutal supply and demand fluctuations.
This explains the cyclical logic of the sector and the pattern of 34 crashes that has historically devastated investors. In its 42 years of trading, this sector has collapsed 34 times independently, falling by more than 30%. Even worse, on 18 of those occasions the stocks lost more than half of their value and on seven occasions we saw collapses of more than 70%.
Falling 70% means you need the stock to rise 300% afterward just to break even. These are not rare accidents; it's simply how this industry works biologically. And these four steps perfectly illustrate how to detect cyclical stock market bubbles before getting caught up in them. It's always the same story.
Step one, a new technological catalyst causes demand to explode overnight. Step two, since there is no product, the prices and profits of these companies go to the moon. Step three, dizzy with so much success, manufacturers invest billions in new megafactories, but building a chip factory takes 2 to 3 years.
And we come to step four. When the machines finally start up and the product floods the market, the oversupply causes a price crater and profits disappear completely. Today we are told that memory is the scarcest resource in the world, but historically we are only living through steps one and two of this trap.
What's setting off all the alarms today is overvaluation. Currently, the stock is trading at more than 12 times its total sales. Its historical average is just 2.2. In other words, we are seeing valuations that break the most absurd records on thebubble.com. And be careful with this, many analysts will tell you to look at the famous price-to- earnings ratio, the P/E ratio, because it seems cheap.
Beware, that is the most dangerous financial trap in cyclical companies. That ratio is dividing the price between profits that are artificially inflated and at all- time highs due to a temporary shortage. Buying a cyclical stock just when its profits look most incredible is often the exact point of greatest risk of losing everything.
Section four, Sialdini biases. Let's see why the human mind and analysts ignore such obvious facts. If the story is so clear, why do the geniuses on Wall Street recommend buying? To understand this, the analysis reviews and analyzes Sialdini's psychological biases that explain the irrational behavior of consensus.
Let's see why biases have an influence. It lines up here. First, scarcity. The news reports all day that HBM memory is exhausted. That fear of being left out makes funds buy at any price. Second, social proof. If all your colleagues online and in banks say they are making fortunes with AI, your brain assumes that the crowd is not wrong.
Third, commitment and consistency. After 3 years of applauding the AI revolution, one analyst finds it too painful to change his mind. It's the sunk cost fallacy. And finally, the extremely powerful authority bias. Today, 52 out of 57 banks say buy. We also see figures of very high authority, including recent endorsements from former US presidents, praising investments in this sector.
And, regardless of political affiliation, when the masses see an authority figure endorsing something, they take it as an absolute guarantee of success. For a contrarian investor like Burry, all these biases screaming "buy" do not invalidate the bubble. On the contrary, they are absolute confirmation that we have already reached the peak of euphoria.
Section Five, the engine of the market. Let's keep going because this doesn't just affect those who buy chips. And this is critical because the cyclical bubble brings a huge systemic risk. Take a look at this chart based on Goldman Sachs data about the S&P 500 index, which basically supports the pensions and savings of half the world.
Of all the expected earnings growth for the S&P, this memory company alone accounts for 16%, and the rest of the semiconductor EAs add up to another 34%. In other words, exactly 50% of the future growth of the entire market depends on a single technology group. There are millions of people buying passive funds thinking they are oversupplied when in reality they are blindly betting that a cyclical industrial sector will never again have an oversupply.
If chip prices fall, as they have dozens of times, half the market engine will stall, and that will be felt in everyone's pockets. Section six, investment strategies. We've reached the practical part. What do we do with all this information? It's time to talk about investor strategies after a crisis, or rather, how to prepare for the imminent change of cycle.
Predicting the exact day and time a bubble will burst is impossible, but relocating capital is possible. After the crisis, investors focus on positions of relative value. This means selling off or betting against cyclical industrial assets that are awash in euphoria, such as hardware and semiconductors, and sheltering that capital in much more stable tech giants.
Businesses with recurring revenue from high-margin software or services, like Microsoft, have real competitive advantages and don't need to burn through cash wildly to survive. In short, you sell the exaggerated buying panic and buy the boring stability that the market is ignoring today. At the end of the day, 100 years of financial history show us that the market will always bounce violently between extreme greed and blind panic.
The news will continue to sell unstoppable disruptions, but to survive and protect money, we cannot be guided by institutional delays or our own biases. We must coldly examine the true return on capital and respect the brutal history of oversupply. And this leaves us with a question to think about deeply.
When the factories finally fire up, the excess of chips floods the planet, and the cycle mercilessly spins downward, what do you think will sustain investment portfolios? The pretty tale of a new golden age or the undeniable reality of financial data? The market never ceases to teach lessons. The secret is simply knowing where to look.
Comments 0