저자: MeetKevin · 외국인
요약 보기
Microsoft가 OpenAI 관계 변화 이후 AI 칩 투자를 가속해 Nvidia H100 등 칩을 대량 구매할 것으로 전망. Nvidia 실적 발표 전 기대치(매출 약 161억달러, 마진 72.46%)는 이미 상향 반영됐고 공급 차질이 가장 큰 리스크. 중국 규제 영향은 제한적으로 봄. 장기적으로 Nvidia 분기 매출 250억달러(약 900달러 목표) 가능성을 제시하고 Nvidia, Palantir, Tesla, AMD, Intel을 AI 유망주로 언급.
다듬은 글
A quick note: the Black Friday offer for Meet Kevin programs is now available at meetkevin.com. In this video, we're going to analyze NVIDIA's profits. This is very important if you have any exposure to the chip sector, investments in artificial intelligence, or if you are curious about what to expect from this upcoming earnings report.
Keep in mind that, after the madness of what happened this weekend, which left many people very sad with the exception of Microsoft executives, who were very happy about OpenAI's move to Microsoft, well, at least a considerable portion of OpenAI employees who are threatening to leave, in addition to those who have already left, it will probably mean more investment from Microsoft in spending related to artificial intelligence.
Now, keep in mind that Microsoft has its own artificial intelligence chips, but they are built with 5nm technology. They 're not as powerful as the Nvidia H100 or the new Grace Hopper 200 combination. These are excellent products. And the reality is that most companies right now are trying to get any kind of chip they can get their hands on.
Here's an example of what Microsoft is building. It's the Maya M-A-I-A 100 AI accelerator. There's also a CPU version that sits on top of this rack, which they 're also manufacturing, or rather, designing. Microsoft is using Taiwan Semi for manufacturing. But the reason I mention OpenAI at the beginning of this discussion about Nvidia is because I truly believe that, for this current panicked transition , I really want them to think about it .
This panicked transition is what they are doing regarding the relationship with OpenAI, which they will maintain at Microsoft with OpenAI from Microsoft. But they will move from maintaining the relationship with GPT-3.5 4.0 to creating their own version, possibly even stronger than what they have with 3.5 4.0.
Now, why am I bringing this up again in a segment about Nvidia? Well, because even though Microsoft is venturing into making its own chips on a basically obsolete chip wafer platform, or in the semiconductor space, the 5-nanometer chip platform, which is probably, to be honest, sufficient for data infrastructure anyway, is still an outdated chip.
You know, we're now getting down to 2 nanometers. Many of the advanced chips are being manufactured at 3 nanometers. The point is, however, that Microsoft is going to be, in my opinion, in a situation where they'll say, "Uh oh, in order to catch up, basically go from zero to the level of OpenAI, we're going to have to invest hundreds of millions of dollars, if not billions, in AI right now." And we don't have time to wait to get enough of our own chips.
Let's buy what we can get now. And that's going to be a combination of Intel chips, AMD chips, and Nvidia chips. In my opinion, they're going to take whatever they can get their hands on . This will be very similar to when Tesla buys H100. Tesla buying 10,000 H100s, thanks to Elon Musk's desire to accelerate FSD as quickly as possible.
Why not just manufacture more Dojo supercomputers? Because you don't have the time, you don't yet have the quality, you don't yet have perfection, and you don't want to hinder your ability to compete while you try to perfect the chip, when you could simply spend the money, buy one already made, so to speak, as it becomes available, and dedicate yourself to your true mission: which for Tesla is to accelerate FSD, and for Microsoft now is to create that competitor to OpenAI.
Let's be clear. I made a video about this before. Microsoft now essentially has de facto control of what remains of OpenAI. And the OpenAI brain that created OpenAI in the first place. So, basically, they have everything. Topic for another video: whether or not we should be afraid of that, general artificial intelligence, and so on.
I already made a video about that. They can just write it on YouTube, Meet Kevin, OpenAI, Microsoft and they'll see it. It's very detailed. So, watch that video. But now let's focus on Nvidia. Nvidia's earnings are tomorrow. The expected implied volatility for one day is 7.11%. It is approximately 2% higher than the average volatility.
