원문 보기
Michael Burry, the investor from The Big Short, just added to his bets against chip stocks on the very day the Nasdaq hit a record. His main target is memory chips, the corner of tech where one stock went up 16 times in a year and a half. His bet rests on one old pattern, and I'll show you how many times that pattern has played out before.
Plus, his own scorecard, including a call he later admitted in public was wrong. First, why anyone listens to this man at all? 20 years ago, Michael Burry was a doctor turned fund manager, running a small fund in California called Scion Capital. In 2005, he started buying a kind of insurance that would pay him if American home loans went bad.
Wall Street banks were happy to sell it to him because almost nobody believed home prices across the whole country could fall. For 2 years, he paid the premiums, the bet looked foolish, and some of his own investors demanded their money back. Then the housing market broke, and the bet paid off with more than $700 million in profit for his investors.
Michael Lewis wrote the story into The Big Short, and in the movie, Christian Bale played Burry barefoot, drumming, and blasting heavy metal in his office. Since then, every time Burry bets against something, the headlines follow. Last November, his fund's filings showed bets against two stars of the AI trade, Nvidia and Palantir.
Headlines said more than a billion dollars, but that was the value of the shares behind the contracts. Burry corrected them himself. For the Palantir bet, he had paid about $9 million for put options, contracts that make money if a stock falls. Palantir's chief executive went on television and called the bet, in a word I'll soften, crazy.
A few weeks later, Burry closed his fund to outside money and started writing for anyone willing to pay. He named the newsletter Cassandra Unchained, after a nickname he says Warren Buffett once gave him. In Greek myth, Cassandra was a Trojan princess cursed to see the future clearly and never to be believed.
She warned that Troy would fall. Nobody listened and Troy fell. Burry clearly enjoys the comparison and in less than 8 months more than 300,000 people signed up to read his warnings. To be fair to Cassandra, she never had a paywall. Now to what he actually wrote this week, the post is dated September 22nd and it opens like a man watching a game he has money on.
We are all watching the big game. The Nasdaq 100, historically overvalued and historically top-heavy, looks about to drag all that weight to a new all-time high. Top-heavy means that a handful of giant companies make up a huge part of the index. So, when they move, everything moves. Then comes the reason for his post.
The chief executive of Acer, the Taiwanese computer maker, is talking about a chip glut again. Acer buys memory chips by the truckload. So, when its boss talks about memory prices, he's talking about his own costs. Burry quotes him at length and one line does most of the work. No way. How could there be a continuous shortage?
China's production capacity has been consistently increasing and there is absolutely no shortage issue. Burry's own comment is careful. Interpreting this news is fraught, but it aligns with what I believe to be true. According to CNBC, in the same post he spelled out his forecast. Over the next 2 years, this shortage will blow off as production catches up and memory will have a down cycle again.
So, he's adding to his bets against Micron, the largest American memory maker, against Nebius, a company that rents out AI computing power, against Palantir, and against a whole fund of chip stocks. And he says he still holds his puts. But, he isn't only betting against things. The same day he bought more of five stocks that, in his words, have corrected tremendously.
The list includes Birkenstock sandals, the grocery chain Sprouts, and Build-A-Bear Workshop. To see why memory chips in particular, you need one idea about how they're different. An AI chip from Nvidia is like a brain, and nobody else makes one quite like it. A memory chip is where the brain keeps what it's working on, and one maker's chip does roughly the same job as another's.
That makes memory closer to oil or wheat than to software, and its price is set by one thing, supply against demand. For the last 3 years, demand has been off the charts because every AI data center needs mountains of fast memory. Look at what that did to Micron. In early 2023, Micron was selling about 3.7 billion dollars of chips a quarter.
In the quarter that ended in May of this year, it sold 41 and a half billion. That's 11 times more in 3 years, and Micron's own forecast for the quarter it reports next is around 50 billion. The stock went from about $65 in April of last year to more than a thousand, and Micron is now worth over a trillion dollars.
