Chip Stocks Are Crashing After the Best Earnings Report I've Seen

Hey everyone and welcome back to another video for today. So in today's video we have to talk about the recent chip selloff and it's a selloff happening despite good numbers from air systems ASML TSMC but ASML and TSMC super important players in this whole AI semiconductor space and yet the names the microns the Intels the neoclouds actually all of the semiconductors are down quite a lot over the last couple of days. Of course, year to date, a lot of them are still up considerably, but right now there is a selloff despite us getting a lot of good information. Now, there are a lot of headlines out there as well. Coreweave, memory names, Penguin Technologies. We have a lot to discuss in today's video, so let's try and cover as much as possible. Of course, we're also going to cover the new announcement from Nebu. Some people say, "Oh, they're changing their business model." No, they're not changing their business model at all. They're just trying to accelerate things in a not a weird way. It's actually a way that a lot of other tech companies have done before. So, we'll cover all of that in today's video. Now, right now, I'm filming this before the market opens. And as we can see before the market opens, we have again the bottom left corner here. That is extremely well extremely red for one day, right? Usually when we see minus 5%, -4%, that's usually for a week. But okay, we're already used to these types of moves on a daily basis. So again, the high-flying semiconductor names are down quite a lot pre-market. Even if we look at the neocloud players, we can clearly see that here as well. Nebulus is down 4.7%, core with 2.7, IN 2.5. Again, all of the Neocloud players are down. If we look at the semiconductor names, all of them are down as well. I mean, TSLC is down 4.2%. it it's ridiculous because you'll see you'll see the report. As for the fintech names, most of them are dumb as well. Big tech is a bit mixed as of right now. Looking at cyber security names, those are mostly green. And so let's maybe start off here with TSM. TSM over the past year has done quite well. It's now experiencing a draw of around 15% or so. But again, we've seen the draw down across the board in the semiconductor space. year to date it's still up 26%. Now TSM we get monthly numbers from the company so we knew that this quarter was a very good one. It was all about well capex are they increasing capex yes or no because remember this is a company that a lot of other companies are relying on and so if they are increasing capex then there is clearly a lot of demand because remember this is a company that is quite conservative. This is a company that does not want to say, "Oh, we are, yeah, capex is going to be hundreds of billions of dollars and then they overbuild and then they're screwed." No, this is a company that is extremely conservative time and time again. And guess what? They have increased capex guidance yet. Again, we'll look at all the numbers in a bit. If you enjoy all these type of videos, you know what to do. Hit all the buttons. We really appreciate that. Subscribe if you're new here. This earning season is going to be one hell of a ride. We also go live a lot with Amit, with Jose, with Tanner, etc. So, make sure you're subscribed. We really appreciate that. Want to support me even further, do check out the link down in the description and the pin comment with the top 10 best stocks to buy now or go to full.com/gotchinvestor. Thank you very much. Now, let's start off with the numbers for TSM. Revenue up 33.7% year-over-year. Net profit up 77.4% year-over-year. And quite a big beat which came in at $22.3 billion versus $19.74 billion which was expected. Gross margin 67.7% estimated number 67.1%. As for EPS per ADR that is $4.31 huge beat expected number was $3.83 operating margin 60.3% also a beat there net profit margin 55.6%. As for some extra data points, wafer shipments are up 16.6% year-over-year, up 3.9% quarter over quarter. Return on equity 45.9%. Operating expenses held at 7.8% of revenue. Inventory days sits at 87, up 7 days quarter over quarter. North America does represent 78% of all revenue. Then if you look at the revenue per platform, of course, HPC 66%, that's up 20% quarter over-arter. Smartphone is down actually 4% quarter over quarter, but that's not surprising. Then we have IoT up 4%, automotive up 15, DCE up five, and others up five as well. Of course, the big story here is high performance computing. As for the technology mix, we had 2 nanometers, sits now at 3%, 3 nanometers at 30%, 5 nanometers at 33, and then we have seven here at 11. So 77% of share wafer revenue comes from 7 nanometers and more advanced nodes combined. As for guidance, revenue guided for Q3, 44.6 to 45.8 billion above the estimates. And here this is implying 37% year-over-year growth. Gross margin guided to between 65 to 67%. Also a beat an operating margin between 56 to 