Transcription
So Nvidia grew revenue 55%. You've got the hyperscalers spending 350 billion, 400 billion. Who can even keep track after a while? Where are we in terms of this story in your view? What ending?
Now I've been doing this job for almost 18 years. I've never seen anything quite like we're seeing right now. AMD was thought of as the alternative. They said they were going to have a GPU chip. What's happened? What kind of inroads have they made? And I will say I've made my career being negative on on Intel. What the hell happened to Intel? Can you, can you just explain to me what the hell happened to this company? At this point, I don't see anything out there that's so wonderful coming out of AI that means the returns are going to be so high. I think the doomsday scenario would be there's they're spending all this money and there's no return, right? And if that's the case, then I think the whole thing would come crumbling down, frankly, for everything else. And that's what we're here to talk about.
[Music]
Hi, this is Steve Eisman and welcome to another episode of the Real Eisman Playbook. And today I'm interviewing Stacy Rasg, who is the chip analyst and semiconductor equipment analyst at Bernstein. Welcome, Stacy.
Good to be here. So Stacy, not much going on in your group. Never boring. Nobody really cares, but let's see if we can mine a couple of nuggets out of what you do. So I do this with a lot of guests. Let's imagine that we're at a cocktail party. I'm someone who watches CNBC, kind of knows what's going on, but I'm no expert. And we get introduced and I find out what you do, and I think, "Oh, this is a great opportunity to learn." So Stacy, give me a five-minute dissertation summary about the boring aspects of what of the stocks that you cover.
You bet. You bet. So again, I am Stacy Rasg. I'm a managing director and senior analyst at Bernstein Research, where I do cover the US semiconductor and semiconductor capital equipment space. Now, I've been doing this job for almost 18 years. I started this in April of 2008, about three weeks after Bear Stearns failed.
Good timing for you.
Time that time that you remember. Um, I've never seen anything quite like we're seeing right right now. Um, and it really is a tale of of of two cities in some sense. We've got the AI trade and the AI stocks, the AI numbers that are just ripping, right? And we've got everything else that, in to be honest, is is kind of lackluster. And it would be very interesting to see what would be happening to this space as well as maybe the the the broader situation if we were not in the middle of of what is a very clear AI boom, because it really is AI that is driving everything right now.
That's the big push.
Yeah, it really is. And just just to give you and you know, just to step back a few years, remember it's only been a couple of years since this really started. Clearly started with Nvidia, right? Um, Chat GPT was sort of the, you know, the the lightning rod for all this. ChatGPT was released in November of 2022. It wasn't that long ago. Um, and Nvidia really started their their run in around May of 2023. They had an earnings call. I remember.
Do you remember that one?
Actually, absolutely. But this is this was this was reaction the analyst community. Yeah. I I mean, the the title of my my earnings recap note was was the big bang, and and and it really was the first they report first quarter 2023. I I can't remember which. Yeah, they have a weird fiscal year. They always report late. Yeah, they have a weird fiscal year. But whatever quarter it was, I just remember I the the the consensus for the following quarter was something like 7 billion in revenue, and they got 11. And today, by the way, 11 would seem quaint, right? But but at the time, I had to look at the release like twice to make sure I wasn't looking at the wrong number. Like it was it was it was a different company. It it was it was it was absolutely insane. And and and I mean I mean they're they're doing, you know, 50 billion a quarter now, like not not 11.
So even even but by the way, before you continue, the one statistic that I always like to say is if you look at just look if you knew nothing and you looked at Nvidia's most recent quarter and you just looked at the numbers and you said to yourself the following like three sentences: This is the largest market cap company in the United States. It's a market cap of over 4 trillion, and they grew revenue 55%. And you and and then you just stop to think about that just for a second, like like wait a minute, the largest company in the United States, just this not a billion dollar market cap company. This is a four trillion market cap company, and it grew revenue 55%. You have to like take a step back and say, and you can't even grasp it, which is what has to happen for that to happen. And that's what we're going to talk about. It's really amazing. Again, I've never seen anything quite like this where and that's the thing. It hasn't it hasn't slowed, right? It's these were numbers that when it started seemed unbelievable, right? And and it's just it's just been going from there. But it would have been even bigger. You have to remember like they got cut off from China, right? It would have been more than even more like without without so. And and we're seeing other other names catching similar maybe not to the same magnitude, but similar types. Broadcom, for example. Uh, we can talk about the whole GPU versus ASIC, but but they're they're starting to see a similar similar type of acceleration, and there's a lot of other names are on the periphery, um, you know, cooling and power and memory and even in the semi space that are doing well. So that's all doing very, very well.
On the other end, you you can take the more traditional semi to the analog space. For example, you have your Texas Instruments and your NXPs and your Microchips and your your your ADIs, your Analog Devices. Um, that actually were super super strong during COVID. Like we haven't even talked about that. There was like a massive, everybody bought their laptop that needed all those chips. So there was a massive overbuild during COVID that some parts of the market are still to this day trying to work off, and we're, you know, we're several years past the the peak of COVID. The analog names are sort of in in that category. They're they're well off the peak. Uh, there's some hope of of sort of a cyclical bottom and and recovering. We're still waiting to see that. That's not a growth story. It it has not been. We we can talk about I've been a little more lukewarm on my on my analoges. We can talk about that. But like that part of the semi-industry, you know, from at least from a revenue standpoint has been doing, I'd say less well, right? And then you've always got your pockets here and there. But but I mean, without the AI story, things would be in a very different place right now. But you can look at the overall semi. I mean, the the SOX, which is sort of the the broader semiconductor index, at least as of Friday, had been outperforming the S&P year to date by almost 1400 basis points. Um, this the group collectively has been doing very well, but it really is on the back of these AI AI names that's sort of lifting lifting the the so that's the rising tide.
Where are I to think of it this way? Okay, so Nvidia grew revenue 55%. There would have been more if there was China, but you've got the hyperscalers spending 350 billion now, Oracle 400 billion. Who can even keep track after a while? Where are we in terms of the story in your view? What ending?
Yeah, that that is the question. You know, the the because if someone was short, if I if believe me, I would never short Nvidia. You have to be out of your mind. But if you had a thesis that you were going to short Nvidia, you would say it's inning seven or eight, and pretty soon the 55% is going to become 10%. Yeah. You know, there's a couple different flavors of of that as well. U and and that is is the bare case is just these numbers have gotten so big so quickly, how could they possibly be sustainable, right? You know, you've got Jensen out there though talking about, you know, in 2030 we'll be doing three to four trillion dollars a year of infrastructure spend. These are just like like unbelievable numbers. Yeah.
So I think there's there's two ways that that sustainability question can go wrong. One one is the purely cyclical. These guys have spent a lot of money. They're going to keep spending money, but they'll they'll take a pause for a bit. There's a digestion. And even before AI, you could sort of look at the hyperscale capex numbers, and they would tend to build and digest and build and digest. It wasn't unusual. And if that and I always say like, I I've covered these stocks a long time. The chances of that at some point happening, it has to be 100%. Only a digestion period. Digestion. It's it's not now. And I've been saying this ever since it started. It's not this year. Doesn't look like it's next year. If if Jensen and Hawk 10 over at Broadcom are correct, like it it doesn't look like it's '27, but at some point, I guess so. I don't know when. Not now.
How would you Let me ask you a question. What would be the sign that would tell you it's happening?
