The Six Five Pod | EP 278: The Foundation of Future AI: From Micron's Memory Surge to Marvell's XPU and Nvidia's $100 Billion Blueprint

On this episode of The Six Five Pod, hosts Patrick Moorhead and Daniel Newman discuss the tech news stories that made headlines this week. The handpicked topics for this week are:

1. AI and Infrastructure Investments: Nvidia and OpenAI announce a major AI infrastructure investment. $100 billion phased investment from Nvidia to OpenAI. Pat & Dan discuss the strategic implications and industry reactions.

‍

2. Oracle Leadership Changes: New co-CEOs appointed –nClay Magouyrk and Mike Sicilia. Safra Catz becomes Oracle executive vice chair. Hosts analyze this leadership transition and its potential impact.

‍

3. TikTok Developments: Executive order to push final decision on TikTok to mid-December. Potential spin-out authorized, contingent on Chinese agreement. Oracle and Silver Lake are potential key players in the deal.

‍

4. Microsoft's AI Strategy: Microsoft adds Anthropic to Copilot. Discussion on the benefits of model diversity in AI offerings.

‍

5. Qualcomm Snapdragon Summit Highlights: Introduction of X2 Elite and X2 Elite Extreme chips, focusing on AI capabilities and performance improvements. Plus, a look at Qualcomm's strategic positioning beyond smartphones.

‍

6. Cloudera Event in New York: Patrick's insights from the Cloudera event and a discussion on hybrid multi-cloud fabrics and AI ecosystems.

‍

7. The Flip: Should Apple or TSMC Invest in Intel? Hosts debate this topic with arguments for and against potential investments in Intel. Plus, an analysis of the geopolitical and technological implications.

‍

8. Market Trends and Earnings: Federal Reserve Chair Powell's comments on stock valuations and a discussion on Micron's earnings and the impact of AI on memory demand. Marvell's special presentation and stock performance.

‍

For a deeper dive into each topic, please click on the links above. Be sure to subscribe to The Six Five Pod so you never miss an episode.

‍

Or listen to the audio here:

Disclaimer: The Six Five Pod is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.

‍

Patrick Moorhead: We absolutely know what we're talking about on the show. This is episode 278. This is Dan and Patrick. We are coming back from being on the road. Daniel and I probably got in bed at 1am last night, but we are here to crush it for you. Daniel, how are you doing, man? Did you get that, did you get that workout in?

‍

Daniel Newman: I did. I didn't go on TV. I'm not famous or important, but I did get a workout in and I'm pumped this morning for this show because man, you look good. I mean, I just, I don't know what to do with that. You're, you're wearing, you know. Can you wear that color though? I mean, you're usually pretty good about that. Are you a little springy for the time of year? Do I have to call you out on this? Do I have to do fashion?

‍

Patrick Moorhead: You know, the problem is, I wear this dark coat on air all the time, many times and it's kind of during my fat days and it's, it's, you know, I peel it in. I don't know why. I just don't buy a new coat.

‍

Daniel Newman: Yeah, you can, you can get there, you can get there. You look good. But you get a new black coat, a new gray coat. No, it's a good, good looking jacket. I just, I've never cared about things like you can't wear white after light. Like I don't care. But you've sometimes called people out on it. So I'm just kind of coming full circle in you.

‍

Patrick Moorhead: You know, I feel a payback. This is a payback for sure. Yeah. And I felt like I had to do that. And I don't know, I see your gray coat looks pretty good.

‍

Daniel Newman: How funny, by the way, is it that you and I randomly ended up in New York? Randomly ended up like, you know, people probably don't realize, like they think because we're usually together so often, but like the fact that we ended up completely at random at two different events in New York speaking on the same day and then came home on the same flight.

‍

Patrick Moorhead: No, I know it was weird and normally when we say random, it's not random, we're just joking. But it was totally random that you flew in on a red eye from the Qualcomm event. And you know, I was doing some Pre-work for the Cloud Arrow event. But no, it's always great to see my bestie and, and it's too bad we didn't get to the city to sit next to each other but it would have been boring because I slept most of the trip.

‍

Daniel Newman: You did. I saw you. I turned around, mask on both masks. You had the face mask, the eye mask. Very impressive, very impressive.

‍

Patrick Moorhead: I should have probably worn a face mask by the way.

‍

Daniel Newman: I think I saw you in a shield full PPI or.

‍

Patrick Moorhead: No, it's actually just a full rubber suit like they were in the foundries. So.

‍

Daniel Newman: Yeah. Well, there you go.

