저자: Kevin · 외국인
요약 보기
에이전틱 AI 확산으로 CPU 수요가 급증하며 AMD·Intel·ARM 등 CPU 관련주가 강세. 엔비디아도 CPU 시장 진입으로 AMD의 경쟁자가 됨. Meta 매수 근거 제시. AMD는 PEG 기준 단기 과매수지만 목표가 944달러 제시. 메모리(DRAM) 가격 상승 지속 전망이나 메모리주에는 유보적. 자막 중 일부 구간 누락·오인식이 많아 수치 신뢰도 낮음.
다듬은 글
Adamus has led to an explosion in CPU-related stocks, and guys, the stock market is full of opportunity right now , much more than just Meta and AMD stocks, which just blew up the market today. AMD is up 9.95% today. Officially, the market capitalization reached 1 trillion. We made a forecast for this stock when it was worth $452, and we told the course participants: " Friends, this asset is about to bounce off this line." Deliciously.
I didn't expect it to take off so quickly. This is what happens in the market. The market can move very quickly. You have six-figure, and possibly multi-figure, exposure on Meta. You have a great bounce from the legendary 544 mark and a breakout of the downtrend . A great Meta move up 11.4%. So many people are wondering what other promotions are available and is it too late to enter?
Well, in this segment, we'll look at other investment opportunities and figure out what's going on. So, the first thing you need to know is that in my opinion, this whole AI chatbot thing is a big opportunity for advertisers. I'm just opening the App Store because when I open it, I usually see Metam Muse ads popping up all the time.
I think advertising is a critically important place to promote these applications, amidst all this AI competition. This is just a small part of what we will discuss in this video, but it is worth noting that the more agents review products for you, the fewer human eyes see them. So whether it's DoorDash or Wingstop or even an Amazon store— which is why Amazon is now banning the Metam Muse app from crawling their site—there are fewer people looking at those sites .
And so, the value of advertising in app stores is increasing, where people actually download things, which is profitable for Apple or Google, where people watch entertainment content , whether it's Netflix or YouTube. For some reason I can't open the App Store right now. I keep getting some kind of nonsense, it's typical Apple.
I type " app" to open my apps, but all I see is Google Apple, or, well, the Apple store to buy the iPhone Duo. I think it's a sign, but it doesn't matter. I notice that many of these apps are competing with each other for advertising space. So, you see Metamuse ads inside the app store. You see GPT being advertised there.
And what is happening? Oh, there it is. What happens is that there is competition for users. Now even Gemini is being advertised. They are all advertised. So, you open the app store today, Google Gemini launches apps. Metamuse used to be there. They alternate them among themselves. Chat GPT is also there .
If you go to the best free apps section and look at the top apps in that category, you'll see that Metamuse now tops the free apps chart, ahead of Chat GPT, which is now in second place. And you also have prediction markets in fourth and fifth place. You don't really see Gemini until the eighth. You don't even see Claude from Anthropic until you get to number 16, which is crazy.
Instagram is in 18th place. Metamuse got the first one. Now, one of the reasons I mention advertisers is because they do n't just advertise on these platforms, whether it's app stores or whatever. They also rent billboards. They buy spots in influencer advertising campaigns. These apps compete to attract as many users as possible to their platform.
I actually think so, or at least as many users as possible. And I really think this will be a boon for advertisers and promotions. This is one of the reasons why I bought Meta shares, because I thought: well, Meta not only pays for the development of its AI, but also benefits from the graphics cards it buys, having over 4 gigawatts of computing power.
The cost of these video cards has only increased. And they benefit from AI performance, and their Muse launch is just incredible. And we haven't even gotten to the Watermelon release yet, which is expected next month for their advanced language model, their advanced LLM . So, what other stocks could benefit?
Well, in addition to the ones we just mentioned, I think advertisers have opportunities. Whether it's Google or something like AppLovin or Netflix, I think they have prospects too. But let's take a closer and deeper look at AMD and Intel themselves, as well as other players in the hardware space. So first, you need to understand that when you advertise these agents, they are a bit different from video cards and chatbots.