Yeah. I mean, you're you're looking I mean, if the instant one of the hyperscalers cuts capex, like it's all over. You'd be too late by by then. But the point I want to make, if you thought it was purely cyclical, it it would be painful. But you could probably feel comfortable buying that dip because you say it's just cyclical and the outlook is still strong, and these things, you know, they they go up, but it's never a straight line, right? Fine. I I think the doomsday scenario would be there's they're spending all this money and there's no return, right? And and if that's the case, then I think the whole Yeah. the whole thing would come crumbling down for all of these games and frankly for everything else. I think by the way, you could make um an analogy to the 1999 2000s. Yeah, sure. By the way, so I was I was on the I was on the sell side back then, and sitting across the hall from me was a young Henry Blodget before he went to Merrill Lynch. And Henry was going out and he was making statements like dynastic levels of wealth are going to be created. The internet's going to conquer the world. And he was 1,000% correct. But but but so much money got spent so rapidly that the returns at first were not there. We had this enormous tech recession, and then eventually we came out of it, and Henry's predictions became true. So at this point, do you I I don't see anything out there that that's so wonderful coming out of AI that that means the returns are going to be so high. I mean, I mean, the search is better, but it ain't crazy better.
That also is an argument that I I actually do think we're still early if people worried about bubbles and we're not. And I and I get the fears, right? The numbers are very big. You know, people start to look at like companies like Nvidia that are now starting to invest more aggressively in the broader ecosystem as well. People start to worry about you're pulling a little bit of a GE vendor financing. We can talk about that. But but people look, I I get it. It raises eyebrows, like I I understand. At the same time, though, I I mean, it's it's not like like during during the, you know, the the tech bubble, like they were laying fiber, for example. They laid dark fiber, stayed dark for 20 years, it didn't get used. Like we're not nobody's buying GPUs and sticking them in a warehouse and and stockpiling. They're all getting used. The demand is is off the charts.
Just interrupt for one sec. The one difference though, important difference between then and now is then you had these rinky-dink companies that had just gone public that had no revenue, had a business plan that were spending money. These are real companies, and this is these are the biggest companies.
My next point I was going to make that's a big difference. It it is, and and people worried about bubbles, and we're not anywhere it might we're not anywhere near bubble territory yet. The valuations are actually fairly reason. They're elevated, but they're actually fairly reasonable. Nvidia's I mean, these guys I mean, if you think the numbers are anywhere close to correct, it's it's not even expensive. It's way. And by the way, Nvidia is way cheaper today than it is before the whole thing started. Like the stock's up a ton. The earnings are up. The earnings have gone faster than the Yes, they have. So, it's not like it's it's these things are not trading at 100 times earnings like they were during the bubble, right? Um, I always joke, OpenAI hasn't even gone public yet. Like, can we be in a bubble? Like if that hasn't happened, like so I don't worry too much. I I lean a little more toward the idea that we're likely more on the earlier side than the later side of of this. So maybe three or four, like whatever you want to I don't think we're in eight yet.
Okay. Okay. So, so let's just talk about the whole like I mean, Jensen talks and tells a story that would like insane. So just talk to me pretty good and he's been pretty good predictor and he's been right, but you you still sometimes can't even believe it. So just talk about how big the opportunity is and and then let's talk about from your perspective, what do you think this whole ecosystem will be capable eventually of doing? Because that's the real important question. What what are the use cases for this?
Yeah, absolutely. So in terms of the opportunity, so these guys are already spending hundreds of billions. He thinks that number goes to two trillion a year, a year. We we'll see. Well, you're seeing countries like Saudi Arabia getting sovereign. And and then again, I've got Altman out there, um, who is very aggressive. And we were talking about this before we started, you know, so you got, you know, you got Oracle with their 450 billion in RPO, of which 300 billion looks like it's OpenAI. Let me just explain to the viewers of viewers what that meant. So Oracle came out and said they they use this term, it's basically a fancy term for backlog, and they said that the backlog quote unquote type backlog had grown to 455 or something 350% in like three months or a year, whatever it was. And and And then they they sort of shut up. And then then people started to dig, people like you. And the word was of the 455 billion, 300 billion was just from OpenAI. And then the skeptics said, well, hold on there for a second, buddy. Brown or Chad, they don't have OpenAI doesn't have 300 billion to spend. They've raised they raised 60 billion so far, but they just got another hundred billion in Fusion from from Nvidia.
Sort of. What do you mean sort of? It it's not 100 billion right away. They'll they'll do it incrementally as they build it out. So, but they So, let's say they got a hundred billion in in the wallet, but they but they're losing money. So, they're spending that too. So, there's no way that OpenAI could spend a $300 billion yet. Yet, yet. Okay, got it. They've got other mechanisms though. Like I said, they're still private. I think what's the current valuation? I can't even remember. like 300, 500, but who cares? I mean, they got the point I was making though was that Altman is very aggressive, right? So he's got that with with OpenAI. He's got this opportunity now with Nvidia where they will be helping OpenAI to build out that infrastructure over time. I think they're Nvidia is going to deploy 10 billion in they're going to buy 10 billion of of equity basically OpenAI um next year. Okay. Um, start with that first trench, and that'll be a gigawatt of of power. And to your question on the opportunity, it's funny, the the limiting factor, it looks like it actually may be power. So how much we'll come to that. But in terms of size of the opportunity, you can sort of think of one one gigawatt. The numbers that have been tossed around, one one gigawatt of power to power a data center, it's roughly call it like 50 to 60 billion, probably of spend to build out that gigawatt for Nvidia. That is probably 30 to 40 billion of revenue opportunity for the stuff that they sell into that infrastructure, right? You know, I mean, look, Altman was out the other day saying he wanted to get to the point where they're building a gigawatt a week, a week, by the end of the decade. You can't build a gigawatt. I don't know what he can do. We We'll see. But I mean, you can start to get big numbers, right? You you can't build from nothing. Things have to change. Certainly things have to um have to change. So the opportunity is still huge. The opportunity is still big. The again, let's come back to the question of clearly these chips are revolutionary. They do things that couldn't be done before. Um, how do you see like use cases? Because like for example, just from my perspective, when I go on Google and I do a search, I now get two responses. I get the traditional Google search, and then I get the the the Gemini AI search. And the Gemini AI search is better, and it's more comp. It's kind of like the college student versus the high school student. It's certainly better, but does it make my life that much better? Not that much better. So, in terms of the use, so the use cases, you've got those kinds of things. There's, you know, I have my OpenAI subscription for 20 bucks a month or what whatever it is, or I I'm Googling and I get my sort. So, those those are one set of use cases. I wouldn't call those revolutionary. They're interesting and convenient. I wouldn't call them revolutionary. Um, I actually, by the way, I wonder if the real returns on this will not necessarily be, you know, we have the best model and we're renting it out for 20 bucks a month. I wonder if it's really productivity savings. So we're seeing massive improvements for example in productivity on coding in coding coding right. Um, we are seeing how much Let's nail that down a little bit. How much productivity savings are we seeing? I mean, some of the numbers I can't I can't remember the Google, some of the others are thrown like half their code now. Half the half their lines of code are getting written by AI. So I don't know how that translates to productivity, but we're seeing a lot more of that right? We're seeing companies that are actually starting to reduce headcount. So a lot of like the the the big SAS companies have been reducing headcount. Software companies. Software companies. We saw, I think it was IBM reduce headcount. This was last year in in at least in their agent was thousands of employees in their HR departments. You can imagine call centers. Like call up American Airlines or whatever to change your ticket, and you're now, you always were able to talk to a recording, which was never that satisfying, but that's actually getting better. Okay. You could imagine, you know, you're going to McDonald's or whatever, you're going through the drive-thru, and and there's experimentation going on there. Um, and but I don't know what that means in terms of like, you know, employment and and and and everything else at the end of the day, but I really do wonder if it's productivity savings that will help this. And then in terms of other broader use cases, you know, we're moving from these one-shot models to what are called reasoning models, where the model itself is uses a lot more compute, but it's a lot more productive. And even to what are called agents or agentic AI, where you could imagine, you know, I want to book a trip to France and look for get me plane ticket options that are leaving on these dates, and I want to do X, Y, and Z while I'm there, and you know, the AI model will just go out and and do all of that for you. And you can imagine it uses a tremendous amount of compute. We're not there yet, but that is clearly where things are going. So I don't really sign up for the idea necessarily that there were that that is that is the bare case, there's no use cases, this whole thing comes crumbling down. I I don't think so.