‍

Patrick Moorhead: There we go. Hey, we had a great show for you today. I mean so much stuff happened. I mean all these announcements. Dan and I were talking earlier this week about keeping up on everything that's going on, it is getting quite hard. But we've got a lot of Nvidia, Oracle, AI gosh, TikTok, Microsoft, more Intel rumors. Imagine that we got the Qualcomm Snap, Snapdragon summit, new CEOs being appointed. Just a lot of stuff going on. So hey, let's dive into the decode and figure out all the decode news of the week. All right, Daniel, we've got Nvidia OpenAI announcing the biggest infrastructure investment in the history of the world. What's going on here?

‍

Daniel Newman: Well, I mean in the spirit of it wasn't enough that last week the Nvidia intel deal shook the world. You know, it's like what is the next thing? So this is a really interesting one after the Broadcom sort of big splash at its earnings about building a chip for OpenAI and there was a $10 billion sort of number that showed up and all these chips that are going to be expected. Nvidia comes in and says OpenAI is the best or among the best in this space and we want to invest $100 billion to help them build out more capacity and grow and scale. Now of course, instantaneously that investment, you know, the whole world comes up in arms and says, oh, they're basically just giving money to OpenAI. So OpenAI can then buy capacity from Oracle, which then Oracle will buy chips from Nvidia. And it's this big circular thing which I think this one's a super bifurcated thing like a company like Nvidia that has a big balance sheet, lots of cash, certainly can choose its investments and if it chooses to invest in companies that will potentially use their technology, that's pretty common in accelerators and incubators like that's actually oftentimes a big part of the investment thesis and criteria is like, we're going to invest in a company that could benefit from our ecosystem. But of course, you've just got a world now where there's so many people that just want this to be a bubble. And so there's a lot of negativity now. You know, a couple things that are interesting. Here is like, one is this thing is not a. Like here's $100 billion. Just go, this is going to be done in phases. This is going to be done as the energy and there's compute gets built out and scales. This is obviously not. And this was an Nvidia comment. They wanted it to be clear, this is not a, hey, use this money to buy compute. Having said that, compute is the number one thing they're going to need to invest in, to scale. So it's kind of like one of those, like, no, you don't have to use it, but you probably will use it for that. Cause. But I think it's important that we demarcate that because there's no strings, technically speaking. There's nothing legally from a strong standpoint that says that's how the money has to be spent. But OpenAI has definitely been one of the biggest winners. They've got the highest valuation, they are the most popular at this point. Generative LLM platform with a ton of research innovation and a huge potential IPO or some sort of in its future. So, Pat, this is smart business right now. Jensen's peppering the market and buying into and investing in the companies that are basically building the rails for the AI future. Hard to blame them for it, but it's also hard to in some ways blame all the people that are skeptical because what you do have right now is the same handful of companies doing a lot of winning and we're still waiting for all that consumption to really take place. Starting to move down the stack into applications and into tools. Who's making money? That's what everyone's watching. And they. Until that happens, I think there's gonna be a lot of pressure that there's a bubble. I just think it's, it's ingestion.

‍

Patrick Moorhead: Yeah, Daniel, I have gone so many different ways on this. And you know, this is not a, just a classic venture investment because it's over years. So there are steps, there are milestones, so there are strings to it. And when you're a VC and you're putting in money, throwing into the pot, that's typically not what you can get. And then on the other side of my mouth it's hands off because they're not getting a board seat. Right. It's in Nvidia's best interest to have OpenAI continue to look at alternatives like AMD, which I think they are still doing. I still think that they're going to do an XPU with Broadcom. I don't think it changes the calculus. In fact Nvidia would want them to do this. What it does do is, is lock in a certain amount of Blackwell GPUs for, for the company. How many? We don't know. It's probably a handshake deal or a verbal. A verbal deal. You know, at the same time we got, you know, this Oracle OpenAI, SoftBank, Stargate expanding it and I think it they reiterated the $500 billion 10 gigawatt by the end of 2025 schedule. They did talk about the number of on site jobs. So they really, you know, needed to make this more real. I think particularly since Microsoft came out with its Wisconsin Fairwater where you know, they're showing cooling, working electricity, they're showing everything. And I love that we're getting into used to be in a CHIPS arm race and now we're in this data center race. Right. We now have XAI and Microsoft. I don't think Oracle has made the claim yet, but the most performant AI data center on the planet. So Stargate's the strategy and Wisconsin the blueprint.

‍

Daniel Newman: Yeah, I think that's a good point. I think there's a lot of people also that want to make this zero sum, that oh it'll be Nvidia and this is going to push out Broadcom or push out amd. I still think it's kind of an all hands on deck thing. I think all these companies still want to build their own infrastructure. Yeah. But I also think that's two or three years in the future. And so the bottom line is even if they get to a good infrastructure product Pat, it's not like they're going to be able to just turn on their custom chip tomorrow and tear out Nvidia. I mean I think the bottom line is we see like three, three to four fifths of all the chips over the next five years being GPUs for AI. Yeah. So it's like there's still going to be a both. Like I'm so tired of debating with people, is it going to be one or the other? It's going to be both.