Usually when we think of GPUs, we imagine learning and, uh, responding to broader contextual sets in chatbots, and working, how to say it, on the front lines . For example, our latest Reuben chip has ten times the bandwidth capabilities of hardware like the H100 chip previously used by Nvidia. LLMs are trying to acquire as many GPUs as possible, but many processes in the field of artificial intelligence run very efficiently on CPUs.
One such task is agent workloads, and that is what Metamuse is all about. That's why many people are now rushing to buy CPU stocks instead of GPUs. At Reinvest, we use a lot of CPU-based machines to run our AI. As a startup, we've seen for many months, since about March, where it's going—this shift to using more and more CPUs, which is why Nvidia got into this game.
Many forget that Nvidia, although it is the main player in the GPU market for us, is counting on getting ahead of its competitors in the CPU space, such as AMD, who specialize in this very field. Although AMD also supplies GPUs. Let's talk about numbers. Nvidia predicts CPU demand of $20 billion for the 2026 calendar year alone.
That $20 billion in CPU demand alone would now be equal to about three- quarters of the gross profit that AMD already makes. This is quite interesting, but many don't consider Nvidia as a player in the CPU market. And I think that's partly why Nvidia is a little behind. Look at this. Nvidia just got back to the 727 mark today, which is great .
We mentioned this this morning in our alpha report. We also predicted that Meta would break through to 700 this week, and it reached 741, which just blew me away. I was expecting 700. We wrote it down. We even created a coupon code in honor of this called "Metaballs" which will only be valid for 12 hours.
Actually, it's probably about 10 hours now. We're deactivating it tonight and also officially raising the price to $900. But those were the predictions we made this morning in the alpha report. The peak of fear was two weeks ago. If 725 holds and Nvidia holds 720. So 725 on Qs holds and Nvidia holds 720.
We're going up. We'll be back to 735 soon. We guessed. We broke the 741 mark on Qs, which was extraordinary. On Friday we had our endurance test. We had Meta. We predicted 700, even though it was Friday, and I even said so in a public video. I'm like, "Dude, it's at 666, and I have a target price of 700 by next Friday." Dude, it's 741 on Monday.
I'm like, " Damn, this is crazy." We noticed Hood testing the 127 mark. It wasn't an option I was seriously counting on, but I remembered it, so I included it here as a oversight. And the next two that we're really looking at are the bullish buybacks between now and the midterm elections, and I think by Black Friday we'll be hitting all-time highs.
In fact, we are now only 1% away from historical highs. So I feel like it's a little bit outdated, but we were right to be optimistic, and I would love for you to become part of the alpha membership, you can join at meetke.com. I'd be happy to see you there. We have a lot to talk about in our alpha report tomorrow morning in addition to our analysis.
But take a look at some of this analysis. So, if Nvidia is predicting... how better to say it? 20 billion in CPU computing sales by the end of 2026. It is worth noting that AMD's total sales volume for 3 months is about 11.5. So Nvidia is planning to make AMD's sales in 6 months in a sector where a year ago they had almost zero revenue, which actually makes Nvidia a real competitor to AMD in the CPU space.
I'm not trying to criticize AMD. I think AMD still has potential. I mean, just think about the benchmarks alone. Their Helios system vs. Vera Rubin: 50% more memory bandwidth, 50% more scaling bandwidth, and about 10-15% more tokens per second. So, the Helios rack has become a great product against Vera Rubin, especially because people need these processors right now.
They have a really high-quality product, and in some tests AMD even outperforms Nvidia GPUs. Vera's part of Vera Rubin's system, their fifth-generation Epic Turin chip, is now projected to account for 50% of the total TAM (total market volume) for CPUs in data centers. And that's why people are going crazy here, because it's an x86 product, which means you don't pay ARM licensing fees.
ARM shares rose significantly today, as companies like Apple also benefit from the development of processors. By the way, remember that OpenAI has purchased tens of thousands of Apple Mac Studio and Mac mini for agent-based artificial intelligence, and ARM licenses the architecture on which these processors run.
So, their shares are up 17% today. Not many people actually talk about ARM compared to AMD. AMD's market cap is around $344 billion, and they are also a competitor. The more Nvidia and Apple become competitors for AMD in the processor space. AMD is doing x86—that's Intel's style, and Intel Corporation is up 14% today, ending the day up 12%.