It's not that there are no use cases. It's just that the use cases aren't good enough to justify the enormous amount of money that's being spent. But again, we've only been doing this for like two years. That's fair. It's still very It's still early. Still very early. In in two more years, let's say by, you know, when when we're into this five years and we're still having the discussion, I start to worry. Okay. I don't think we're there yet. Too early yet. Yeah. I still Yeah. And I also think that the to your earlier point, the companies that are spending this money are not idiots, right? And they can see things that we cannot see. A lot of this a lot of this used to be was all very open back in the early days, like when ChatGPT, the the model structures and everything were very, everything's very closed now. Nobody's really, it's it's a competitive threat to put what you're doing out there. So these guys can actually see things that we can't see. Um, and I don't think they would be spending money willy-nilly just to spend money. They they have a purpose behind it. So okay, I don't understand. So when Broadcom comes out, Broadcom had an amazing quarter. They did. And this is not a company that I I'll be the first to admit that I know that well. And they start talking about um custom-made chips that they make for Google. Talk to me about the world of GPU versus custom-made chips. What's that all about? I really don't understand it.
You you so custom-made chip. You may hear the term ASIC. That stands for Application Specific Integrated Circuit. It basically means custom chip. But I'm going to use that term just okay. So it's not so there are a couple different mechanisms to get the compute that is needed to do AI. So one is one is GPUs, and it's a whole other discussion how that came came to pass. But one is GPUs, but the other is is custom silicon. You you don't have to use a GPU. You can design a custom piece of silicon to do this. And the idea would be, you know, you're no more you're not paying the Nvidia tax. The margins are pretty high. Um, and ideally, you can customize that chip to be more efficient for your specific workloads, for the workloads that it is designed for, versus like a GPU, which is general purpose and may do everything well, but may not do individual things perfectly. So this is why we we've been seeing this um and and a number all of the hyperscalers are working on their own custom silicon. I will say that the only hyperscaler that has really deployed this in any great volume is Google, and they have a product that they call a TPU. Tensor Processing Units. And they've been working on this for 13 years. They're on their seventh generation. Amazon has a version as well. They call it Trainium, and they're on their I can't remember third or fourth generation. It's been
So what do those chips do that the GPU doesn't do or do better?
So they can do anything that the GPU can can do. So I I would say you you tend to design a a custom chip for large stable internal workloads. Like it it costs a lot of money and time to design a chip. You don't want to do it for a workload volume that's very small. You want a a lot of compute, and you want that workload to be stable, because the issue with the ASIC is it's not flexible. Like there's no free. The GPU is flexible. The GPU is programmable and flexible. So there's there's no free lunch, right? So I'm willing to stipulate that an ASIC in theory should be more efficient for the workloads that you are designing it for. Otherwise, why are you bothering? But but again, there's no free lunch. There never is. It's not flexible. If your workload needs change, if your model structures change, you may need need to spin a new chip, whereas the GPU can can can be more flexible to handle that. So my own view is in if you do the math, by the way, in 2024, just on on the on just looking at the value of the silicon itself, the ASICs were probably low double digits, 10, 11, 12% of the total of total of what of total um spend on AI silicon processing silicon. So not that you have to remember, Nvidia doesn't just sell, they sell racks and and all kinds of stuff. If you just looked at the silicon spend in 2024, the ASICs were probably low double digits. I bet this year they're probably mid-teens. Okay. If I was to look forward, could they be 20 or 25% of a much bigger pie?
What's causing people like Google, some other companies to buy more of these ASICs as opposed as opposed just calling up Nvidia and say, "Ship me some chips."
Well, they they still buy a lot of chips from Nvidia, too, for sure. Right. Um, so I would I would say for internal workloads that are where there still is a lot of dynamic uh motion and and they're not nailed down, they're still buying GPUs. I'd also say for the companies that have large public cloud businesses, like like Amazon or Google, where they're renting out data center capacity to end users, those customers do not want to use custom silicon. They want to use GPUs, and that's because of Nvidia's software ecosystem. It's it's called CUDA, that everybody uses. And so, for example, the the custom silicon is available. You can go to Google Cloud today and rent a a TPU instance. They're available. No, nobody bothers. Nobody does it. They want So, they still buy a lot of GPUs for the public cloud as as well. Um, I think the whole question, it's funny, around the stocks, the narrative seems it tends to swing back and forth between GPUs and ASICs. And after Broadcom's report, where they took up their forward outlook for their custom silicon pretty significantly, I'd say the needle swung a little more back toward ASICs. I'd say after Nvidia's announcement with OpenAI, it swung a little more toward GPU. Broadcom's actually like incremental announcement was around OpenAI ASICs, and people were like, "Oh my god, they're never going to buy GPUs again." With Nvidia's deal with OpenAI, clearly they're going to be buying a lot of GPUs. I think all of this personally kind of misses the point though. The right question to me is not nec is not really which one's winning or losing. Right now, the question is, is the opportunity in front of us still big or is it not? Because if it's still big, if we're still early, then they both thrive. If it's not so big, they are both screwed. Okay. Right. It's working.
So your attitude is we're still early. The pie is in. So both will do well.
Yeah. And you have to, but if you don't believe that, you shouldn't be invested in in the AI space. I think that's that's Listen, if if I didn't believe that, I'm I'm going to leave here. I'm going to go change my portfolio. So you have to believe that we're we're still we're not at the point. And you you can look at some of these. So So with Broadcom, Hock Tan's the CEO. He's, you know, 72 or 73 years old, right? Um, everybody always worries he's going to leave. He just signed up to stick around until 2030. Really? And he's got some new targets for that retention that are pretty significant. So for their AI revenues, which by the way include both says custom silicon as well as networking.
I'm sorry, say that again. Sorry.
For Broadcom's AI revenues that in it includes custom silicon as well as networking. They'll do $20 billion this year. They're implicitly guiding for 40 billion plus next year in '26. And for for doubling. Yeah. Yeah. 100% over 100%. And sometime between 2028 and 2030, he's his targets, his baseline target is 90 billion, and his stretch target is 120 billion. Wow. He doesn't have to stick around. Like he's he's he's in his 70s. He's already a billionaire. He doesn't have to stick around. He's sticking around. He sees a reason to stick around, right?