‍

Patrick Moorhead: Yeah. No, I agree. Hey, a lot of big Oracle news too. Gosh, Oracle's in the news as much as, as any of these folks. And we saw some changes at the top. We got two new co CEOs coming in, both who've been on the Six Five, Mike Sicilia and Clay Magouryk. So you have, you know, OCI meets industry, industry leadership and then Safra becomes the executive vice chair. So you know, it looks like an orderly transition. And you know, when you're talking about this much capex coming in, putting Clay into one of the driver's seats makes sense. And, and then Mike, as it relates to the entire software stack and all the industry that goes with it, I think it makes sense. I don't think that two in a box is a perpetual thing. 99% of the time they come in and they pick one. It's funny, companies like Huawei and Samsung have rotating leaders that come in. I don't know if they're going to be doing it, they're probably not going to be doing that, but I think that's, it's absolutely fascinating. And then by the way, you know, they get this day job too with TikTok to you know, Oracle is, is part of the contingent to babysit the new algorithm that they're gonna have to create here which you know, it's so funny, they said the valuation of US TikTok was $14 billion. You got to be kidding me. I mean you probably understand math more than I do. I don't understand that. But essentially the president did do an executive order to not only push the final decision to mid December, but authorized a spin out. And it's contingent right now upon the Chinese agreeing to this deal. And you would have Oracle and Silver Lake who would be the primes on this. The algorithm would have to be rewritten. And it's so funny, people are thinking, oh, oh, it's going to be so easy to rewrite the TikTok algorithm. My gosh, the TikTok algorithm is the highest level of value and the biggest differentiation that goes into that thing. So this is not a trivial thing to do. And then I'm wondering who's going to do that, Daniel, is it, you know, they're going to start a new company. Are they going to hire everybody from Meta that Zuck hired in his hiring spree? Anyways, in the next three months we're gonna have to see where this thing lands.

‍

Daniel Newman: Yeah, I mean my read on the, the CEO is very similar to yours. I have not seen historically dual CEOs work particularly well. So it'll be interesting to kind of understand how they're going to split. I think I'm guessing it's kind of a bit of a cloud side and then industry side would be my guess, but I don't know for sure. But we met both these folks. I think they've both been on our Six Five pod at different times before they became CEOs. But this one will play out. I do think it'll probably end up landing with one. I mean what a tenure from Safra. You know, she wasn't a super public CEO in terms of like a regular, you know, she made her appearances very rare but the execution was great. And of course, you know, the one thing about that company is Larry's always there.

‍

Patrick Moorhead: So live to be 150 folks.

‍

Daniel Newman: So he is, he's a mover and a shaker. So he, you know, Larry's, Larry looks better at 80. Was he 81, 82, you know, then I did at 40. So good for him. He's well put together. I mean the Tick tock thing, I don't know anyone our age that's using tick tock should ask why.

‍

Patrick Moorhead: You know, I put my weekly video on there, so.

‍

Daniel Newman: I'm kidding. I gotta give you, I gotta give you crap for it though because like, no, I know it's because I did forget about that. But you know, it is creepy.

‍

Patrick Moorhead: It is a creepy platform. I mean anybody can message you, anybody. I mean I look at some of the people following and it's just half of them look like bots. But yeah, it is interesting. Hey, let's jump in. In the ever present open model war and AI use case war, we now have Microsoft adding anthropic to Copilot. What's going on here, Daniel?

‍

Daniel Newman: Just model diversity, this is a pretty straightforward one to me. I mean one is we saw how straight all in commitment to OpenAI. It looks like it might have a happier ending, but there was certainly a potential of a trail of tears that was shaping out there and all the cloud providers have been doing this moving towards a model diversity and Azure AI, that's been something they've been doing for a while. But in Copilot it's also like, hey, as you're building your enterprise solutions and you're trying to deploy Copilot in your applications, which, this is where the AI is going to be consumed, this is where enterprises are going to get value from it. It's not always one size fits all with models. And Anthropic has some tremendously good models and they're very respected, reputed and competitive. And for some companies, certain outputs, maybe it's related to tables and creation of numerical content where Anthropic seems to do very well versus ChatGPT, which does very well with contextual text and semantic. And so you might want to pick that. You might want to give your vibe coders and developers the opportunity to play, you know, play and flex across these different models. And you know, adding Anthropic was just a signal that, hey, you know, yes, we're committed to copilot with OpenAI, but we are also committed to giving optionality to our customers.