So for AMD, think x86 and Intel architecture. For Apple as a player in the processor market— think ARM chips. And of course, don't forget that Apple itself also benefits from advertisers in the App Store . Nvidia is also trying to gain access to this market because it understands the potential of the transition to agent-based artificial intelligence.
Their current goal is to reach 50% of the processor market share by 2030 . Again, this figure is slightly lower now. They are at around 30 or so percent. However, they expect significant revenue growth. AMD's revenue is forecast to grow by 72% in the period to next year. Total annual revenue is expected to reach $88 billion.
If you compare this 88 billion per year only with the income that Nvidia receives from data centers, then Nvidia earns as much as AMD as a whole company in one quarter. So it's worth understanding the comparison of these players. Nvidia is now five times larger in market capitalization than AMD. We'll compare them in more detail later, but that's why people are thinking: if the future is in processors, and Nvidia is a $5 trillion company, what's stopping AMD from growing as well ?
Nvidia sees these trends, which is why they are also entering the processor market and should be considered as a player in this area. So we have AMD, Intel, ARM, Apple, and of course Meta, which is causing this hype today. And Nvidia, what else do we have? We'll go through the numbers so you can get more context.
I just don't want to get too deep into the numbers because sometimes they get a bit boring. But let's take a moment to consider who else is worth considering. Of course, Intel. We've talked about them, but why are we even paying attention to Intel? Since Intel is projected to double its processor sales between 2026 and 2027, maybe even both years.
Their data center and AI revenue in the second quarter of 2026 has already grown by 59% year-over- year. And this is only with a 9% increase in the number of units, which is simply crazy. So you put in 9% more processors, but your prices increased by 48% to cover the difference. This is pricing power.
So Intel is also observing pricing power. So, they are not only a manufacturer, but also a chip designer that makes their own processors. Therefore, this is a big win for Intel as well. And, of course, we have serious supply constraints on all sides. And this is where Nvidia has a slight advantage. Uh, and that's why I do n't want to talk too much about Nvidia here.
I own shares in many of the stocks we mention here, so I'm not playing short anything . I'm pretty optimistic right now. I just want to be clear, you know, I'm trying to voice my biases. I have a stake in these things, you know? But I want you to see the commitment or the difference in commitment between these companies, because I think it's important.
AMD has commitments from today through the end of 2031 and beyond. Most of the commitments fall on the initial period. AMD has approximately $30 billion in liabilities. it's a commitment, as they say right here , to purchase wafers, substrates, and components from third-party companies. You need components to actually make processors, you need to have inputs, right?
So they have about $30 billion in liabilities. Nvidia, however, has $334 billion in commitments for around the same time, in the same timeframe. So, this means that Nvidia secured 11 times more supplies, being a 5 times larger company. Again , I think you can see here that this is not going to be a one- horse race.
This will be a race of many participants where they grab as many resources as they can, both for CPUs and GPUs . To be fair , for Nvidia, a big part of this will be about memory. A significant portion of this will also be part of their GPU business . But memory is very important . I mean, if I open up the memory expectations right now, I have a screenshot here, uh, just with the prices.
Oh, actually it's about processors. This is processor pricing. Intel CEO emphasizes the acute shortage of server processors , as Intel's available capacity meets only 50% of real customer demand. According to reports, Intel will raise prices for PC processors by about 10%. And even greater price increases are expected for data center models.
The reason for this difference is that these are different lines and different product quality. The PC processor market is relatively weak. The gaming industry is actually in decline for most developers and manufacturers, including Nvidia. But what is growing rapidly is data centers. So the fact that Intel is raising prices on consumer processors is indeed surprising, but it just shows how serious the shortage is.
In any case, DRAM prices are expected to peak only at the end of next year. So , memory prices continue to rise . That's why AMD and Nvidia are joining this trend, trying to provide as much computing power as possible now. And when I say " computing power," in this case I mean DRAM and the components to power it.
But it's Nvidia that does this. You know, they get twice as much adjusted for size as AMD, probably because they have more ability to pay and they can write bigger checks. I find this an interesting comparison, just as it is interesting to analyze the income statement of both companies to see the margins and their growth.