So, okay, let's move on to some tangents. Sure. Let's go to AMD. Yeah. So, by the way, for those people who don't know, the CEO of AMD and the C and and Jensen of Nvidia, I think they're like second cousins. They're like distant cousins. Third cousins. Good gene pool. Yes. So I remember like two years ago when I was looking at AMD, when when this whole story started, the the bull case was that at some point companies are going to say to themselves, I can't be 100% beholden to Nvidia. I I want to I want to buy GPUs, but I can't buy all my GPUs from one company because if Jensen sneezes, I got to run over with a tissue. Yeah. So AMD was was thought of as the alternative. They had they said they were going to have an a a GPU chip. What's happened? What kind of inroads have they made? What do you think is realistic?
Yeah, you bet. So they've made some, and and I will preface this by saying I always tend to be a little lukewarm on AMD, which means I tend to miss the hope phase. We've been in the hope phase like like like this year, but but it but it's fine. Right. The stock's up a lot again on hope because they have not really blown numbers away on the GPU side. This happened last year as well. So last year, this they were just getting started, and they they had said, you know, we're going to do $3 billion in AI revenue, and folks in my seat, not me, but folks in my seat said, "Oh, no, no, we just did our Taiwan checks. They're going to do 12 billion." 12. And then they wound up for last year doing five. Last year. Last year. This is all last year. 2024. And five is a lot higher than three, but it's by the stock ran to 180 on on this. I remember. And five is a lot higher than three, but it's a lot lower than 12. The stock got cut in half, but it's a lot higher than zero. But is higher than zero. And I don't want to knock AMD too hard. So they did five billion last year from nothing. It was zero the year before. So objectively, that's pretty damn impressive. Yes. That being said, in this I mean, Nvidia did a hundred billion dollars, give or take. It's it's just not that big. Right now, that is also the bull case on AMD, though. The bull case is pretty simple. It is hey, Lisa, the CEO, Lisa Lisa said the TAM is 500 billion or whatever it is, and if they just get 5%, that's $25 billion. That speech is a speech that every like internet company would the TAM is a trillion. If we get 2%, we're worth a fortune. Now, to AMD's credit, they don't have to go out on two do $200 billion in revenue. 25 billion would probably be enough, right? That would be okay. And right now, like again, people are are back in that hope phase a little bit. The stock has actually been decent year to take a bit of a breather in recent weeks. But the real, so most of the tech stocks as well. Yeah, that's fine. But I look, the the real hope for them, it's they've got a new, so they they've had this this product that's been competing with Nvidia. It's called the MI300, which has been that was at 5 billion last year. It's kind of and so how is that chip compared to Nvidia's chip, just from a pure engineering?
Well, it's it's well-engineered, but it's just behind and behind years. It's it's Nvidia. Yeah, just in terms of raw performance. Sure. Couple years. How give me some numbers like how much?
Two years. Two years. But but they're trying to close the gap again. You talk about the both. So the other issue though that Nvidia that AMD has is software ecosystem. So even if the chip was perfect, like in terms of like raw performance, um, it probably still is preferable to use the Nvidia systems.
So let's expand upon that. When I'm a CTO of a company and I I need I need a whole GPU ecosystem. So tell me what I get. Obviously I get chips. When I buy from Nvidia, but what else do I get from Nvidia with as a package?
You get CUDA, which is the name of their I mean, in terms it's CUDA is a few things. So CUDA is their sort of GPU programming ecosystem, okay? And it is a few things. It it is the sort of like the programming environment to program the GPUs. That that's one thing. Fine. Um, it is also, in some sense, a set of optimizations that happen at a lower level that they've worked out over by almost by trial and error over many, many years to make these things run very smoothly and efficiently and and have the time to market, have the ramp-up time be very, very fast. And then thirdly, there's a lot of application-specific stuff that they've built on top of this, depending on what you're doing. So, for example, if you're doing robotics and you're using AI for, they have a whole package, it's it's called Isaac. It's, you know, it's pre-trained models and libraries and and all kinds of things that go about that. If you're doing, you know, medical imaging and diagnostics, they have a whole thing. It's called Clara. If you're doing quantum,
What does that do?
It again, it it's a AI pre-trained models, libraries, like all kinds of things to be deployed specifically against those applications. Clara would be for medical.
So you're writing applications basically on top of their software?
Yes, absolutely. Yeah. And they've got things that could that can very much help you depending on what and they've got a gazillion of these application-specific. You know, and what does AMD have? Not not much. They're working on it. So they their software platform is called ROCm, and they've been working on it for six or seven years. And and again, they're they're they're making progress. It's better, but but It's also a moving target because Nvidia is not saying keeps going. Yeah. In fact, you go to any Nvidia event, you go to GTC, or you go even when they were talking at CES or whatever, and and Jensen will get up for two hours and give this big keynote, and it's as much about the software as it is about the hardware. So, they've got everything there. So, even if if AMD's parts were dead on competitive, which they are not, they don't have anywhere to the ecosystem. And they're trying. I don't want to knock AMD, they're trying, but but it's hard, right? And again, it's it's a moving target. And just think of it, just to put it very simply, if you're you're a CTO, and you want to set this stuff up, you're buying GPUs, you can buy Nvidia GPUs and set up, you're probably up and running in days. So why would you buy any So why would you buy anything from AMD?
Well, people have been buying less from AMD, right? I mean, again, there is that aspect of we don't want all our eggs in one basket. Although even there, I've got other alternatives like what? Broadcom, you know, and I get the need for a second source, but is it is it AMD or is it is it Broadcom, right? I mean, that that's a real question, especially for the hyperscalers who are all working on their own silicon. You see some small deployments. So, for example, like a Meta uh deployed the MI300 on their Llama 405b AMD's chip. That was AMD's uh chip from last year, right? On their Llama 405b model. So that was exclusively MI, but nobody used hardly anybody used the MI, the Llama 405b model. Hardly anybody used it. Um, OpenAI, Sam Altman supposedly is working with AMD on their MI450, which is the one that comes next year, right? And AMD had Altman up on stage with them. That being said, it looks like Altman's working with everybody and and with Nvidia and Broadcom in a in a much bigger capacity, probably. So AMD is making progress, and and that's the dream, though. I I mean, it's that they don't need to do 100 billion. If they did 20, sounds like for them, really to go to the next level, they have to really improve their ecosystem. So I would feel again, I told you I tend to be a little lukewarm. I don't hate it, but I tend to be lukewarm. I would feel better about AMD if I really felt like lots of big customers wanted to use their parts for their own sake because they thought they were better. And I'm not convinced that that's opposed to just an alternative. Yes. And I'm not convinced that that's true. Okay. But but I don't want to knock AMD like like like I said, from what they've done objectively, it's pretty impressive. And you take a look at like an Intel. They were trying to build up an AI product as well. They had something called Gaudi, and I mean, they were supposed to do $500 million of of sales last year, which is tiny, and they couldn't even hit that. They had to walk that target back.
Let's go back to Nvidia for a second. Um, how important is China for Nvidia?