‍

Patrick Moorhead: Yeah, I mean spot on, Daniel. Competition is good. And while IBM was the first company to really lean into open models and then AWS, Microsoft has embraced it as well. And I think they always did embrace it. But when you're leaning into OpenAI and OpenAI is crushing it, right, You. You have to give precedence to that. Plus Microsoft owns a certain percentage of the. The. Is it the nonprofit or the. Or the profit Daniel there. So it's a pretty darn good investment if you truly believe they can get to $125 billion of revenue in 2029. So anthropic is going into Microsoft 365 copilot and it's going into Copilot Studio where you can make your own agents and products. And my gosh, my son Pico just absolutely loves Claude code.

‍

Daniel Newman: Yeah, we talked about that at lunch last week.

‍

Patrick Moorhead: Just absolutely vibe coding to the world. So let's jump to the next topic. Gosh, you know, coming off a week that Nvidia invested in intel, they're now in more intel rumors here and one of them is about Apple. And you know, for Apple and Intel. It really would be pragmatic and prudent for Apple to be doing this. Risks them from tariff risks, capacity risks, and Apple is one of the most competent teams out there, tons of resources. And I. And they are a company who could pull off a dual source between TSMC and Intel and it just makes sense.

‍

Daniel Newman: So look, this is smart business. Maybe we could debate this in my opinion, though, Intel's, it's sending around the, what do you call it, you know, the, the collection basket right now. And it's basically saying if you believe in American exceptionalism and the comeback of American manufacturing of semiconductors, you need to invest in intel and, you know, a lot of kinds of bad takes out there. Well, you know, Apple doesn't use intel for any chips. It doesn't matter. The real question is, does Apple want to get access to capacity if, say, President Trump changes its deal with TSMC and says TSMC needs to do A, B, C, D and E in order to continue to not pay big tariffs, and not have big expenses? So, you know, there's a lot of different ways this could play out. This could play out with different, you know, some different IP types of sharing agreements. This could play out with talent sort of being used in, in development, in the development and R D side. This could play out with, you know, a certain amount of, of each of the Fabless needing to be produced at an intel foundry. So that basically TSMC needs to enable intel to, to handle 20% or some volume, which is the optimal outcome that I always thought intel needed was sort of a forced hand. Now again, we can argue whether that's the best thing. I'm not agreeing that it's the best thing. I'm just saying that a forced hand could create the urgency for the IP sharing required for Intel and TSMC to be compatible, to be successful. So that's what I think is ultimately happening. I think there's a bit of a, you know, process of going around and raising money from everybody that could benefit. And I think you and I both said this in different variations after the Nvidia investment. That's like, look for the rest. I think I tweeted after and a lot of people thought I was crazy. I mean, right after the Nvidia, when I said, watch, Apple will come, there's gonna be an Apple investment next. So I didn't necessarily predict that it would be solicited by Intel. I thought Apple might just run and do it because they want to get the, you know, the political capital and maybe a little forgiveness for its slow movement out of China or anything else it wants to do by agreeing to maybe put some money into this intel investment.

‍

Patrick Moorhead: Yeah, it's interesting. We can talk about this a little bit more in the flip, but I think tsmc, I think, think where they could benefit from intel the most would probably be packaging. I mean, Intel's packaging is second to none in terms of optionality. They've got capacity and I think it's a big opportunity for them. Hey, let's jump into the Qualcomm Snapdragon summit. Daniel, you were all over the airwaves talking about this. You were actually in Hawaii. I was watching it from my hotel room in New York City on my PC, sitting in bed. And it wasn't the same. But it is sometimes easier to pay attention when you're not, you know, sitting in a room full of people.