57% versus 57.7% which was the estimated number. As for peacefill year capex outlook they raised it. That's now between 60 to 64 billion up from 52 to 56 billion. management guided capex over the next 3 years will be even more significantly higher than over the prior 3 years. So whoever thinks that we are at the top near the end of this cycle clearly one of the main players here is telling us no uh that's not going to happen anytime soon. They also announced an additional hundred billion investment in Arizona lifting its total US investment plan to $265 billion. The eventual US footprint could reach 10 FABs and two advanced packaging facilities. The four new facilities are expected to focus on two nanometer logic, though the final mix could shift to three logic fabs and one packaging plant. Construction timing remains undisclosed and will depend on market conditions. And so here again, what do we have? An excellent quarter from TSMC. if they are seeing such a huge increase. I mean, don't be surprised if we're going to see another great quarter across the board, whether it's Micron, AMD, Nvidia, even Intel, all the other players out there. Clearly, the momentum is still there. We're going to have the big tech players report over the next two two weeks and a half or so. I'm expecting again big capex numbers because well when you hear what ASML and TSMC have told us over the last two days I don't see how that does not happen. So to me here as well the AI story continues but sentiment is important because we started getting some headlines. Cororeweave explores hedging against a memory price drop. So, Corife is exploring put options and other derivatives to protect against a major future decline in memory and storage prices. Reuters report Kore has not executed hedges. Discussions remain early stage. Cor supply agreements with Sandisk and Micron include price floors and ceilings creating a reason to manage downside exposure. So, what the heck is going on here? Well, this report says that you know what, Corwave is thinking about the next couple of years. And if in the next couple of years spot prices are going to come down, well, okay, great for future contracts, but since Corif already has supply agreements with Sandis and Micron, they are going to pay a certain price no matter what. And so to protect themselves from paying over market spot prices, well, they're exploring put options and other derivatives because they think that okay, if those headlines come out, if this is true, right, memory prices are going to come down, then probably the price of SanDisk and Micron, the stock is going to come down as well. So this is how they protect themselves. Now, according to Evercore ISI, they're telling us that memory supply stays tight through 2027. Well, guess what? That's exactly what Micron has told us as well. So, DM and NAN constraints are expected to worsen exiting 2026 and persist through most of 2027. AI infrastructure demand continues absorbing capacity, limiting near-term pricing relief across DRM, NAD, and HBM. Evercore does not view Core's hedging exploration as a near-term change to its AI infrastructure thesis. Again, it's about headlines and sentiment right now. And how does this affect, for example, Dell or other players? Well, server impact for Dell. Lower memory prices could pressure Dell's year-over-year pricing in AI and traditional servers where higher component costs were passed to customers. Lower component costs could also unlock delayed deployments by improving server affordability, availability, and customer budgets. Memory and flash prices has spiked more recently. We've seen that across the board, and we've heard that from every other player in the industry. So, of course, right now when we get these headlines, woo, the industry is cyclical and high prices typically ease as new capacity ramps. But new capacity doesn't ramp anytime soon. And this whole hedging thing apparently is something that's very popular with energy firms and airlines, which brings me to the next topic here, and this was also a headline. So, Penguin Solutions, apparently a very good company that I'm starting to look into more and more. a cheaper alternative to more GPU. So, Penguin Solutions plans to launch its memory AI KV cache servers in Japan in Q4. It gives local AI players a lower cost way to scale without adding more Nvidia GPU servers, storing LLM short-term memory outside the GPU targets, existing GPU memory bottlenecks, improving utilization rather than replacing Nvidia servers outright. Right? This is a solution that you put on top of the Nvidia stack. We also had Rebellion's NPUs tested as lower power inference auctions. So, Japan's Toman devices and Tomorrowonet are testing Rebellion's NPU based servers for AI inference. We knew that inference is going to change the game. NPUs are optimized for inference, not training, and are seen as more power efficient and price competitive than GPUs for serving production