It is important, and and but they're not selling anything in China right now. Well, there two reasons why they're not selling. The reason why they're not selling now is because the Chinese won't buy, right? Well, they're they're being ordered not to buy. The government not to buy. The companies want to buy. Yes. Is that by the way, this is very similar to what happened to Apple around 2012, 2013, where all of a sudden press came out about Apple where the the press reports were saying Apple is not a good citizen. All of a sudden they had problems. I mean, look, there were some comments from the administration. How did Lutnik put it? He said, you know, we're selling like our fourth best GPU, it's not like the latest, but we want them addicted to our technology. And it's what was that line from that Marvel movie? He's he's out of line, but he's right. Right. But he is out of line. I I wouldn't have put it quite that way. And I think there was some stories that the Chinese were irritated by that. Tommy, probably true, right? Um, but it is important, and and by the way, Jensen has has not hidden from this, and and I like the fact that they've addressed this dead on. So there are two reasons that it's important. One one is just the purely financial. Sure, China is a big market. Yes, I get all that. That's actually not the most important. The most important reason for them to sell into China is it gets back to that ecosystem question. There are Chinese alternatives today that have better performance than what Nvidia is allowed to sell in on raw performance, because Nvidia is not allowed to sell the latest, latest, greatest. They have to sell constrained parts. Huawei, for example, has a part called Huawei's a local Chinese big local player, and they have a better chip than what Nvidia. Better better on raw performance. It's certainly not better on power efficiency, and but that's the Chinese have unlimited power. They can just build more coal plants. They don't care, right? Um, and it has better raw performance than what Nvidia's allowed to sell, but it doesn't have the ecosystem. Again, the same thing, they're hard to use. So the Chinese developers would rather use the Nvidia parts. There's a lot of demand for that. If you do not allow if Nvidia is not allowed to compete in China, what you're effectively doing is incur, in my opinion, you're encouraging those local developers to coalesce around a local player like a Huawei and make that ecosystem more robust, and eventually it it's it's robust enough where they can use it, because that's how you make these ecosystems. Or else you need a developer population that that that's working with them. And once I've got like a robust competitor in China, assuming they could get capacity, which is a whole other question, but if they could get capacity, why why would What do you mean capacity? Could they make enough chips? Right. That's a whole other thing that Huawei has an issue with, because we put other constraints on China's ability to manufacture semiconductors.
So what are the constraints on Huawei building enough chips?
They are the US semiconductor capital equipment players can't sell tools into China to make ASML. Well, ASML is Dutch, but I mean, so sort of peripherally, but AMAT and Lam and all these guys, okay, in the US cannot sell tools to make leading-edge chips. So they could build their own tools, couldn't they? There are some Chinese guys that that do tools, but not leading edge, and they're not as good, but they're trying. But these whole issues we're seeing around Chinese competition in AI, we're not just seeing it in AI, we're seeing it in semicap. We're seeing it analog, because we're forcing this is a whole other conversation. We're forcing the Chinese to be creative, right? You know, and it gets back to my my point on on on AI in China. If if we force the Chinese to coalesce around say Huawei, and they get because they're not idiots, right? They're very smart and and good engineers, and they build up that ecosystem. Now all of a sudden I've got a robust competitor in China, and then why does it stay in China? Maybe it gets exported, right? And now I've got a robust global competitor. So that's why it's important for Nvidia to be able to sell. And by the way, I would make the same argument for the semicaps and for the analogs and and everybody else who has have constraints selling stuff into China. We are forcing China to be creative in all senses of that word, and forcing them down paths. I mean, the the the way that Huawei is doing their chips, they're making quote purportedly leading-edge chips at SMIC, which is a Chinese um semiconductor manufacturer, contract manufacturer, and the way they're doing it are ways that you would never want to do. They're they're doing techniques that would not be cost-effective if they had to compete with that stuff globally, but they have no choice. So they'll do it, right? And eventually, do they get better? Eventually, they'll get better. Yeah. And so I I wonder if like in 10 years, we will collectively discover that we've created a monster, right? I am. And now it's too late, by the way. Even if we pulled all the export controls away, the Chinese can't stop, right? They've got to double and triple down on everything on everything that that they're doing. So, I think it's important for Nvidia to be able to compete in China. We'll see if they're able to.
Let's broaden out a little bit to your other stocks, and we'll talk about Intel last. The good, the bad, and the ugly, as I say. Just talk a little bit about I mean, obviously there are tentacles to the whole AI story. If I mean, everybody's an AI story now, right? Even if even if you're not an AI story, right? So just for example, you need memory chips. If you're going to if you're going to build GPU data warehouses, you need you don't just need GPUs, you need other kinds of chips. So which companies have done well because of this? And which companies have not done well and why?
Yeah, you you bet. So some of the peripheral plays in semi, you have memory, you've got semicap, right? Um, you've got some of the the the guys that do like the power management like in the servers. Um, if you stretch a little farther out, there's guys that do like the cooling and the cabling and vert and and the optics and that kind of stuff, like the optical connectors. So, there's the old JDS Uniphase, which I don't even know what it's called today. I don't know.
You know, I I don't cover the optical space. I let somebody else deal with with that. Um, but the most direct plays are, I mean, it's probably memory. I don't cover memory. It's a colleague of mine, but he's he's been a memory bull and this is part of the reason. So, we like Micron. Yeah, he he likes Micron. There's Micron and Samsung and Heinix. And so in general, he he likes the the memory stocks and they've done done pretty well.
So so these AI chips use a special kind of memory. It's called HBM or high bandwidth memory. It's a type of DRAM. There there's there's two broad types of of memory. What's called DRAM and NAND. NAND is like what's in your phone that stores your photos. Mhm. It's it's nonvolatile. So if I turn the power off, it still holds the stuff in memory. Right. Right. It's actually remarkable that they've pulled NAND chips out of fires and still been able to read the stuff off. It's pretty amazing.
Um, the other is is you have what's called DRAM. DRAM is like the system memory that's in your computer. It's much faster than NAND. It uses more power, but you have to rewrite to about every 100 milliseconds or so. When I pull the plug on the computer, I lose everything that's in the memory. It's what's called volatile. So AI chips use a type of DRAM that volatile memory. It's called HPM, high bandwidth memory. And if you were to look at at a at a GPU, an AI GPU, it's typically got the GPU die in the middle, the the logic die, and it's got these little chiplets on the side. Those those are the memory.
And all by the way, there's another thing that's benefiting here. It's called packaging. Um, you have to combine all these chips together. It uses a technique. It's called CoWoS, that um, it stands for chip on wafer on substrate. I'll describe it. Um, imagine I have my my GPU and a bunch of memory dies. All of those chips together are are connected together on a piece of silicon called an interposer. Typically, it's silicon. They're making it out of other stuff now, but typically a piece of silicon. That interposer has electrical connections that allow all these chips to talk to each other. And then that interposer itself has connections that that connect it to to a type of what's called a substrate that allow the that that allow the overall thing to connect to the other parts of the circuit. So chip on on silicon on substrate, chip on wafer on substrate, CoWoS. It's a type of advanced packaging that's done mostly at the at TSMC, who is what's called a wafer foundry. They are the company that manufactures the vast majority of these chips. They do the packaging as well. So the packaging space, and especially some of the equipment guys that sell them, have also benefited from from this. And in fact, when people start go to Taiwan and they do there's channel checks for like how much supply is quite often they look at the CoWoS that the packaging supply as a measure for how many of these chips are actually going to get get built.
But all of these guys have benefited to some degree. I I'd say the the biggest benefits have accrued to the guys who are doing the compute. It's been primarily Nvidia and and more recently Broadcom. You've had some other of the second tier guys that have done okay. You've had you like like for example, AMD and Marvell. AMD and Marvell. Yeah. Marvell also does like custom chips and they do optics and things. Um, some of the guys that do power stuff, some of the more analog type names, you know, you have Texas Instruments and Monolithic Power and Renaissance and Infinity and all these other guys as a as a smaller part of a much larger business. They may do a data center power and so they've sometimes there's a narrative there and maybe they they catch a bid.