‍

Daniel Newman: There's lots of announcements going on. You know, Snapdragon continues to really focus on building those next generation, highest end, highest performing chips for, and again, not just for phones. And that's one of the biggest challenges of Qualcomm. And Snapdragon is being seen as more than a handset company. And I think they're working really hard to do that. You know, first of all, they obviously are going with what's next on the, on the phone. I think one of the most interesting things for the handsets to me was the work that's being done with the PC. We knew that last year the AI PC was going to be this kind of big trend. Did it end up being a super cycle? It did not. Will it end up being a super cycle? Not until those killer apps are figured out. I think it's still going to move a little slow, but we still want better performance out of every device. And so if you're using a PC and you want to have highest performance, you want to be able to run models right on your device, you want to have best graphics, best imagery camera, you know, Qualcomm and quote unquote, Snapdragon, which by the way, they really don't use Qualcomm at all. It's been a hard habit for me to break with it because I still call, kind of think of them together and actually their CMO, Don McGuire did a big thing about Snapdragon. In fact, the brand itself now is apparently worth $65 billion. What was what? Yeah, just the Snapdragon brand alone, you wouldn't have seen a Qualcomm sign anywhere in Hawaii. There wasn't one. But anyways, you know, just quickly and because there's a bunch of different things you could, we could touch on. But I thought what really caught my attention was the X2 Elite and the X2 Elite Extreme, which by the way, it's not really going to be two different series. They're not going to create a perpetual like extreme series that's going to be required every year. But this year they, they basically did find a way to basically crank out some serious performance. You know, I think it was like 50% better multi core performance and almost 40% on the single core. But I also, you know, it's 18 cores now, which is about a 1/3 improvement from 12. And this is the third generation Orion CPU. You know, overall though, Pat, you know, I'm looking forward to signal 65 on our team doing a lot of the testing, didn't do as much comparison with Apple this time. More focused on comparing with the other traditional PC companies, Windows companies. But they did definitely show some good results. Very, very high powered, very fast. I use one of these every day, Pat, you know, and I definitely enjoy the kind of longer battery life which I think they're seeing. Like, you know, almost, almost 43, 40, I can't remember 45, 43% more power efficient than the last generation. So that's another really good thing with X2 and X2 Elite Extreme. So you know, like I said, they had, you know, announcements around handsets, announcements around wearables, announcements around the devices and I think they got a lot more going on. I mean, you know, I talked about this, this is more what I talked about on the TV's path. It was like, I think Qualcomm's interesting because they also have a lot going on. Like with automotive, they've got stuff going on with data centers, they've got stuff going on with robotics which will be super interesting. And this whole kind of narrative around personal AI is going to be an inflection to see if the company can really figure out a way to be the critical component to that everyday personal assistant experience that is going to be AI because faster phones, faster chips, longer battery life. It reminds me a lot of what I kind of say that has bored me about Apple. So there has to be something more and so it has to be something more. This stuff has to start to turn the corner and make AI super personal and super available. And CEO Cristiano Mann did do a whole keynote just about that.

‍

Patrick Moorhead: Yeah, that's a good breakdown. Daniel. Two things for me. First of all, the strategic framing of Qualcomm for the next five years. And Christiano went through six things. AI is the new ui, smartphone centric to agent centric, the required changes in computing architectures, models becoming hybrid edge, data being relevant for personalization, and then the future of perceptive networks. He has talked about this before in different ways, but I think this was the most mature encapsulation of this and I while this wasn't an investor conference, if you look at the downstream opportunities for the company beyond, beyond the smartphone which doesn't necessarily have to be the center of anything anymore. It could be your watch, it could be your, your meta glasses, it could be something in your ear. I thought it gave a good idea where at a minimum where they are investing and we should, should take hold. And then you connect that with the industrial IoT you connect this with the automotive and in the Middle East they had made, they had talked about a future data center product. I think the Qualcomm of five years from now is not going to be the Qualcomm that we see today. The benchmarks of the smartphone and PC Snapdragons were literally off the chain. And what I listen to without Signal 65 doing the testing in an OEM system. What I can say is that at a minimum they are equivalent with Apple on the smartphone. And then when it comes to the PC you know, we don't know what AMD is cooking up, we don't know what Panther Lake is going to look like but my gosh, those benchmarks were so far ahead on single threaded and multi threaded it was just an absolute jaw dropper. And architecturally on the highest end on the extreme part they integrate memory in a form factor called a sip. And what it does is it removes latency, improves performance for bandwidth and I think versus an M4 and an M4 Pro. I feel very comfortable that Qualcomm is going to be super competitive here. I'll be doing my rounds with all the OEMs and Microsoft to get to see what they thought about this as well but overall, overall it was good. So Daniel, while you were in Hawaii I was hanging out in, I was hanging out at Cloudera in, in New York and I, I put a lot out there on social media but here's the net net this whole idea of hybrid multi cloud fabrics that I've been talking about I and I annoy people about Daniel or you can sign up for one company and you can flex between on prem Legacy, on prem cloud and all all of your public cloud. They have finally architected that system and I think will be available by the end of 2026 with agentic frameworks, with toolkits with an end to end AI ecosystem. So they absolutely made I, I wouldn't say put the hurt on Snowflake and Databricks because they're crushing it out there but I think what they did is they brought out a more compelling offering compared to them who both of their competitors don't really even incorporate on prem data. So Daniel. Okay, let's get into the flip here. We are going to debate whether Apple or TSMC invest in Intel? Let's flip the coin and see what we get back here.