I think if anything goes wrong here, you'll probably see either a 10% drop or a 10% rise. That's my assumption. Now, it could be in any direction. So, we have no idea. But, if I look at this and think, "Okay, if the implied volatility is only 7% on an AI bet right now, I 'm probably considering straddle-type options." This isn't personalized financial advice, but I want you to know where you stand in relation to Wall Street expectations and where we have some new expectations for Nvidia.
So, let's review all of this. We'll also be reviewing some price targets. Now, the first thing I want to do is analyze their latest press release, because there you will see what their previous estimates were. And their previous estimates for this quarter match what Wall Street is valuing right now, so you'll see how excited people are, okay?
Here are the estimates. Wall Street is currently valuing $16.1 billion in revenue. They gave a projection of 16 billion dollars. So, they are leaning upwards on that, and as for the margin, Wall Street also values a positive margin of 72.46 against the guidance they offered. Therefore, they are already leaning upwards in their estimates.
I think the biggest risks, aside from not meeting these estimates for Nvidia, will be what they tell us about the offer. For Nvidia to continue generating profits, we need more and more supply. In the latest earnings call, Nvidia told us that it expects more offerings virtually every quarter for the next year.
But if there's a problem with the supply, the share price could fall very quickly. It's well-rated right now , and that's an optimistic rating based on what we see. That does n't mean they can't surpass an optimistic forecast, it's just harder to surpass something optimistic. It's easy to overcome something pessimistic; it's more difficult to overcome something optimistic.
Personally, I'm not that worried about the China problem and the sanctions. We are still waiting for more information on what will happen with that chip debacle for selling to China, which currently represents 20 to 25% of Nvidia's revenue. Nvidia has already told us that they do not expect the impact of the sanctions to be material, given that their revenues are increasing explosively.
I mean, more than $10 billion per quarter in data centers alone. It's absolutely insane. And the market continues to grow. Despite increased competition from AMD, Intel, Microsoft, and Google, with all these companies creating their own chips, the market is growing substantially. We'll review some estimates in a moment, but I want to be clear.
Personally, I think it's good news that the gaming sector is overcoming a bad streak . China doesn't worry me that much . The offer, that would be the big red flag. If we have a delay in the offer, that will give competitors more time to catch up. Note that Nvidia is also moving from an 18-24 month renewal cycle to a 12-month cycle, giving them a slight advantage over their competitors.
It's like getting a new iPhone every year and trying to stay ahead of the competition, while the competition is now basically trying to outdo you. And that's a little more difficult to do. But look what Goldman Sachs has to say. Goldman Sachs currently has a price target of $605 for Nvidia. They maintain that this represents approximately 24% upside potential for the stock.
They believe you will see a big change in spending. This is just a fancy way of saying, " Look, Google spends $20 billion on plant, property, and equipment in one quarter." Okay, in a quarter that's almost a billion dollars a day allocated to plant, property and equipment. How crazy is that? Simply write those checks every day.
I don't envy their accounting department. Anyway, this is a massive amount of money, and Goldman suggests that you'll see much of that investment moving away from regular data centers, general-purpose computing, storage, and cloud computing, right? And it will be transferred to artificial intelligence.
Google already reports that 80 % of its users who publish ads are already using some form that it's not just a matter of buying the chips, it's also a matter of running these machines efficiently. So, for example, if you have server racks like this and on top you have, say, an AMD CPU, or maybe you use the GH200 CPU.
And then on top of or underneath that it has a bunch of H100 GPUs or A100s or whatever. You need to connect these server racks and connect them internally with the highest quality data infrastructure possible, which is basically a fancy way of saying you need the good stuff, the good Ethernet cables, okay?
And guess who makes the really good server Ethernet cables? It's a product called Infiniband. Infiniband is basically your data center-grade Ethernet or robust, high-speed cabling for AI with data center quality. InfiniBand is, once again, very, very similar to Ethernet. I want you to see what it looks like because I think it 's great.
This is what it looks like. Frankly, it looks like an Ethernet cable fitted with a different connector. This is just a quick Google search of what InfiniBand looks like and, you know, look at this, 200G, man, 200 gigabits of data. Let's go all out! That's incredible. But what do you need to know about this?
Nvidia owns it. Yes, Nvidia owns InfiniBand. These people are absolutely brilliant. You have to remember that about Nvidia. And look, this stock could plummet horribly if they fail even a little. So I'm not here to, you know, get on my knees for Nvidia. I'm just saying, it's fine, they have the cabling infrastructure.