You mentioned Verive and some of the other cooling names that you know they've had cycles depending on where they are, but like a lot of these things are are moving from what are air cooled to what's called liquid cooling and so that's benefiting some of these other names. Um, what happened to Texas Instruments recently? Yeah. So TI, what the stock got pummeled on their last earnings. It wasn't AI related. Uh, what what happened was, I should step back. Analog in general. These are guys that made that primarily do older generation chips, a lot of industrial, a lot of automotive, that kind of thing. Um, these all got really really inflated during COVID. There were massive semiconductor shortages and what happens when shortages, people tend to order more. Right. Right. So these guys are they ripped and in many cases, we are still working it off like years later. That's that's how inflated it got.
Um, this year people were getting excited in analog that maybe we finally hit bottom and we could start to get recovery and TI has been calling for for recovery. Um, as as have others. Um, as we get through the year though, it's looking increasingly likely though that to the extent that we're getting any kind of recovery, it's certainly not V-shaped. It's a lot rounder, like maybe it's U-shaped, right? And TI, like frankly, during conference season into last earnings, like like last May, June, got out and and they were very bullish and we we had them at our conference and and I got up, sat next to the CEO and he said, we're growing 13% in the first half and we're going to do even better than that in the second half. And like your gross margins next year, they're way, they're they're low. They have other things they they we should when we talk about tariffs and things, TI's investing a lot of money in capacity right now in the US, which is lower, which increases depreciation, is lowering their gross margins. But anybody said your gross margin next year, as it turns out, none of that was true. So why do you say it? They don't run a backlog model. They have a lot of inventory.
What do you mean they don't run a backlog model? Normally in semiconductors, you you customers will place orders for delivery at a later date and it gives you some visibility of demand going out. TI effectively has 100% availability of 100% of their products all the time. Well, that's their they have like 250 days of inventory. It's purposeful. They they they don't really run a back. What that really means is their visibility is zero. Okay. Right. And and I'm not saying like it's bad. Like it just is. And so I suspect what happened in the beginning of last quarter. I you know, I bet orders were really strong in the beginning of the quarter and he's feeling good. He feels good. He went up and he and he said it and then he got caught with his pants down. It happens, right? I think that's what I don't think it's any more than that. And it's funny too, because their actual earnings when they reported were not awful. It was a modest beat and raise versus sellside expectations. But he had gotten the buy-side expectations up. If he hadn't talked like that, um, I think the numbers would have been fine, but the stock would have been 190 going into the print instead of 220. And where is it now? It's like 180, right? So, I mean, that's that's that's what happens.
Um, and it was a little bit of a change. Um, the older the the prior CEO, Tavis, at TI, who I like, by the way, and I and I like TI as as a as a company. I really do. But he's a new CEO. He's been in the seat for a year. They they had the activist, the Elliott guys were in there for a while. Part of their the the activist was they wanted more communication. So that may be part of it as well, to be honest. I think TI was communicated less. They were doing just fine with less communication. The old TI model was basically not to say anything ever and it was working just fine. That's what happened. Okay.
So all right, let's talk tariffs for a second and then we'll finish up with Intel. What's going on? Give me the whole tariff situation. Well, and we we don't really know yet. We're still we're still waiting. Um, you know, there's something called the the the 232 investigation, which is a sectoral tariff investigation for semiconductors, which was opened up on April 1st, and I keep looking every day to see if they filed the damn report, and we haven't seen it yet. So, right now, we're kind of going off of, you know, Trump speeches and and and news leaks. Um, the most recent thing that we heard last week is it looks like the government's talking about they might want customers to have a one-to-one, basically for every foreign semiconductor that you purchase, you need to purchase a US semi-manufactured semiconductor, else those foreign semiconductors will get tariffed at a higher rate. And we don't know what that is. And there's also been some news that they may be thinking about tariffing not not just semiconductors, but semiconductors inside other devices. And I think this is important. What I So, and I'll hold this up. So, this is a this is a smartphone, right? This costs like $1,000, say, whatever it is. There's probably $500 worth of semiconductors inside this smartphone. Okay. Okay. And as it turns out, the US $500 worth of semiconductors out of a phone that cost $1,000 probably. It's probably 50 the 50% of the BOM is probably semis. Okay. Just just very rough. But the point I want to make is is everybody's worried about, oh, they're going to tariff semiconductors coming into the country. The US does not import that many raw semiconductors. I think in 2024 it was about 40 42 billion dollars or something like that. I thought they import stuff from TSMC all the time. Yeah. But no, but where do those go? Right. They don't go to the US. They go to to China and the phone gets built in China and then we import the phone, right? So most semiconductors enter the US inside other things, inside something else, in a phone, in a PC, in a car. We don't import that many raw semiconductors. And so if the government really wants tariff policy to be effective, they probably need to go down to like like we're importing a phone. You're going to tariff potentially the semiconductors inside the phone. And there were some news stories last week that suggested they may be considering this. This has been tossed around once in a once. The problem is the administrative overhead to monitor or something like that is is an absolute nightmare. I don't I don't know how the industry would deal with it, but it but it may be happening. So I I don't know yet. I think right now most investors are just assumed they're just assuming the Trump taco basically that like, you know, if it comes on, he'll give it and it sounded like he'll he'll give some leeway maybe. He seems to be giving leeway to Apple. Yeah. And again, you kind of have to, right? I mean, just just given you have to remember semiconductors are the most global of any supply chain, right? These chips cross, I don't know how many different borders before they finally end end up where they are. It's very difficult. So I don't know how they'll deal with it, but something will probably be coming down like in the relatively near future.
Um, at the same time though, you know, we are trying to encourage more local manufacturing of semiconductors and, you know, the the Biden administration was trying to do this with the CHIPS Act. Trump administration, you know, is trying to do this with tariffs and I mean, maybe it's effective. TSMC is building in Arizona, for example. They originally going to spend $65 billion there. They're now going to spend supposedly $165 billion there. So they're building more. And then I mean, it's a segue into Intel. Let's go to Intel. Back to our cocktail party. Okay. I say to you, Stacy, do me a favor. What the hell happened to Intel? Can you Can you just explain to me what the hell happened to this company? That's probably a topic for an entire podcast on. We're going to do the short version of it. And I will say I've made my career being negative on on Intel. It's been the key. Yeah, that was sort of like the big call. That was your big call. When did you make that call? 2012. October of 2012. What did you say in 2012? I mean, back then it was it started with pricing and then it started, I mean, it branched out into, you know, they were getting eaten alive by mobile and they missed that transition and then costs were going pretty much everything that I said was going to happen now is happening. Um, although I would say the magnitude is much worse than I ever thought it would be. Like it's really amazing what happened. It it's a few things. Um, so and it's both on the everybody focuses on the process technology, but there's stuff on the manufacturing side. There's also a lot of stuff on the product side. So on the manufacturing side. This is what everybody mostly looks at. Intel used to be at the bleeding edge of semiconductor manufacturing. In fact, Moore's Law, which presumably everybody hopefully has heard of this idea that every every two years semiconductor transistors on chips were getting smaller and delivering a lot of value. Gordon Moore was one of the founders of Intel and and you know, one of its CEOs and I mean, that's that's Moore's Law. That's what it was named after. So they were at the lead and and they lost that leadership. And and it didn't just start. It started 10 years ago. They started to slow down their their trajectory of of of process um technology migration to the point where their primary competition, who was TSMC, caught them and surpassed them. So the best transistors in the world today are no longer made by Intel. They're made by TSMC and TSMC is several years ahead. And so Intel has been scrambling like crazy to try to catch up.