AMD's margins have grown rapidly. We went from 39.8% margin to 53.7%. This is truly impressive. But it is worth noting that Nvidia has 75% of the gross profit. And AMD just came out of the red with a net profit of 17.2%. Nvidia receives 63% of the net profit. So the numbers are drastically different.
AMD's operating profit for the three months was only $2 billion . If I look at Nvidia's earnings report, I see operating income of $63 billion . This is 30 times Nvidia's quarterly operating profit. And I'm not saying this to say that Nvidia is better. I say this to explain why the market believes AMD has so much growth potential.
But at the same time, I feel like Nvidia is also being left out a bit . It's like she's not participating in this party. Nvidia's valuation is very low right now . This is an estimate that is only a fraction of AMD's estimate. AMD is currently trading at a PEG ratio of around 1.5–1.6 at the price it just rose to.
Nvidia is trading at around 0.7 . So, AMD shares are now more than twice as expensive as Nvidia shares. Is it possible that Nvidia will catch up, or will AMD slow down? Perhaps. I'm not saying that will happen. I wouldn't bet on a downgrade. And the reason is that we are becoming a bit overbought. And the reason why this might be desirable is because we are becoming overbought.
I'm not saying it's time to sell them. I'm not going to play for relegation. But if you look at the daily chart, AMD is now significantly overbought— nearly 73 on the relative strength index. And on the weekly chart, you just crossed the 70 mark to 70.1. So, this is a bit extreme. And if you compare it to Nvidia, I have n't done that yet.
Let's take a look. Nvidia's average weekly performance is around 59, and around 58 per day. So, you can see, based on relative strength, how Nvidia is lagging behind. But those who are really growing are AMD and, for example, Dell . Okay, Dell— mostly due to local deployment of CPU and GPU. Quite impressive.
Okay, let's move on. So, let's go back to Intel for a moment, since we're jumping from topic to topic here. Data center customers are going crazy for processors, in part because of agent-based artificial intelligence. We know that Intel's capital expenditures are growing rapidly, as are almost all infrastructure projects.
Whether you look at Meta, which still has 15 billion in free cash flow, which is actually impressive. But Intel is an expensive manufacturing development where you take all the money you earn and invest it in building production capacity. AMD shouldn't do this because AMD can produce products together with Intel or Taiwan Semi.
And the advantage of this is that AMD will almost certainly always be more profitable than Intel. There is something to think about. This is not a criticism of Intel. It's just a comparison. ARM also has a pretty high margin because ARM will probably reach around 17–18% of the total CPU market, and I think they will grow to 40–50%.
And you will be competing with AMD for x86 architectures as ARM will likely remain the product of choice for Apple and Nvidia and they will come into the CPU market. Everyone knows that you can make a lot of money on processors, so the competition is coming. To be fair , the main players right now are AMD and Intel, and I think that's why they're so successful right now.
As for partnerships, I think this is very interesting. So we have Apple and Nvidia with ARM. Google is building its own chip. Axion based on ARM. There is AWS Graviton. There is Microsoft Cobalt. They are all based on the ARM architecture. ARM also collaborates with Broadcom and Qualcomm. They collaborate everywhere .
AMD also has some promising partnerships. Look at this. AMD has offered call options, known in corporate circles as warrants, to Meta and OpenAI. This means that Meta, amid the rapid rise in AMD shares, will likely now have to recognize a profit on its AMD options, which allow them to buy AMD shares at one cent apiece .
They can implement this at any time: for OpenAI, from now until October 5, 2030, and for Meta, until February 23, 2031. So Lisa Su is reaching out to companies like Meta and OpenAI and saying, "Hey, could you buy our six gigawatt chips? In exchange, we'll give you stock." This dilutes shares, but to be fair , it's just a different form of cyclical financing compared to what Nvidia does, which invests excess cash flow, which they have significantly more of than AMD, directly into the company, thus creating demand.
Take a look here: this is Nvidia's cash flow statement. Nvidia's cash flow statement : If you look here, the purchase of stock and debt securities—they literally bought $42 billion of other people's stock or $21 billion of other people's debt in just the six months ending July 20th. Their free cash flow is $70 billion.
AMD can't afford that. They simply don't have such assets. If you look at their free cash flow, it's $4 billion in six months. This means that Nvidia is 17.5 times richer in terms of cash flow. So AMD is like saying, "We'll just strip down." "We'll just give you some of our stock instead ." So, you know, we can't do you a favor.
