What will Micron's gross margin guide be?

0:00Thank you.

0:30See how things have changed since that August episode where we left it last time. So catch us up, maybe rank it in terms of importance or in terms of the scope of your focus, how much how much room it's taking up. How would you share with investors how you've been thinking about the tech space recently? So I think the you know, the first and the most obvious thing that's really happened since we last spoke is that whole, you know, theme of AI safety concerns and that whole, you know, slowdown weekend we had with Anthropics Amodi coming out and saying, we need to pace the frontier, Sam Altman agreeing, and then Elon Musk agreeing as well.

1:04I don't think you really see these three agree on something very often. And so it does stand out, I think, that the three of them came together to say, hey, there's actually some concern. And then if you look back, you know, the couple of, you know, a few weeks that came before that, we also had, I would say, headlines building up to this. And essentially, we had a lot of headlines from, you know, the hugging face incident with OpenAI, to other incidents that conveniently came out back-to-back, even if they had happened earlier this year, but were only recently disclosed, similar to the Gemini hacks that happened in May, but were only, you know, talked about this month.

1:40And then, you know, you think about how conveniently these, you know, hacks built up into the slowdown weekend. And you also have the 27-year-old researcher that resigned from Anthropoc talking about, you know, the end of humanity and these, you know, existential risks that AI really poses. And so when you take all of that, and ironically, you look at that, you look at the market reaction to that, the Monday after that weekend, markets sold off. And that's when we told the Tech & Trans community that big tech doesn't grow conscious overnight. And we don't think that this translates to any real fundamental slowdown in the near term.

2:18Instead, it's much more accurate to look at this in terms of who benefits from this kind of conversation. So that's something that we're going to get into. But just for the sake of looking at the month more broadly, since we last spoke until now, you see that red Monday happen. And then a Monday after, you see the entire market in terms of semis really trade up, actually rally in reaction to MUSE, meta-releasing MUSE and agentic AI. So the market really leaned into the slowdown fear. And then a Monday later, reversed and leaned into the agentic AI greed. So while, yes, we're hearing a lot more, I would say, scary headlines over the past month and since the third quarter began, you know, not just from the UN speeches that we're seeing this week and that resignation, but also from, you know, mounting credit risk, inflationary pressures, rising yields, you know, the U.S.-Iran war and everything happening there.

3:09Yet still the market decided to lean into that mini, you know, news Agentekeye rally. So I think that goes to say that we're still in a market that's more greedy than fearful. And I think that's, you know, something worth keeping in mind as we head into Q4, because I don't think the bubble is going to pop this year. I think that's still a 2027 conversation. But similar to what we talked about in August, I don't think it's just going to be, you know, one bubble pop and then that's it. I think we're going to see in Q4 some reset to the supply chain. And that's something that I'm also excited to kind of talk through in terms of the inflationary component costs that have really gone up during this year.

3:47And we see those really, I would say, rebalancing or being more accurate to end demand come Q4. Is that one of your main data points that you're going to be looking at heading into Q4? Yes. So we're definitely really watching the supply chain closely. Memory is the component that has increased in price the most. But I think the alarming thing right now is that everything has gone up in price. So even the PCB boards have gone up in price. So I think we're getting to a point where the price increases are creating end demand destruction. And that's something that has been flagged a lot more as we head further into the second half of the year, both in terms of PC TAM and smartphone TAM contracting.

4:24So the IDC now sees the smartphone TAM contracting about 17 % this year and 27 % in the second half of this year. And they're blaming memory and higher component costs on that. PC shipments are also expected to plunge close to 19 % year over year, with the worst of that really happening in the second half. So the situation that we are at today, just to put together a PC, you can no longer have a PC at the price point of$399, right? Everything has gotten, I would say, boosted up into a higher high-end product price range, and Apple will win the customer every time at that price range. So this is creating end-demand destruction both on PC and on smartphone.

5:07And so what we're watching is when does the other shoe really drop? When does the double ordering, the inventory hoarding that's happening in response to either supply chain, risk anticipation, or fear that there's going to be further price increases, what does that cool off? Because the only thing really working in this market is AI, right? We know from everything else that everything else is pretty weak elsewhere. So it puts a lot of pressure on AI, but it also means that at the end of the day, the supply chain, supply demand dynamics do, at the end of the day, find a balance. And so we're looking for when that balance happens, when we need to step out of those trades that have been puffed up by these price increases.

