원문 보기
Welcome to the Wall Street Lounge. Today we're going to look at what exactly Michael Berry is up against right now, and I mean exactly , right down to the execution prices. Got it. He personally revealed this at the end of September, and it's pretty aggressive. These are three names: Micron, Nvidia, and Palantir.
Wait. Micron is one of the strongest stocks in the entire market this year. That's right, it has grown by 281% since the beginning of the year. And he bought put options with a strike price of $500, which is about half of today's price. It is rarely possible to find out the execution prices. This is true.
He publishes the expiration and strike in his bulletin, so for the first time we can check the math of the bet instead of guessing. $500, and Micron is worth $1,088 today . So, he needs the price to fall by half. Even more than twice. And it expires in June 2027 . So the bet is that Micron will fall in half in about 20 months.
Ahem. This is a bold prediction. That's right. And most importantly, Berry himself said he had shortened his terms. In August, he talked about 2028. Now he has postponed them to the near future. Does the expiration date change its risks? Fully. A typical short position can be held for years if you can withstand losses.
A put option has an expiration date. If the stock doesn't fall by that date, the money he paid will be gone. A. So just saying that stocks will fall someday is not enough? That's right. He placed his bet on June 2027 . That's why shortening the terms and switching to put options are actually the same decision.
For those who don't know him, who is Berry? This is the investor who was featured in the movie "The Downside Game." He predicted the housing market crash in 2008 and made a fortune shorting mortgage bonds. Ah, that guy. So, he predicted everything correctly again ? This is where you should be careful.
He was also a bear in years when the market continued to rise , and then he was wrong . He has one resounding victory, he is not sinless. Do we receive official reports from him? No, and that's important. He closed his fund in 2025, so there are no more quarterly 13F reports. Everything we have comes from his own bulletins reporting this.
How did this even become known? In three stages. On September 22, he wrote that he had increased, what he said were, significantly short positions in Micron, Nvidia, Palantir, and the semiconductor ETF. Of course. And then September 28th—the most important thing. He wrote that as of today, put options have completely replaced my short positions.
Fully? Fully. And then, on October 6, he published that the market was in the first stage of grief— denial—and that, based on 2000 and 2008 , this stage lasts 6 to 9 months. Are ETFs there too? Yes, SOXX, a semiconductor index fund . So, he is betting on the downside of not just three companies, but the entire sector, plus the Nasdaq 100 index itself, using put options for September 2027 .
Individual names, sector and index. All together. So this isn't a statement that he doesn't like one company, it's a statement that the entire AI sector is overvalued. So where exactly are these bets? Micron, June 2027, strike price around $500. Nvidia, also June 2027, with strike prices in double digits.
Palantir, September 2027, just above 100. Double digits for Nvidia? It is trading at 237. That's right. So, the stock would have to fall more than 60% before this contract would have any value. He also has puts on Nvidia for September 2027 with a strike price in the mid- hundreds. Each of them involves a serious fall.
And besides these three, there is something else. He opened a new short position on MetLife using long-term puts with a strike price in the seventies. Insurance company? That's a whole different story. That's right. There is also Caterpillar , December 2027, around 400, and Oracle, December 2027, in the mid-double digits.
Meanwhile, he closed his short position on CoreWeave and wrote that he was unable to buy puts there at a price that suited him. He admits what he could not achieve. This is one of the reasons why this newsletter is worth reading. Why did he abandon short positions altogether? His own words are quite clear.
He wrote that fundamentally he is moving the timeline forward, and therefore wants more leverage. Sounds logical. The second reason is the price. He argued that puts are cheap because the VIX and similar volatility measures are unusually low. Why does a low VIX make puts cheap? Option prices are based on how much the market expects prices to change .
When everything is calm, such insurance is cheap. As soon as the sale starts, the same put costs many times more, and you can't go in anymore . Ah, it's impossible to buy fire insurance when the house is already on fire. That's the idea. He buys while it's quiet. The downside is that if nothing happens, he will simply lose the premium paid and the option will depreciate in value.
Let's look at them one by one. First Micron. Yes, and this is the basis of the whole position. This is the asset that grew by 281%. The biggest winner is also the biggest target. Here's what this deal looks like. Let's start with stocks. Micron is up 281% this year. A year ago, in 2025 , they closed at $285, and now they are at $1,088.
This is almost four times more. Nvidia rose 183%, while Palantir only rose 9%. So, it was Micron that made a real breakthrough. And it is on them that he puts his "short" most aggressively. Is memory usually this explosive? Memory is one of the most cyclical industries. When supply is limited, prices skyrocket.
When new capacities appear, they collapse. The demand for AI servers has been greatly driven by the shortage. So the recession is finally coming. Historically, yes. The question is never if it will happen, but when. Berry gave the date: June 2027. What do the real numbers look like? Honestly, very convincing.
In the 4th fiscal quarter of 2026, reported on September 30, revenue was $ 54.2 billion, up 379% year-over-year , with a gross margin of 87% on a non- GAAP basis. 87% gross margin for a memory manufacturer? Yes. Annual revenue was $133.2 billion, and earnings per share increased by 811%. The forecast for the next quarter is $ 61.5 billion, another record.
This doesn't look like a company that's about to collapse. What about data centers? The core data center segment accounted for 18 billion, 33% of total revenue. DRAM — 39.8 billion, an increase of 343%, and NAND reached 14.1 billion, both figures are record. And they spend even more? Much more. About $11.5 billion next quarter and about $25 billion in the first half of fiscal 2027, and most of that is construction that won't add output until the end of 2028.
So what is he really worried about ? About the offer. He wrote that memory manufacturers are ridiculously expensive, and the price could drop sharply when cheaper Chinese products appear and the shortage disappears. Does he have evidence? He cites the head of Acer, who said that in the memory of the older generation, there are now more sellers than buyers.
