원문 보기
I think this is a fundamental shift towards purchasing computing power. That's what we're talking about. This is Nvidia, this is SK Hynix and this is Micron. The money is going into another area, and I'm not sure it will come back. ServiceNow, CRM—they all fall in the premarket. A more pessimistic theory, Jim, is that this is part of the shift to direct corporate installations.
We've already discussed these things. I think the Starbucks story is much more important than many people think, because they're trying to save $400 million on IBM and Microsoft. Microsoft is gradually growing. I spoke to Marc Benioff and he would tell you that this is complete nonsense. That means his business is doing better than ever.
Things are going badly for Adobe. ServiceNow will tell you that their business is not just better than ever, they will surprise many skeptics. I just know that this is not the place to be because it simply doesn't resonate. And where should you be? Look at this SK Hynix . Look, SK Hynix, David, what do you think their multiplier is for next year?
Eight times. 3.8. You're kidding. No, I'm not kidding. Wait, what? 3.8. I thought eight was pretty good. You were wrong. And so it was. It's kind of a joke this morning: if the KOSPI is indeed bursting a bubble, it's the cheapest bubble that's ever burst. Of course, it all comes back to the idea that usually the profits of such companies are reduced by a quarter.
Remember, Micron was losing a lot of money. And you know, when they were losing big money, people would say to Sanjay, "Why don't you spend more money?" Well, I'm losing so much money. What money should I spend ? And yet he continued to invest, which is why he feels better than others. We have even less capacity now, don't we ?
If we hadn't spent this money. Calculation. It's about computing and cybersecurity. Cybersecurity is what people think about first. We know this. But when I say "computing," I mean everything we're talking about. It's about expenses. Remember yesterday you said, "Listen, you're spending 50 billion—that's just the entrance fee?" Yes, they have an earnings report on September 2nd, so I chose an expiration date of August 31st.
I analyzed the expected movement. I sold a vertical call spread of 385/395. I got about a dollar and forty cents. It is currently trading slightly cheaper, but has an 80% chance of closing out of the money. So in the short term, if you want to hedge a long position so that the stock can go up and you stay in profit—that's a high probability short call spread that collects theta, and of course with limited risk, Joe.
When I look at something like this, that's all I want to do; if I'm neutral to bearish, I don't want to get overwhelmed right now. I want to give the stock room to recover not only before Nvidia's report next week, but also before their own report on September 2nd, of course, before this event. Yes, and I think when you sell that kind of premium before an event and can hold out outside of that event, you have to think about what catalyst could move the stock and put that call spread at risk.
And it will be over before the Broadcom report, so you do n't have to worry about that. And right now we don't see much strength in semiconductors in general. In fact, they had a rough week or two. So, in my opinion, these stocks were probably a little overvalued by the negative. But if you think about what could cause these stocks to create a problem for this spread...I don't see many catalysts ahead of the report.
Coach. What is your approach to your deal example? I actually entered the reporting cycle for this company. And this is truly the moment of truth. This is a way to make money on a few things I see now. Namely, we've talked about this before, Diane, the idea is that they can form support around the 360 level, which we've already bounced off of twice.
And we are right at the 200- day moving average. There are a few things in terms of charts that are shaping up for this deal. Another thing I noticed was the skewness of call options. So, how much are investors paying for growth call options now compared to what it was over the last 30–90 days. And they start overpaying for it again.
So you see that a lot of people are betting on this stock to go up. And you can take advantage of this as an options trader by doing something similar. This is a butterfly strategy with a torn wing with an expiration date of September 4. So I buy the 360 strike. I sell two 390 strikes and then buy the 410 strike.
So I have a long vertical spread of $30, which I reduce in value by selling a short vertical spread of $20. Its value will peak at 390, which is quite close to the expected price movement on September 4. Maybe a little below that level. This gives you the opportunity, by investing about $8, to get almost triple the profit if we reach this goal.
Another important aspect of considering call skew is that I can sell two 390 strike calls at a higher price than usual given the bullish demand. I'm trying to take advantage of this, and if this move happens and I'm right, approaching the 390-395 level, certain things will happen. Those "in the money" options I bought with a strike of 360 will increase their delta as the price increases, and the 390 strikes will take a significant portion of the expected volatility, which is beneficial to me since I opened a short position on them.
Sure, you could implement it at a 1 to 2 ratio, but I Nvidia is approaching another earnings report. But the most important story may actually lie one level below the GPUs. For years, investors have considered memory to be one of the most cyclical areas of semiconductors. As demand grew, manufacturers increased capacity.
Eventually, supply caught up with demand, and prices fell. Margins shrank, and the cycle began again. Artificial intelligence challenges this patters models become larger and more complex, the amount of data they need to process and store is constantly increasing. This means that each new generation of AI infrastructure puts more pressure on the memory supply chain .
And now the supply is not keeping up with the demand. Building up modern memory capacity takes years. You can't just decide today that you want more HBM and get millions of additional units next quarter. That's why the current deficit could last much longer than investors initially expected. But here the situation becomes complicated.
When companies start to make excess profits, competition naturally emerges. If you get more than 18% margin, competitors don't need any invitation. They see opportunities. So the real question isn't whether more memory will eventually become available—it will. The question is whether a new offering will arrive before demand for AI slows.
This is the fight that investors should be watching. And so far, the demand side remains incredibly strong. Hyperscalers continue to invest massively in data centers. AI labs need more computing power. Businesses are beginning to implement AI into real business processes. And beyond current models, there are huge potential applications in robotics, self-driving cars, industrial automation, and super-scale AI infrastructure.
All of these areas require more computing power, and more power requires more memory, which is why some investors are starting to argue that memory is not just another component of an AI system. It becomes strategic whole equation right now. The company no longer just sells GPUs. It builds a complete computing platform around hardware, networking software, and systems.
And her next big opportunity— Vera Rubin. An important development is that Rubin is moving from a roadmap to actual production and implementation. This is important because investors have been waiting for evidence that Nvidia can successfully transition from one major architecture to another without losing momentum.
If Rubin is successful, it could be another big growth engine for Nvidia in 2027 and beyond. But there is a catch. Nvidia can only sell a system if supporting components are available, and HBM is one such critical limitation. This created an unusual situation. You may have customers who are desperate to buy Nvidia systems, but Nvidia is struggling to produce them. The bottleneck is still somewhere inside the memory system.
Comments 0