‍

Patrick Moorhead: These are the four they absolutely should invest in. I feel like this is the easier of our simulated discussion points. So I think the 4 on this is for Apple. It's, it's de-risking, right? If you look at all of the silicon that's being forecasted for AI until 2030, TSMC does not have the capacity to do that. And well, what does that mean? I think you'll see TSMC announce a bunch of new foundries too, to fill in the gap. But what it does do is it makes Nvidia TSMC's largest customer and the customer they can make the most money off of. Before this it was Apple. So it's not that Apple's not important to TSMC anymore. They are. But I think that Apple needs optionality and I think by investing in intel they can give them the capacity to de risk not only from a tariff standpoint but also if nothing else, just a volume standpoint. And they could do it in a way that de-risks it where they don't have to go all in and not use TSMC for a certain price product. And listen, I'm not confused. It takes a lot of resources to be able to do that. But Apple has the resources that I think has the biggest motivation next to Nvidia to be running wafers through them. An investment would make sense I think from a TSMC standpoint. I think that they would, they would look very good in the eyes of the United States that if they aren't going to build some mega factories then they can at least fund intel who can fill up Columbus, Ohio and ostensibly compete with them. But they could also watch the intel stock that would likely go up in the stratosphere if they got a wafer customer like Nvidia or Apple. So I think it'd be financial upside for both of them as well as my gosh, can you imagine if, if, if Apple invests in intel what that would do to the stock with just a lingering of a, of a yes, we are going to run either wafers or packaging through them. So yes, they should be investing in Intel.

‍

Daniel Newman: I don't see any reason that either of these companies, you know, superior process, superior technology you know, Apple ditched Intel a long time ago because it could do better on its own. It took only a couple years and ever since then its products work better. You know, you want to go back and use an intel powered phone. Apple, you want to go back and use an Intel powered MacBook? I don't think so. Everybody that's experienced the benefit of these Apple architected chips and any of their devices have found a better experience. They just outperform just the power, just the power envelope. Just the power envelope alone is enough. But everything else, smoother, smoother graphics, everything that you run on that device is better without Intel. So I bring the intel back into the folder. You know, it's like when you finally got out of that terrible relationship with that person that was mistreating you, are you gonna run back to them? Don't be Stockholm syndrome. For Apple to go back, it would be like Stockholm syndrome to run back to Intel. What are they gonna beg for a little capacity in an imaginary foundry that has no way for customers? Oh, we need you so badly. What is it just because Nvidia did something? No, Apple runs its own race and it's gonna continue to do that. And freaking tsmc. TSMC doesn't need Intel. TSMC is the controller of the AI world right now. They have more deals. You know, everyone thinks Nvidia controls the flow. Nvidia. Because if all TSMC has to do is say, look, we're going to give a little more capacity to someone else this month, Nvidia's numbers plummet. And you know what? We've seen that the power of TSMC has been in its ability to control its pricing. Probably arguably the only monopoly bigger than Apple, the Apple App Store is TSMC itself right now. Because there is no way to get done what you need to get done without tsmc. So again, what are you going to do? Prop up your biggest competitor? You're going to help your biggest competitor get back on their feet. We all know how that worked when Microsoft put a few dollars into Apple and then for the next 20 years it had to look up at Apple's market cap. Yeah, Microsoft caught up finally about three CEOs later and with a little luck in a massive trend line. But I don't think that NTSMC wants to pave the way. What's next? Are we going to give Blackwell and Ruben to China so they can catch up with us too? This is a terrible idea. Terrible investment thesis. Don't do it. Don't give your money away. Just set it on Fire dump it in the ocean. It would be a better use of your dollars than giving it to Intel. Oh, by the way, this is a simulated argument. I don't actually believe any of that.

‍

Patrick Moorhead: Now what do you really think, Daniel?

‍

Daniel Newman: I mean of course you gotta, I'm the American exceptionalism guy. Like I want intel back. So you know, and by the way, I don't, it's not an Intel love affair for me. I respect Intel and I want to see Intel do well, but it's more. I just don't want to be dependent. These are just some technologies that are too critical and TSMC is kind of doing half the business. They've brought half of what we need here. But the IP thing's a problem. I can't tell you. Do too. How many arguments do you get on with people online when you're explaining this and they just give you dumb comments about how TSMC has already got plants here. Yes, we understand that, but that doesn't mean they gave us any of the IP that we would require. If China, you know, sent the Navy in the South China Sea to undo Taiwan, we're going to be stuck in current generation technology for the next 20 years. We need competitive offerings and so that's why we have to do this. Apple has the balance sheet, you know, it spends 600 billion on buybacks. Why not give $10 billion to Intel to help it get its foundry business up and running. Maybe like you said, maybe it's a packaging deal to start because that seems to be where Intel is doing its best work right now with wafer opportunities in the future when the process gets there. And of course tsmc, you want to not pay tariffs, you want to have an exemption to be able to keep doing business here. Maybe you're going to share some of that critical IP or maybe you're going to help us get processes to parity in an area where we can at least manufacture some percentage of the chips here in the US and by the way, that's a big ask, it's a big if. But the alternative is they're going to have to look at the bottom line. The bottom line is, if he puts 50% tariffs on these things, that's going to totally change the calculus. But I also don't know, I mean with the amount of money being spent, maybe that would all be absorbed. I mean they keep raising prices 20 or 25% of the time and everybody just pays it.