We'll just give ours. We, we'll just give it to you in a different way, essentially. Okay, that's acceptable. AMD doesn't have the same margin performance as Nvidia, but it's trying to achieve it. That's the whole point. Again , we've already looked at these margins here. 62% of net profit compared to 17% for AMD.
This is the growth side. People think Nvidia has reached a ceiling in terms of margins and the volumes it has access to. But people forget that Nvidia has an entire processor business that is starting from scratch to compete with AMD. Remember, Nvidia's CPU sales are expected to be 20 billion within the scope of integrated sales , where even if we double the data center sales right now, we'll go here and say double it and cut it.
So, let's take the quarterly revenue from data centers at 6.6. 7. Let's take 6.7*4. That's 26.8 billion. And then let's just double that. Okay, let's be very generous. That would be 53 billion in processor sales in 2027. Nvidia expects to sell for 20 billion in 2026. Let's say they increase it by 50%.
That will be 30 billion. AMD actually has or AMD actually has competition breathing down their necks for these processors, and that's Jensen. He doesn't want to lose to Lisa, and I get it, you know, they're related, which is the craziest part. But it's like a horse race nostril to nostril. This is very interesting to watch, but AMD will be the main beneficiary of the margin growth.
And you actually see that comment right here. Look at this. Uh, AMD Instinct, which is their GPU product, was recorded last year, uh, or a larger share of those sales, which led to, uh, higher data center revenue and therefore a favorable product mix for margins. In simple terms, their margins have increased because of these products, like more sales for data centers.
Of course. You will make more money selling to data centers than to consumers. Of course. Now again, this is the type of analysis, by the way, that we try to do in alpha membership. If you're not with us yet, remember, go to meetKevin.com and join us. You get all nine courses. You receive an alpha report every day.
People leave comments in the chat: " Alpha membership paid for itself in about one transaction." Obviously I can't guarantee or verify this, but it sounds really cool. Uh , these are just some of the predictions we made in the alpha report today. Uh, you know, also based on all the predictions we made on Friday.
And so we're going to, uh, raise the price by $900 for the alpha membership. So you should make it by midnight tonight. And this is a meta balls coupon code if you go to meetc.com. We changed the price and then looked at Meta shares. We are like this: we don't know. Should we do a quick sale? So we're doing a quick sale on it.
This is a victory for you. This is a victory for us. This is a win for Reinvest. Everyone wins. Do you know how bad that is? Good. Let's continue to look at this. AMD also acquired ZT Group, a company that helps create a more versatile computing infrastructure for hyperscale businesses. In simple terms: yo, we're increasing our appeal to data centers because the margins are good there, man.
There doesn't have to be just one winner here. This is a big win for AMD. I'm a little worried, I have to be honest. I wouldn't add AMD at this stage. It can continue to grow. It's just that right now she seems a little overheated and somewhat euphoric. I think there are other possibilities. There are a lot of them, at least Apple.
Apple is a bit expensive. You can look at some ASIC manufacturers, or I would say designers, because most of them use what is it called? TSM for production, I do n't have that. Good. I do n't have what I'm going to show you. So I don't...I think it's important to share your bias. I think most people don't do this often enough.
The last thing I want is for someone to think, "Oh, Kevin is showing off his assets ." All my stocks could depreciate to zero, it doesn't matter. If all my real estate depreciated to zero, I would cry, but I don't think that will happen. I think we bought properties in very good locations. But anyway, Broadcom, man, it's right at the 355 mark.
It turned 355, but in my opinion, this is the next breakthrough. This is something you should really pay attention to. Cerebrus, I also think, has hit rock bottom. I do not own Broadcom. I have a stake in Cerebrus. A little, just a little. You know, you kind of average out the cost of entry into it.
I think it has become attractive to buy below 250. So , 250 and below. I think Cerebrus is worth buying. I don't think she'll stay there long. Robotics is the next direction. Forget about agents for a moment. Robotics: You need Cerebrus and Nvidia. These are your main bets in robotics. And there are many others.
We have them in the research tab of the stock app, other companies benefiting from robotics. But these are two really attractive options. Okay, what else do we have? So, let's get this straight. Muse moves in a parabola. It was launched on September 8. It's been about two weeks since it came out and it's working really well.