That's the boom Berry expects to end, and a boom this size makes the old pattern very easy to miss. Most news stories stop at the forecast, so I went back 30 years. Here is Micron's stock since 1995, and every red zone is a moment when memory stopped being scarce. 1996, down 82%. 2002, down 90. 2008, down 91.
2011, 2016, 2018, and 2022, each time about half or more gone. Eight times in 30 years, Micron's stock has lost about half its value or more. And the most recent one isn't ancient history. It ended just a year and a half ago, in April of last year, down 58%. The business itself swings just as hard. In the fiscal year that ended in 2023, Micron's sales fell by half in a single year.
Why does this keep happening in an industry full of very smart people? Because a memory factory takes two to three years to build, and everybody decides to build at the same time, when prices are highest. By the time the new factories open, they all open together. The shortage turns into a glut, and prices collapse.
Then nobody builds, a shortage returns, and the wheel turns again. The memory industry has the memory of a goldfish, which is ironic given what it sells. Investors call this the capital cycle, and there's a whole book about it from the London firm Marathon Asset Management. Its idea fits in one sentence.
High profits attract new money, and that new money eventually kills the profits. Burry wrote about exactly this two days later in a post that opens with a question every investor should ask. Where are we in the capital cycle? He measured how much big American companies invest above what simply wears out against the size of the whole economy.
By his count, it's now higher than at any point in nearly 40 years, except the aftermath of the dot-com peak. And you can see the building in Micron's own reports. In the first 9 months of this fiscal year, Micron spent $19.6 billion on factories and equipment, almost double a year earlier. Several of those new plants are going up in Idaho, and according to Acer's boss, Chinese memory makers are adding capacity without limits.
A research paper from a Washington think tank put it plainly this month. Chinese overcapacity could become the next memory market risk. So, the supply is coming, and the only real question is when. Now, his scorecard, because he'd want you to see it. The housing bet was right, but it came about 2 years early, and those 2 years turned some of his own investors against him.
In January 2023, he posted a single word to his followers, sell. Two months later, he posted again, I was wrong to say sell. And last December, he retold the whole story himself, the admission included. It takes a special kind of confidence to publish I was wrong and still charge for the newsletter.
As for last year's bets, Nvidia is higher than when those puts made headlines, and Palantir is a little lower, so for now, call it a draw. So, here's the one doubt I'll raise, and it isn't about his logic, it's about the clock. Howard Marks has a saying he learned in the early 1970s, and he still repeats it.
Being too far ahead of your time is indistinguishable from being wrong. The memory cycle has turned eight times, but nobody, including Burry, can tell you the month of the ninth. And here's the line most headlines skipped. In the same post, Burry wrote down in advance what would make him back off. I will watch, however, and look to cover should the market take these up to new highs again.
To cover means to close the bet, even at a loss. That's the whole difference between a profit and a professional. Before he bets, he writes down what would prove him wrong. Most of us have exit rules, too. They're just called, "I'll hold until it comes back." So, what does this mean for you if you're not shorting anything?
If you own a Nasdaq 100 fund, you already own a big slice of these chip stocks, so look inside and see how much of your money rides on one boom. If you own a cyclical business like memory, remember the old rule, it looks cheapest at the top when profits are huge and most expensive at the bottom when profits vanish.
And please don't copy a short seller's trade. A put can expire worthless, and Burry can afford to be early in a way that most of us can't. In the myth, Cassandra was right about Troy, and it didn't save Troy because being right without timing is only half of the story. Burry has been Cassandra before, and he has been simply wrong before, and he writes about both.
What he offers this time isn't a date. It's a pattern that has repeated eight times, and a rule for what he'll do if he's wrong. That's worth copying even if you never make a single bet against anything. Now, I'd like to hear from you. Is this time different for memory chips because of AI, or is the wheel about to turn again?
Write different or wheel in the comments and tell me why. And so I know how to pitch the next video, was this too simple, just right, or too deep? Just write one word, simple, right, or deep. Thanks for watching, and if you made it to the end, subscribe because next time we're opening another letter from a legend.
And if memory chips fall while you're watching this, remember you heard about the pattern here, not the date.
Comments 0