‍

Patrick Moorhead: Yeah, and this whole one to one tariff thing reminds me of we're going to shut down the government. Right. Like nobody believes any of that nonsense. Okay. And yeah, another topic, good discussion here, folks. Hey, let's go to the last segment called Bulls and Bears. We're going to talk about earnings trends in the market and stocks, basically. Stonks. That's where we are right now. Let's hit it. All right. As you would expect, Paul, in his words, stock prices appear fairly highly valued. Danny, what is it? What does that even mean?

‍

Daniel Newman: Oh well, I mean, I think Fed chair Paul was probably saying that look, we've had a nice run up, stocks are valued well. We've got higher, looking for higher forward multiples than we've had in a long time. You know, there's these comparisons everyone keeps making to the bubble of the Internet. And while it isn't quite that frothy and the companies I think are much more well funded and, and have much more diverse moats than some of those companies did back then. We've seen, you know, some stock prices, especially some companies like the Oclos and the Palantirs that have run hundreds if not thousands of percentages up before they really hit any revenue scale or capacity. So I think what he's saying is that the big companies have been valued pretty fairly. Even companies like Google that were well behind and well undervalued have sort of caught up to value after some of the, you know, some momentum and some policy decisions were made. And that, you know, while we still have some employment strain and we still have some inflation strain, maybe assets have come out a little hot and so the market hated it. They've been selling all week. It's been down ever since he said that. But at the same time, like markets are efficient, you need a little selling. Like you can't just go up into the right. So while it frustrates people, you know, and by the way, while I, I think that chair Paul has been super political, I actually don't think he's wrong about this. I think he's actually pretty right. I've seen some of my things triple and I've had stocks back that have tripled in value since April. Were they undervalued and now overvalued? I don't know, but I think that's a pretty good six month return. You know, like if inflation's 3% and I'm getting 300, I'm, I'm pretty happy. And if it comes back a little bit, the last thing I'll say is, you know, everybody always screams like oh my God, the stock price went up so much. I should have bought more. Good, let it come down now you can buy more. So I don't know like I think he's, I think he was pretty right on this one. I think the equity markets have had a very, very nice run. And you've seen it with the NASDAQ 100, you've seen it with the S and P. By the way, last point I'll make on this is every analyst got this year wrong. Like every, every one of these banks got the S and P wrong. They all got inflation wrong. Because retrofault we didn't call out here also kind of called out that inflation didn't do what he thought it would do.

‍

Patrick Moorhead: So I was, that was really surprising and it was good to see, you know he didn't say I'm wrong but he said it in the kind of weasel cryptic words that he used. And you know the message was really just about two, you know, two sides of the coin. Kind of like the flip, inflation tilts up and employment tilting down. Right. So you know what, what is there, what is there to do here? Yeah, the Fed didn't call it a bubble but it also didn't say oh this is a bargain. So it was kind of in the middle tilting toward, toward the bubble. So let's get into some earnings stocks and some chip companies. So Micron, the crush on every vector you could possibly see, it's all in the back of HBM. I hopped on Yahoo Finance and talked through that in the context of are we in a bubble or not? And also talked about circular, circular purchasing. We, we have here nand, NAND and dram right that nobody wants to talk anymore. That is getting lower supply, higher demand and therefore prices are going up. This is just the circle of life in the memory and storage market here. Possible tailwind on this 6.2 billion dollar CHIPS act agreement here and the first disbursements are finally flowing in there. The US government didn't make an equity investment in Micron like they did in intel, but they still have cash coming in. You know there was some, I would say some technical jargon, technical discussions on X hey this HBM4 base die. They tried to do it themselves, but they failed. They had announced they were moving it over to tsmc. You know one way to look at that is that hey it's great a higher probability of execution. I think the negative bear case is that hit a hit on, hit on profits. Hynix is doing the same thing. I don't know if it matters, but it's just amazing to me that we are spending this much time on companies that people didn't even care about years ago. It's amazing. Amazing. But the depth of the conversations, you're. Ray Wang probably did like 27 posts dissecting the HBM technology. It's pretty amazing.