He sends emails for people, books trips. I haven't tried the desktop app yet, but it seems to have a PC version as well. I only used the mobile app. That's pretty good. advertisers here, when we were talking about the fact that we made a prediction about Meta on August 28th that this stock was about to take off, and it did.
We made this prediction when the price was around $566. I mean, you can just look, go to my video, type in YouTube "meet Kevin Meta stock is about to skyrocket" and watch the video from August 28th, there's a lot of detail in there. It was right there. It wasn't a particularly interesting day or anything.
That is, the stock moved 0.27 that day. You know, two days later she fell. It doesn't matter . But, uh, but the rapid growth still clearly happened. There are still opportunities here. We talk so much about AMD and Nvidia , and I don't want people to forget that Meta is still very cheap in my opinion .
The next stop is here—826. We did make a temporary line here at 732, but we already broke through it today. I don't even think we stopped at that line. Yes. So, this line is currently unsuccessful. We drew it today as "possible ". No, she looks bad right now. Uh, 735 on QQQ was another one that we completely crushed today.
And at least we got a generous stop on my line . It's another gray line, so it's not perfect, but we got a good stop right there. But the 700 was great. It is worth updating this line. It was wonderful. We bought on the dip at 700 and sent a buy alert to everyone on this position. Uh, what else do we need to know?
So, Muse for Mac, uh, and a bunch of third-party Oh, Muse for Mac came out today? Wait. There's a note here that Muse for Mac was released today. Maybe that's why I haven't had time to try it yet. Meta's Muse has arrived on Mac. Oh, three days ago. Oh, okay. So , that's why I haven't seen it yet. OK.
The desktop application is out. In any case, for me it's a stock worth $1,800. Meta has so much money now. So, my target price for Meta is around $1,800. If you look at AMD and other similar assets right now. Nvidia, I expect over $600 per share, and yes, that would probably make it the first company with a market cap of 10 trillion.
And, honestly, she deserves it. If I look at AMD, we have 747, which means we're currently trading at 82 times 82 , and the growth rate is the current forecast divided by 4.55, 82 divided by 55. So that gives me about 1.5 times. This is what we talked about earlier. But it still provides potential for growth.
Its margins are n't even close to as good as Nvidia's, so you need to lower your growth expectations slightly based on the PEG ratio, but it still justifies a stock price of around $944. So you see that a lot of stocks, like Nvidia, I think, have already doubled . I think Meta could easily double. For me, AMD still has 50% upside potential.
Oh, she just has a lot of momentum right now. Okay, that clarifies those targets. I know some people have had questions about this. Let's look at this...oh, yeah, one more thing. AMD Venice, one of their latest products, is expected to deliver 2.8 times more agents per watt than Vera. Although, to be fair, we're comparing newer with older, so there's still some work to be done, even though Vera is still a relatively new chip for Nvidia.
But that's the aspect of racing. We talked about Helios, which is a rack system. Nvidia probably wins in full rack systems. And right now, there's such a deficit that they're, frankly, all winners, so it does n't matter. We talked about Intel. We talked about ARM. We talked about the $20 billion forecast for processors from Nvidia, which is a "bullish" signal.
Just remember that Nvidia's growth rates will likely look higher because they're starting from a zero base. So you start with essentially infinite growth. And now Jensen comes out and says, "Yeah, the 70% growth we expected for the business as a whole will probably end up doubling." Initially, there was a problem with supply constraints.
Another thing I've talked about a lot also has to do with memory. Modern GPUs are becoming so efficient at processing bandwidth that they no longer need as much high- bandwidth memory. But, but, but, but, but it's worth saying something about agent-based AI, which is actually different. So we have to discuss this.
This is important. OK. So we'll draw this together . Let's say you have... I'll just use the Blackwell 6000 because I know it has 96 gigabytes. Reuben seems to be, uh, Reuben is about 228, right? It doesn't matter. I'll just use this. Let's call it 228 gigabytes of RAM. Okay, I could be off by a few gigabytes, but let's say we have 228 gigabytes of RAM on the Reuben chip.