Something else? He also noted a report that a Chinese manufacturer had increased DDR5 yield by over 90%, commenting, "That's huge if true." Better product yield figures mean much more cheap supply soon. But Micron claims otherwise, right? Exactly the opposite. During the earnings conference call, management said it expects significantly tighter supply and demand conditions in the memory and storage market in 2027 and 2028 than in 2026.
Tighter—does that mean the deficit will continue? Yes, and they went further , stating that they see no prospect of supply and demand ever balancing out. So, the company and the short seller are saying completely different things. No one yet knows which of them is right. Is the company simply praising its own achievements?
Maybe, but they backed it up with concrete facts. They stated that most of their high- bandwidth memory shipments for 2027 are already contracted with significant price increases. So, next year is mostly sold out already? That's their claim, which means that the June 2027 deadline for Berry falls right into the period that the company says is already booked.
Number two, Nebius. And this is a completely different profile than Micron. What's the difference? Micron is profitable and finances its development with earned funds. Nebius is still losing money, spending far more than it earns. Nebius rents out AI computing power. They install GPUs and sell capacity by the hour .
How is business? In the April -June quarter, revenue was $582.3 million, up from $105.1 million a year ago. This is a growth of 454%. Five and a half times. Adjusted EBITDA was $236.2 million , a positive figure compared to a loss a year ago. The annual revenue figure at the end of June was $3 billion.
Do customers pay in advance? A significant amount, yes. They sign contracts and receive an advance payment that covers part of the bill. This really helps. So why are these shorts? Structure. GAAP net income was a loss of $190.4 million , and capital expenditures for the quarter alone were about $5.7 billion .
Wait. 582 million in revenue and 5.7 billion in expenses? Right. They have about $8 billion in cash, but the capital spending plan for the full year is $20 billion to $25 billion. This model operates on a continuous funding basis. So, rising rates cause immediate harm. This is a pressure point, and probably what Berry was aiming for.
They also announced a $4.5 billion convertible bond in August. Did they change the forecast? No, they left it unchanged. Annual revenue of 7 to 9 billion, income of 3 to 3.4 billion, EBITDA margin of about 40%. Business shows results. The only question is financing. The third is Palantir. And this is the strangest of the three companies.
Strange in what way? Because there are no problems with business at all . Palantir sells data analytics platforms to governments and large enterprises. Revenue for the second quarter was $1.935 billion , up 93% . 93%? The commercial sector in the US alone grew by 149% to 764 million. Their " rule of 40" score reached 155%.
These numbers are impeccable. And GAAP net income was $1.062 billion . This is not a story about burning money. And the cash flow? Adjusted free cash flow 1.22 billion, margin 63%. They raised their annual revenue forecast to $8.15 billion . If business is so good, why open a short? Rating. Palantir has grown by only 9% since the beginning of the year.
Business grew by 93%, while the stock was virtually flat. Wait, doesn't that make them cheaper? You can also look at it this way . Berry sees it differently. Stocks have skyrocketed in recent years , and now it's paying off. Its low strike price of $100 is a statement that the revaluation is not even halfway complete.
How much do they cost now? $194 with a 52-week range of $106–$207. That's the upper limit, but the momentum has worn off, and its paths run until September 2027—the longest of the three. He gives this position time. Is he just selling or buying something else? Here's the most interesting part. In the same post, he disclosed five long positions: QXO, Build-A-Bear, Sprouts Farmers Market, Birkenstock, and Mercado Libre.
No name from the field of AI. None. Building supplies, toys, groceries, footwear and e-commerce in emerging markets. He said that they have all been significantly adjusted and look quite attractive. Sell AI, buy boring. That's the essence of one-line trading. Do these five companies have anything in common?
They are all outside the topic of AI and are far from their peaks. Real income , real profit, without the hype markup. Classic value thinking. And what does he think about time ? He believes that everything has barely begun. But this is his prediction, not a fact. If you compare all three, what do you see?
Nebius leads the way in growth with 454%. Micron —379%. Palantir—93%. All indicators are very high. What about risk? Nebius is the most vulnerable. Loss-making, large capital expenditures, dependence on financing. Micron is profitable, and the company insists the deficit is ongoing. Palantir is profitable and has already managed to stand still .
So which shorts are most likely to work? Probably Nebius, because it is the most sensitive to financing conditions. The most difficult one is Micron, where the business is operating at full capacity. And it was on Micron that he bet the most. Which is the most amazing thing. He made his biggest bet on the hardest case.
Let's put the complexity into numbers. Micron needs to fall by about 53%. Nebius— over 60%. Palantir— about 45%. Nvidia— about 37%. They are all almost twice as much . And options have an expiration date. If the stock doesn't fall enough by that date, the contract will be worth nothing. Unlike a short position, he cannot simply wait longer.
Should ordinary investors copy this? Honestly, no. If you make a mistake over time, the entire prize will be lost. A professional planning this as part of a larger portfolio acts very differently than a private investor copying a deal. It 's better to use this as an analysis tool. That's right. Take this as a reasoned bearish scenario and use it to stress test what you already have.
Another look at the situation. This bet is n't really about whether the demand for AI will stop. It all comes down to two things. When memory supply catches up with demand and when high cost estimates will be revised. So, let's summarize. On September 28, Berry replaced all of his short positions with put options and moved his AI “bubble” forecast to 2028.
The targets are Micron, Novius, and Palantir. And each of them should fall by 40-60% before the fixed expiration date . What is most striking is that Micron itself claims to have no understanding of when supply and demand balance out. We will find out who is right by June 2027 . Thank you for watching.
If this analysis was useful, like it and subscribe for new in-depth analyses of the US market. Mhm.
Comments 0