They have the software moat called CUDA, not Cuba, CUDA. That's their software moat. If you want to create a cutting-edge artificial intelligence platform, you 're probably building it with CUDA software. And now they have both CPUs and GPUs with their Grace Hopper 200. This, this is something really important.
The problem is that Wall Street is already including inflated expectations in the price. So, yes, you could be harmed by this. Am I going to trade based on these results? Absolutely not. Am I long on Nvidia? Of course I am , hell yeah! Do I think Nvidia, Palantir, Tesla, AMD, and Intel are your best bets in AI?
Of course. I think Microsoft probably has some good things going for it too? Yes. Like what we talked about this morning. We touched on this topic this morning in the live market opening broadcast , which, if you're not already familiar with it, you can join completely free of charge every day the market is open.
Just type Meet Kevin Live into YouTube, subscribe to that channel and you'll see the live streams . You can join them every day, at no cost. It happens just before the live broadcast for course members. And I'll also be live soon on the Meet Kevin Politics channel if you want to watch it, or on the podcast channel.
They are all linked below. But anyway, regarding Microsoft, it's worth noting that I personally believe there are some disadvantages for Microsoft going forward. Don't get me wrong, I absolutely love Microsoft, but this is what I see when I look at Microsoft. I see risks of deflation leading to lower rates , which in turn leads to a desire for growth at all costs.
And as you move away from an interest rate-sensitive environment, you move away from Microsoft as a safe bet, which is a very robust safe bet and has performed very well. And I believe your growth will potentially be much greater in these bets I just mentioned. Now, if inflation stays high, you'll be much better off at Microsoft.
Okay, that's my thesis. Personally, I believe inflation will plummet, and that's why I like Nvidia, Tesla, and Palantir. Very good, excellent. So, what else do we need to know about Nvidia? Okay, let's continue with the analysis of Goldman Sachs. We have Goldman Sachs' price target of $610; Comments on infrastructure continue to support his track record.
As for video games, they are overcoming the lowest point. Investors are expected to focus on new supply chains; We already know that. Here is your forecast. They believe that if one is pessimistic and eliminates generative AI, Nvidia still generates about $10 billion per quarter in data centers, which is basically where we are now, about $ 40 billion per year.
Personally, my target is 100 billion annually, which is approximately two and a half times the current estimate, and is about ...let's see, 100 divided by 65, which is Goldman's bullish target. I am 50% above its bullish target. So, if I were to translate that into a target price, it would probably be around $900 for Nvidia in the long run.
That wouldn't be my 12-month goal. Although short-term market euphoria might value Nvidia at 900, if that were to happen in the short term, I would probably sell, because it's definitely a long- term target for me. Reaching $25 billion per quarter at Nvidia will likely be a three- to four-year move.
You really have to double it and then add another half. So I think that will take a little while. I would give it a growth rate of 30%, probably a little more to get where I need to get. If I calculate a growth rate of 30% for 3 years, I arrive at about 22 billion. Yes, I'll probably have to be at 35% to reach my goal of 25.
Yes. Yes, you have to grow close to 35% annually to reach my 25 billion per quarter, but I think it's realistic. We'll see. Anyway, that's my belief, and I do think there's a possibility that the market will value that euphoria. I still think there are a lot of bears regarding Nvidia. They think AI is a passing fad, which, don't get me wrong, consumer AI is to some extent, but I don't care about consumer AI.
I care about enterprise AI because that's where the savings will be and where the capital investments will go. So this is my opinion. By the way, I'll be live tomorrow afternoon at 1:00 p.m. Mark your calendar. 1:00 p.m. November 21st. Mark your calendars: live stream. It will be a live broadcast at the close of the market.
Meet Kevin live. Well, that's where Nvidia's profits will go . We'll cover it together. Ready? Get ready. Come on. And of course, my usual legal disclaimer. over and utside of the Houzz Law, nor do I currently act as a market maker. Boom! Well, there you have it, folks. Thank you so much for watching.
I'll see them on that live stream . Remember to go to meetkevin.com to see my Black Friday deals on amazing wealth-building programs . Bye bye.
Comments 0