For LLM, you may only need to start Frontier LLM. You might only need this part here. So, let's call it the loaded Frontier LLM. OK. Uh, now, what makes AI hallucinate less is the presence of context. Okay, context goes here. Okay, so this is a model. This is part of your VRAM for the model. This is your contextual part.
Most text threads today don't really hit memory limits. So what's happening is that you're filling up, maybe that's what it means, you actually have a lot of what's called "hanging" memory. You just don't use this section. So now, there's something particularly important for agent computing that actually plays into the memory side of things: you come in and say, " Okay, we're not going to let this all hang, because we've already loaded the model layer, and we can achieve this with a single set of agent computations." What can we do now that we have all this extra stock that is actually beneficial for high-bandwidth memory like SK Hynix or Micron?
We can run what's called parallel processing, and we could go here and say, " This is going to be agent one, and this could be agent two, agent three, agent four, agent five." And now they're all using the same loaded LLM model, but we're actually executing and running five different contextual inferences at the same time.
So your chat could be here, Kevin's chat could be here, your mom's chat could be here, your dad's chat could be here, and your coupon chat could be right here. Of course? So you can get five different agents running on these chips with more RAM. And while there is this fear about unused compute power, which is true for training advanced models, it's worth considering the counterargument: agent-based AI can help you utilize that unused RAM, which in turn contributes to greater optimism about rising DRAM prices.
DRAM prices increased by about 95% in the first quarter of 2026 , right? They are predicted to grow. Although the pace will slow down. The peak of price growth will likely occur at the end of 2027, as significantly larger volumes of supply will begin to enter the market. DRAM is finally becoming a commodity, but to be fair, all three DRAM manufacturers have sold out of all their high-bandwidth memory .
And it seems that SK Hynix's capacity will be the last to come online, sometime in 2028 or 2029 . So there is still a lot of work ahead. So there is still time to invest in memory. This is another beneficiary of the AI stack for agents. So think for a moment about all these agent AI money-making opportunities , because how many people do you know actually use agents?
This is probably only a fraction of those who could use them. So I believe we are at the very beginning of this stage. We are in the early stages for Meta . We're still early days for AMD, although I believe the stock is overbought in the short term . Early stage for Nvidia, which sounds crazy because it's a five trillion market cap company, which means it's early stage for Intel and ARM.
Regarding memory manufacturer stocks, I have somewhat conflicting opinions. I see arguments in their favor. But I'm still hesitant about the memory. And this is an early stage for advertisers. I mean , eventually everyone will have a Grokbot and all that. In fact, I think that's why SpaceX fell a percentage today, because Meta has shown such great results.
You know, people think, "Oh, everyone's going to use Meta." This is probably bad for SpaceX. This is a race. It does n't really matter." You know, the real winners are the chipset sellers. I think it's AMD and Nvidia right now, but that's pretty incredible. So this isn't the boom that's close to the end of the cycle.
I know some people are like, "But Kevin, you made a video where you talked about two paths to a recession." And it was. And in it I said that I buy in a recession because there's no sign of two paths to a recession, but we should be mindful of those two paths so that we can be prepared and know when things turn around.
That's a major red flag. So join me at meetke.com. Use the metaballs coupon. One other point I want to make . I just want to thank all of you for being here. I mean, I don't know who's going to listen to this long rambling about AMD and agents. If you did, make sure to leave a comment. But thank you so much for being here and support me.
We've been releasing a lot more content over the last month. I'm having more fun posting videos on YouTube than I've ever had. Our views are almost triple what they were. Our impressions are tripled. Our views are more than doubled and are approaching triple what they were. So we're doing great. You know, it helps the startup grow .
It helps the channel grow. And I really enjoy sharing these insights and information. And I think it also makes our Alpha report better every day. The more we get involved, the more we see and the more we can factor into our predictions. Of course, it's impossible to be perfect. We're not always perfect, but I think we're getting a lot of wins so far.
I'm really happy about that and I hope you can benefit from that . So, as always, I'll take the hit so you can make money. Oh my God, poor Kevin just lost a lot of money. Oh, but it's nothing. I'm with you. With you. Oh my God, poor Kevin just lost a bunch of money. Oh, it's nothing. I'm with you. Let's go. Come on . Push. Buy on the dip. Buy on the dip. Let's go. .
Comments 0