‍

Daniel Newman: Yeah, I mean, look, I, you know, I've been kind of screaming and yelling that I think Micron's one of these lagging beneficiaries of. Of AI. People are gonna have to. There will always see, this is one of the weird things, Pat. There will always be something that'll look more boom and bust. Even with AI, it's not going to ever just completely smooth. Right? But if you actually look at PCs, for instance, it actually is fairly smooth. The unit differences in a bigger cycle is like 5 to 10%. With memory, you had these cycles where it was like up and then down and then up. And what AI is doing. When you have two annual releases of next generation AI infrastructure coming out by the biggest companies in the world, being deployed by the biggest hyperscalers in the world, in a time when capacity is totally constrained, there's going to be a ton of pull through for the memory because you cannot, you know, as their CEO Sanjay Mumrotra said, you cannot do compute without memory. Like, you literally cannot. There's no way to get around the need. Now, are there other types of memory being developed to potentially scan baby? Yeah, there are things being done. Having said, at this point, HPM which is next, and then in the future generations five and beyond will continue to be utilized to support these next generation AI architectures. As long as that's happening, Micron has that tailwind now. It still has the more boom and busty parts like automotive, it'll have parts like, like it has handsets and it does memory for PCs and that stuff will be less consistent, but that's going to become a smaller and smaller percentage. Kind of like how we saw the pivot with Marvell, which we'll talk about momentarily. When it went away from all one side of its business, all the AI. Micron's kind of moving in that direction too. And if you're sort of tracking the market, if you're investing, you got to look at like, are you looking for the AI tailwind? Having said that, Micron's got to probably have several more quarters of proving that this is true and that what I'm saying is a true theory. But at this point, if our forecasts for AI infrastructure demand continue to align with what we're saying, there's going to be a lot of HBM pull through. And by the way, as the HBM is more scarce, the demand will stay high. And the pricing power, which is something that most of the time memory companies struggled with pricing power during those down cycles, that pricing power will stay higher. So they will see margin enhancement, which we did see this quarter.

‍

Patrick Moorhead: So why don't we jump in Marvel here. Last topic that we're going to be discussing here, Daniel, they did a kind of a special presentation, a little bit of a reframing on a couple things of their stock. Absolutely ripped.

Daniel Newman: Yeah. You know the companies, it's like maybe they listened to Hawk or got a bit of a move out of that Hawk playbook. But like Hawk never really says a win or a design or who likes he's very careful. But at the same time there's no way you get off a couple call without understanding the insinuation that they got that win and that they're going to get that revenue and it's going to come in that forecast. CEO Matt Murphy from Marvell seemed to have a little more of a problem kind of trying to answer that question for, for its investors of what's going on with next gen designs, what's going on with design wins. And right now, let's just be candid, despite the fact that Marvell has a really successful networking business related to all the AI transport, everybody cares about XPU. So getting out and in front of that story and reframing that and winning the market to believe that it is a, it is a strong player that's able to confirm and win designs and keep customers. The other thing like trying to kind of confirm is that this idea of everything being sole source, I mean even Google, if you know, like their latest TPU, there's a MediaTek TPU and there's a Broadcom TPU. There's a very realistic chance that with some of these custom designs that Marvell might get part of a win and another company might get a part of a win as well. But I think just setting the tone that A can't really fully disclose things B, they believe and they're confident in their future and the wins that they have, they're forecasting to it. I think just resettling did a lot for the market.

‍

Patrick Moorhead: Yeah, I think it did. And the buyback didn't hurt as well. It showed. Buybacks always show a lot of confidence. And the crazy part is you just take, you just take Marvell's networking business and, and put it up against some of these smaller networking companies and even the valuation of some of these non-public startups, are incredibly lower from a PE standpoint than any of those. And you know, I won't name names here but you know they're, they're direct competitors so you know something is amiss and I think a reset is required. It also doesn't help that all of Taiwan has local heroes who are trying to leak and damage the reputation of the company as well. You have a lot of short sellers and you got a lot of longs on the competitors that would like to see Marvel go down. And I think they're doing a pretty darn good job. So folks, that is the show here. Thank you for tuning in. I hope you have an amazing week and you spend good time with your family. Maybe you got to relax, got some amazing workouts in you. Pick up your peptides and your supplements, we hope.

‍

Daniel Newman: So many peptides, dude.

‍

Patrick Moorhead: So many, so many peptides. If I can have a peptide that can mean that could make me have to sleep less. I would just load up on that stuff. It'd be crazy.

‍

Daniel Newman: Four hour sleepers. We'll talk about that another time.

‍

Patrick Moorhead: Yeah. Anyways, thanks for tuning in everybody. We appreciate you hitting that subscribe button. Thanks for being part of the community and take care.

‍