And why couldn't they? Can't they catch up? What happened to this company? This is hard, right? So the the idea that with with every single process technology generation, we've had players that have fallen off the treadmill. It it's not new. I mean, you go back 20 years, there were a dozen different folks that that that could do leading edge. Now there's three. And in reality, it's really one. It's TSMC. And then you have Samsung and Intel. And I would say Samsung and Intel, maybe they haven't fallen off the treadmill, but they've got one foot like dragging. And we've never had an example of a company falling off the treadmill getting back on. M and it may just be that TSMC is the last man standing. Um, and you have to remember, these are the most complicated things that humanity has ever built. I'm amazed any of this stuff works at all. And I used to do this. Like my PhD is in is in semiconductor. I used to build semiconductor manufacturing equipment. That was in in my prior life 20 years ago. That's what I did. I have a PhD from MIT. Dungeons and Dragons days, way back when. I have a PhD from MIT doing that. I worked in IBM TJ Watson Research Center, their advanced lithography and project. This is what I did. And I'm amazed any of this stuff works at all.
Why are you amazed any of this stuff works at all? It it really shouldn't shouldn't. You think about this. I mean, the the size of the things that we're making, we're literally reaching atomic dimensions to the point where you like literally can't make the stuff smaller. Um, they're printing features that are orders of magnitude bigger than the wavelength of light that they are using to image them. I mean, it it's astonishing to me. Like, I'm sorry, smaller smaller than the wavelength of light that they're using. It it it it shouldn't shouldn't work. And it does. And we make a trillion chips a year. It's really amazing to me. So the idea that it's that they're having it's hard to do. And I don't know exactly why. You could argue they made bad decisions in hindsight. There were manufacturing techniques that maybe they could have used that they chose not to because they thought they were too expensive. And the other thing is is, you know, they've had personnel issues. They had a big layoff in 2016 that where they're widely believed to have laid off like some of the best and the brightest, the older the older folks. And so it's a lot of things, but they had big issues on process and and it's very difficult to to catch up again once once they go. Now, that being said, on the product side, they had a bunch of issues too. So they missed many new market transitions. So they clearly missed mobile, right? And and in fact, there's an article in the Atlantic from like, I can't remember 20 2012 maybe. Um, I'm quoted in it, but it was an article with Paul Otellini, an interview with him, who was one of Intel's prior CEOs, and he was talking about mobile and he basically said, we had the opportunity to bid on on the chip in the first iPhone and we turned it down because because we didn't think that we could make it cost our cost model said we couldn't make it cost effectively and we didn't think there's any volume. And as it turns out, our cost models were wrong. We could have made it profitably and the volumes were a hundred times like what we thought that they would would be. And and maybe they missed the iPhone, maybe the world would be a different place. And to the actually tried to scramble and get into mobile several years later to the extent where their but their parts were so bad, they literally were paying people to take them. It was called contra revenue. And in their worst, they used to split this out as a segment. In their worst quarter for mobile, they lost $1.1 billion on minus $6 million in revenue in that segment. The quarter after that, they combined it with their PC business so we couldn't see it anymore. So they missed that. And then more where is they've clearly missed AI. Right. Right. I mean, that that goes without saying. Like they've got no no AI sort of to speak of. But even on their traditional business, I mean, you look at AMD, who's their primary competitor, and we were talking about AMD in the context of AI, but AMD also makes PC chips and they make server chips. They're eating their lunch and they've completely eaten their lunch. And and here I think it was real arrogance. And and I'll give you an anecdote. I was in an Intel event and I I won't identify this person. I was speaking with a very senior Intel executive and we were talking about, you know, their their products and this is in the hallway, right? And and and I said, "Well, aren't you worried about AMD?" And this person looked down their nose at me and they said, "Oh, you mean what with that chip? They're going to launch in a year." And then they spun around on their foot and walked away from me. So, they were very very arrogant. And just to put this in context, you go back to 2014, 2015. The controversy on AMD back then was legitimately, are they going to go bankrupt or not? It was a sub $2 stock. Okay? And you know what their server take servers because that was their most profitable business. Their server market share was .1% on a revenue basis. Do you know what it is today? They just crossed 40% last quarter. Wow. So they destroyed it. Destroyed them. Yeah. Yeah. It's it's amazing.
So how do you let that happen? I don't know. And what what do you think of their their foundry thing that they're trying to do? The problem is they're having trouble making parts for themselves, let alone anybody else. Right. There's no customers. And now they're trying to tell us that eight, they said eight. So they they've had they've had this whole what what the prior CEO, by the I should even step back. The prior CEO was Pat Gelsinger. Foundry was his thing. He thought that was the the salvation of the company. But the problem was his view was, if we build it, they will come. And and can I can I curse on this on this podcast? He invested absolutely absolute tons of money that they didn't have to build out capacity that they had no volume to fill. Um, and and frankly, you look at, I think their their manufacturing, their foundry business last year lost $13 billion or something. I mean, they went, they're burning tons of cash. Um, and now and they had this what they called five nodes in four years. They're trying to catch up on the process tech. I wouldn't say that that's gone well. You know, I I could go through the whole thing, but I I mean, they'll say it's success up lately. Hang on. Let let me finish this because this it's it's really amazing. We're still at the cocktail party. We're still at the cocktail party. Um, so you ask like, why do why do they have any foundry business? So they have their current process. It's called 18A. This is the culmination of that five nodes in four years. Um, it's supposed to have its first internal product for for Intel themselves coming out end of this year. It's called Panther Lake. It's a client PC product. More volume next year. No foundry customers on 18A. They're trying to tell us now 18A was never meant to be a foundry node, which in my opinion is That's not what I'm not following you. Say this again. They're making um parts for themselves on this this process node. Don't worry about that. The the name of it is 18A. They call the process 18A. Okay, don't worry about what it means. It's it's the fifth node of their five nodes in four years. Okay. Okay. Five process uh technology transitions for themselves. They were supposed to do uh other customers on 18A as well as as a foundry. Foundry. So they were just making chips for themselves. Just for themselves. Okay. So they're supposed to make they're telling us now that 18A itself was never meant to be a foundry node. We're never meant to make products for other customers. That is not what they were telling us before. Um, I think what it is is true is that they they did not engage properly with customers on 18A. Now they're saying the real foundry node is going to be 14A, right, which is the next the next one. And and they're trying to engage with customers early and they'll have an 18A version for foundry customers at some point, probably the next couple years. It's called 18AP, right? But there's no there's no volume. There's no customers yet. And and look, there's been a lot of speculation. It gets to the point where like you ask, why is Intel up now? It like it's and I wrote it this morning. I think it's Donald Trump wants the stock to go up. I think it's as simple as as that. Like they're signing deals with other potential customers like mostly equity deals. So the the government took an equity stake was $9 billion, which by the money that Intel was supposed to get for free anyways. They're supposed to get this under the CHIPS Act, right? Lathan came in and said, we want some stock, which fine. I mean, look, better better to you have to remember Trump was calling for Intel's CEO's head, right, not that long ago. So maybe that's worth 10% of the company to not have them doing that to have them on your side. Um, and then they've since signed equity deals. They they gave SoftBank took a $2 billion stake. Nvidia recently took a $5 billion stake and and signed a little product deal with them as well. Um, and maybe we'll see others, but my guess is the equity stakes are probably more at the behest of of the government. And frankly, Trump literally tweeted out a picture, an AI generated picture of himself watching Intel's stock price go up. He I thought it was fake when I saw it and to it was absolutely real. He tweeted it or Truthd it or whatever you want to call socialed it. So that's a bull case I guess of sorts, right? Um, but as far as you're concerned, nothing's really been fixed. I think absolutely nothing's been fixed been been fixed. And you have to remember all the money is good. Intel doesn't really I mean, they need money, but I mean, they're not in imminent danger of bankruptcy yet. Like they've got enough money to operate. Um, what the money would do would be help them build out capacity, but they have no customers, right? And you have to remember like, so there's another bull case which is the government will force customers on customers to use them. I think that's totally ridiculous because I looked this way, if if Intel can prove that they can make parts in high volume that meet spec at a good cost that are available, yeah, which is table stakes for if they can do that, they'll have customers lined up around the block to use them. People want to use Intel, but if they can't prove that they can do it, what customer of their right mind would ever put any meaningful volume. They're nobody, right? What they need is capability, right? Capability would let them attract customers.