5:46And I'll just add one more thing here. Something that's interesting to think about as we talk about shortage. Yeah, there may be a shortage in the realm of AI in terms of what we need. And we're still as an industry gauging how much is needed. And so while that is true, I don't think this is a broader overarching shortage. Because if you remember back in COVID, that was a real shortage. Right. People miss numbers. Dell miss numbers. GM miss numbers. Ford miss numbers because they actually could not get the components to build the product and get it off the shelves. Today, no one's missing numbers.

6:18We're actually seeing guidance raised. But this is all driven by higher selling prices rather than more unit volume being sold. So what's scary about that is that you've got to pay attention to when that higher price could come down. Right. Because it's inflationary. So that's where we're watching the semi-supply chain to kind of pinpoint when that reset is going to happen. And right now we're watching it for Q4, especially as far as the analog guys go, because those saw a huge amount of pull in earlier this year. That was part of our bullish thesis on the analog sector. And now I think it's kind of peaking out there.

6:50So, yeah, we're definitely watching that really closely. And, you know, taking it back just more broadly to that, you know, AI slowdown and then the recovery from there. I think the other interesting thing to throw in from the supply chain perspective and the semi perspective is basically how different the reaction is from Jensen Huang and Hook Tan versus Sam Altman and Dario. At the end of the day, these guys have been in cahoots, right? You've got Jensen and Hook. They're really backing and dropping an opening eye, whether that's through financial guarantees and backstops. They're funding these guys.

7:26And they made a point on both of their earnings calls, which happened since we last spoke, to really pronounce that, right? Jensen got asked on the earnings call about, you know, quote unquote, circular financing. And he was like, our only regret is that we didn't invest more and sooner in OpenAI and Anthropik. So for him to come out a couple of weeks later and dismiss the slowdown concerns as, you know, these guys are young. They're still figuring it out. They have no idea yet what they're doing. It's just there's becoming, I would say, a certain extent of mismatch between those sides. I don't know if you noticed that in specific.

7:56No, it's interesting. We've also been talking about kind of the difference in management styles in the tech space in some recent episodes. Anything you would add to that picture specifically on the management side as we're talking about that point? I would say, and this is my personal take, I think that when we look at, I think there's a world of difference between the Semi guys and their management style versus, I would say, the Frontier Labs. You know, when you think about the steps that Jensen Huang makes, you know, he goes out to Taiwan, he goes to the supply chain. He is very much embedded in the supply chain.

8:32He talks to Taiwanese ODMs. He's also very much connected to the NeoCloud customer group. So there's a lot of networking to do with this because at the end of the day, you want to secure the supply chain. And this is, you know, Jensen is the poster of that, but we see this happen from management across the supply chain in the semiconductor space. I don't think we see the equivalent of that on, you know, the U.S. AI lab front. So it's much more, I would say, through execution. And at the end of the day, you know, these companies have been here for a very long time on the semi front. And I would agree to an extent with Jensen that, you know, Anthropic and OpenAI are very new.

9:09They're babies, essentially, in the industry. And they are being treated by babies in the sense of how much funding is going towards them and how much they're going through trial and error. Even the fact that NVIDIA bought Hugging Face, which was the one that got hacked by OpenAI, it really also just extends this idea of, hey, I'm going to just buy out the sector, fund the customers, and make sure that everything is tight and neat because that's the kind of management style. Jensen's not only managing NVIDIA at this point. He's really managing the supply chain in the broader AI trade, I would say.

9:40So I think, yeah, I think there's definitely, it's interesting to think about management style as well on that front. We had Alex King on from Growth Investor Pro. And to your point about the semis, I guess at the end of August, he was on saying don't sleep on software. That was the headline on that episode. And the episode this week was don't sleep on analog semiconductors. What's your thought about that differentiation, A? And what else would you add to the picture of the kind of subsectors? I would say, so I actually saw that headline. I have it on my watch list, but I still didn't get the chance to listen to the episode.