workloads. Tommen's president has said NPUs will become a strong option for building AI infrastructure signaling disruptor support for post Nvidia chips. Now again the Nvidia chip story don't forget it's a hardware plus a software combination right you can run these chips extremely efficiently if you use of course the Nvidia software stack if you try to do a mix of everything else might not be as efficient cost efficient or efficient when you look at inference and so again we get headlines here Japanese firms are looking beyond simple chip swaps towards memory optimization and specialized inference semiconductors Nikke frames this as part of a global post Nvidia movement including Google's TPUs and custom AI chips which could reshape GPU supply demand and pricing over time. Now let's focus a little bit here on Penguin Solutions and why they've become quite popular in this space. So they are seen as an AI factory platform not just a hardware vendor. So they have here a couple of solutions clusterware AI, memory AI, compute AI, origin AI and then services. So they are saying something that everybody in the industry has been saying aentic AI is shifting the bottleneck from compute to memory. So what they're telling us here is this 2x higher inference performance from the memory AI KB cache server eight times lower time to first token latency and four to five times more cost effective than GPU HBM they recently reported their quartile figures revenue was up 48% year-over-year non-gap diluted earnings per share was up 79% year-over-year both of them here huge beats three segments in the business advanced computing up 4% year-over-year. That's 137.6 million. The most important one here is integrated memory up 111% year-over-year. Now 57% of the total company's revenue. Then we have optimized lead up 7%. Legacy business doesn't really matter, but that was 66.1 million. Fiscal year guidance was raised for the second time this year. Net sales growth outlook raised to 22% up from 12%. Non-GAAP gross margin output raised to 28.5% help by favorable memory pricing. Integrated memory segment now guided to grow 90 to 95% for the full year and preliminary fiscal year 2027 viewed from approximately 30% growth in both sales and non-GAAP EPS from the fiscal year 26 midpoint. Full guidance will be given in the next earning score. Now, as of right now, you might say, "Okay, this sounds very very interesting and and all of this actually is quite bearish for Nvidia." Well, point number two here, Penguin Solution is named an Nvidia AI factory specialized partner and Dell Technologies Global Alliance, America's AI partner of the year. They are working together, right? This whole thing of right now memory prices are going up. So, everybody has to pay up, pay the price. memory players margins probably might have peaked or will be peing over the next couple of quarters. Well, what have we been saying? We've been saying that all of the players in the industry do not like to pay these types of prices. So, they're going to look at solutions, right? You think Nvidia is happy with this? No. And so, they're looking at solutions, which in this case is Penguin. Yes, quite a strange name. And so, if you look at high memory prices, all you can do is smile and wave, boys. Smile and wave. Now, how does this impact the rest? We we talked about the Nvidia, we talked about the AMD, of course, they are going to do whatever they can to make sure that you optimize for memory, but what about the rest? Well, what about Micron? What about Kore? What about Nebus? Well, for a Micron, as we've said, I think revenue for Micron and these types of players will still continue to go up because there is just an insane amount of demand and capacity is not going to come online for another 12 months or so or more capacity. But margin wise maybe pricing power is starting to peak. As for the rest a kobe or a nebuse remember cheaper solution means more influence. More inference means more business for these types of businesses. So that's a good thing here. But remember we had deepse drama of a year and a half ago. Everybody was freaking out. It wasn't bearish for Nebuse. Was actually according to them their best week they ever had. And so here as well, I think as memory prices are going to come down or these companies can do more with the memory that they already have, it's bullish for the space. And as for Nvidia, we've seen some headlines come out more recently, right, with some analysis. Oh, huge delays happening. Well, apparently directly from Nvidia, they're saying there are no delays happening. Every year actually, we have these types of headlines come out, whether it's Nvidia or anybody else, mostly Nvidia because that is the bigger name. So, and and it usually also happens right before an earning season starts, which the timing is is quite interesting. But remember, Nvidia always talks about the same thing is the total cost of ownership, right? It's about how much does it cost you