What do you think of their their foundry thing? The problem is they're having trouble making parts for themselves, let alone anybody else. Right. There's no customers. And now they're trying to tell us that eight, they said eight. So they they've had they've had this whole what what the prior CEO, by the I should even step back. The prior CEO was Pat Gelsinger. Foundry was his thing. He thought that was the the salvation of the company. But the problem was his view was, if we build it, they will come. And and can I can I curse on this on this podcast? He invested absolutely absolute tons of money that they didn't have to build out capacity that they had no volume to fill. Um, and and frankly, you look at, I think their their manufacturing, their foundry business last year lost $13 billion or something. I mean, they went, they're burning tons of cash. Um, and now and they had this what they called five nodes in four years. They're trying to catch up on the process tech. I wouldn't say that that's gone well. You know, I I could go through the whole thing, but I I mean, they'll say it's success up lately. Hang on. Let let me finish this because this it's it's really amazing. We're still at the cocktail party. We're still at the cocktail party. Um, so you ask like, why do why do they have any foundry business? So they have their current process. It's called 18A. This is the culmination of that five nodes in four years. Um, it's supposed to have its first internal product for for Intel themselves coming out end of this year. It's called Panther Lake. It's a client PC product. More volume next year. No foundry customers on 18A. They're trying to tell us now 18A was never meant to be a foundry node, which in my opinion is That's not what I'm not following you. Say this again. They're making um parts for themselves on this this process node. Don't worry about that. The the name of it is 18A. They call the process 18A. Okay, don't worry about what it means. It's it's the fifth node of their five nodes in four years. Okay. Okay. Five process uh technology transitions for themselves. They were supposed to do uh other customers on 18A as well as as a foundry. Foundry. So they were just making chips for themselves. Just for themselves. Okay. So they're supposed to make they're telling us now that 18A itself was never meant to be a foundry node. We're never meant to make products for other customers. That is not what they were telling us before. Um, I think what it is is true is that they they did not engage properly with customers on 18A. Now they're saying the real foundry node is going to be 14A, right, which is the next the next one. And and they're trying to engage with customers early and they'll have an 18A version for foundry customers at some point, probably the next couple years. It's called 18AP, right? But there's no there's no volume. There's no customers yet. And and look, there's been a lot of speculation. It gets to the point where like you ask, why is Intel up now? It like it's and I wrote it this morning. I think it's Donald Trump wants the stock to go up. I think it's as simple as as that. Like they're signing deals with other potential customers like mostly equity deals. So the the government took an equity stake was $9 billion, which by the money that Intel was supposed to get for free anyways. They're supposed to get this under the CHIPS Act, right? Lathan came in and said, we want some stock, which fine. I mean, look, better better to you have to remember Trump was calling for Intel's CEO's head, right, not that long ago. So maybe that's worth 10% of the company to not have them doing that to have them on your side. Um, and then they've since signed equity deals. They they gave SoftBank took a $2 billion stake. Nvidia recently took a $5 billion stake and and signed a little product deal with them as well. Um, and maybe we'll see others, but my guess is the equity stakes are probably more at the behest of of the government. And frankly, Trump literally tweeted out a picture, an AI generated picture of himself watching Intel's stock price go up. He I thought it was fake when I saw it and to it was absolutely real. He tweeted it or Truthd it or whatever you want to call socialed it. So that's a bull case I guess of sorts, right? Um, but as far as you're concerned, nothing's really been fixed. I think absolutely nothing's been fixed been been fixed. And you have to remember all the money is good. Intel doesn't really I mean, they need money, but I mean, they're not in imminent danger of bankruptcy yet. Like they've got enough money to operate. Um, what the money would do would be help them build out capacity, but they have no customers, right? And you have to remember like, so there's another bull case which is the government will force customers on customers to use them. I think that's totally ridiculous because I looked this way, if if Intel can prove that they can make parts in high volume that meet spec at a good cost that are available, yeah, which is table stakes for if they can do that, they'll have customers lined up around the block to use them. People want to use Intel, but if they can't prove that they can do it, what customer of their right mind would ever put any meaningful volume. They're nobody, right? What they need is capability, right? Capability would let them attract customers.
What do you think of the CEO? Well, and but and finally, their lack of capability has nothing to do with money, right? They've had plenty of money. It's their processes. It's their and and they've had plenty of they've just made bad choices. So, what do you think of the CEO? So, I like Lipu. I've known Lipu a long time. And I think if anybody can can fix Intel, it it's it's him. Now, that being said, he's not a magician either. He's got to play the hand that he's dealt. It's not a strong hand, but you can see some of the stuff that he is doing like it it works versus like the prior when Pat, so when when Pat Gelsinger came in, you know, like people thought he was a white knight in the stock white group. I'll be honest, like and I he was in my view, he was delusionally optimistic. You have to remember he came in, I don't know what kind of a CEO comes into a turnaround and acts like Pollyanna, which which he did. He literally said, I've looked at everything. Everything's fixed. AMD's in our rearview mirror. He talked about the stock being a double double. We're going to double the earnings and double the multiple. He definitely doubled the multiple because the earnings like went to zero, right? I mean, and he put out these, I mean, they had an analyst day in February of 22 where they put out these ridiculous targets. We're going to do $120 billion in revenue by 26 with no foundry and I mean, they're doing less than half of that. It was crazy. I originally thought he was just pandering to Congress to get the CHIPS Act passed, because that's a lot of the CHIPS Act was came from him. I've since come to to the view that he believed those targets. Wow. And and he finally like they finally booted him, you know. Yeah. So, this guy's good. Lipu was good. He hasn't come in to be he's ly tried to he said, we're going to underpromise and overdeliver and it's all about the customer. He's trying to get new engineering leadership in there. He's trying to remove like layers and and get much farther down into the organization so he can know what's going he's doing absolutely the right thing. I I like Lipu a lot. Can he turn it around? Call Call me in three or four years. I I don't know. Okay.
This has been great, Stacey. Thank you very much. Oh, you bet. My pleasure. Anytime. Wonderful interview. I learned a lot. Yeah. Thank you. You bet.
[Music]
This podcast is for informational purposes only and does not constitute investment advice. The hosts and guests may hold positions in stocks discussed. Opinions expressed are their own and not recommendations. Please do your own due diligence and consult a licensed financial adviser before making any investment decisions.