to generate X amount of tokens. And if we're getting a Penguin solution that sits on top of an Nvidia stack and that generates more tokens or less money, it's bullish for Nvidia. It's not too bearish for Nvidia despite the headline saying, "Oh, looking at the post Nvidia era, right? That's not really the way we should be looking at these things because let's be honest here, I don't think Nvidia is getting replaced." Moving on to the last part here and that's talking about Nebus. So, Nebus launches asset light infrastructure partnership model which lets infrastructure partners deploy Nebus full stack AI cloud platform in their own AI data centers. It pairs Nebus systems architecture, software stack and customers with partner capacity. It gives data center developers, infrastructure investors, regional operators and national AI projects worldwide a route to tap the fastest growing AI cloud market. It generates a high margin revenue stream for Nebus with minimal incremental capital requirements and comes at a time when AI compute demand continues to outstrip supply. Now people are saying here oh this is a completely new business. the business model has changed except no this is just Nebus trying to accelerate their own uh business right bottleneck number one is of course building the data center building a data center takes time takes a lot of capital what Nabius has and what's been the the bull pieces from the start is of course the software stack and so if they are going to partners and these partners have data centers already built or they're building data centers but they lack the software side of things when Nebus comes in puts that in the partner is now ready to go, ready to start generating money because suddenly they have the software solution that Nebus provides which of course is already extremely successful otherwise we wouldn't have seen this happen or we would not have seen Nebus be so successful and so it makes absolute sense here this is not Nebu saying oh what we're doing right now isn't great no it's Nebu saying what we're doing right now is very good we'd like to accelerate things the way to accelerate things is this type of model for now. It's just them diversifying their way to grow and increase revenue substantially. They also said that they've already entered into initial arrangement under this asset light model. I don't know if we're going to get a lot of information in the upcoming earnings report and call. I would assume maybe a little bit. There are going to be some questions about that, that's for sure. But to me, this makes absolute sense. It doesn't mean that the whole thesis for Nebus has changed. The whole business model has changed. On the contrary, it has just accelerated. I think this is another excellent move from Ardi and the team. He said, "Our new assetike model gives infrastructure partners a flexible way to benefit from the explosive growth of AI, right? Building data centers, getting the GPUs, that's one thing, capital intensive, time consuming. Well, once you have that, how do you monetize it? Other than yeah, renting out, you would like to have a solution like Nebus. You would like to have a partner like Nebus because Nebus brings the demand. Partners can begin generating a return as soon as capacity goes live. Partners get fully owned AI infrastructure assets built to Nebu standards plus a fast route to serve the AI cloud market. Again, if this was a company that its core business was already struggling, then I would say, okay, this is a complete new pivot. Doesn't make any sense. But this is a company Nebus that has been exceeding expectations, that has been executing like crazy. And to me, this is just them scaling much much faster. And so, yes, going back to what one of the executives has said more recently, could we see over the next 12 13 months tens of billions of dollars in revenue for Nebus? I guess it would seem that this is already an early success. And so, while right now we're seeing quite a lot of red here, pre-market for these types of names, do I think it's completely over? No. I I just think it's maybe a little shift in sentiment, headline driven moves here. we are going to get into the early season where you're going to see all of these good players generate more and more revenue become much much better companies. I think there is yes a buy the dip opportunity and me personally I'm looking of course at Nebuse around $190 per share or so I look at Micron right now though with Micron do I need more Micron probably not do I need more Nebuse probably yes even though it's already a huge position for me but yeah I I just think and with AMD I already told you if it does go way below $500 per share if it happens great if it doesn't it doesn't but looking the market cap, looking at the huge potential of this company, this is one I am willing to buy the dip on. So, all in all, that's all for me in today's video. I hope you enjoyed. See you all in the next one. Bye-bye.