10:11But I would say in terms of software, I couldn't agree more with him. I think we're seeing a lot of rotation into software, specifically into software more broadly in Q3, but also into the software names, the cybersecurity sector in specific that could benefit from all of this AI safety concern theme playing out. At the same time, when it goes to analog and thinking about the subsectors under semis, I think analog, there was a great play there. I wouldn't agree that right now analog is going to have this breakout moment. I think, you know, when we look at the analog peer group, analog guys, essentially, I would say they're at higher risk of seeing double orders and pull-ins because there's not as much risk to hoarding the inventory that's within the analog sector.

10:53You know, unless we're talking about data center or enterprise data kind of segments, that's where I think about monolithic power, which I think is still an attractive name as far as, you know, the AI famous position, although they had a great quarter last quarter. So we're thinking about them more as a trade right now because we think it's going to be hard to follow that up into next quarter. But more broadly, I think when we think about the analog peer group, they are exposed to a couple of different segments. So, you know, I think one of the tickers that was mentioned in that episode was TXN, if I'm not mistaken, right?

11:22That is, you are absolutely correct. Texas Instruments. And Texas Instruments is great, right? They're a huge company. They're a great company. But at the same time, they have exposure to automotive, they have exposure to personal electronics, to data center, to communications. They're pretty, I would say, diversified, which is a nice thing if we were in January and they were going to benefit from double ordering or pulling across these sectors. But now I think it could work against them because end demand across these sectors is pretty weak. So against automotive and personal electronics in a lot of these segments, end demand is really waning.

11:55and that's because of demand destruction that's happening from higher component costs. So I think that as we head further into the end of the year and early next year, I think specifically early next year, analog is at a higher risk profile because a lot of that double ordering has caused them to report above typical seasonality. And so expectations have also ran up from there, kind of treating this analog sector as an AI trade because a lot of these guys have exposure to data center segment. But the data center segment is too small a part of total sales to move the needle substantially if the double ordering from everywhere else falls through.

12:31So that's where we see the risk with analog. The story, I think, is more attached to that double ordering. And we saw this with TXN last year when we had Trump's Liberation Day back in October, not October, apologies, back in April. And then the quarter after that, TXN and analog group all reported really great numbers because there was so much anticipation of supply chain risk that everyone double ordered and hoarded it. And that lasted for, I think, two quarters that they reported above seasonal results. And then it kind of fell through from there. So that's what we see happening again this time around.

13:06The timing is a bit more difficult to pinpoint because you also have that wild card of data center that's in the mix. But I would say early next year is when we really see a higher risk on the analog peer group. And then, if I may, just taking it back to the semi-divide with U.S. frontier labs, I think the really interesting thing to point out about Jensen and Tan is that essentially both of them on their earnings calls earlier this year or earlier this quarter actually guided out a year. And that's never happened before. We've never had an annual outlook from NVIDIA or from Broadcom. And so it's interesting to put that into context of these two companies endorse the U.S.

13:49frontiers. They guide a year out, which we attribute to them trying to convince Wall Street that, hey, you know, we're guiding a year out. We have more to show. We have more growth coming. It's a very, you know, supply constrained environment. And at least, you know, NVIDIA said, at least we're going to grow 70 percent next year. And then Broadcom guiding two years out, right, and telling us that Anthropik is going to be their largest customer next year, followed by OpenAI. And then, you know, the U.S. frontier is talking about a slowdown. And then the semis guys kind of just dismissing that entire conversation.

14:20So I think when we put this together, big picture, everyone is really trying to serve themselves. So in the sense that the way that we're reading it is the frontier labs are saying, hey, if, if, you know, they want to, they want to blame someone, something if, or someone, if things go south. And the easiest, you know, easiest person to say, hey, keep us in check is the government. We know that that's, you know, at least from our perspective in our research, that's not, you know, it's not that they grow conscious overnight again. At the end of the day, we had the Trump administration try to intervene with, you know, Fable 5 and with Mythos earlier this year.

14:54And even with OpenAI's own models, I think it was 5.6 that they tried to intervene with. And I think the Trump administration has been intervening beyond, you know, the expertise in AI that they actually have. So they've been, you know, they've been putting their hand in the AI scene for sure. So it's funny to see these guys who push back against that kind of intervention come back and re-invite it. So from our perspective, these guys are kind of more looking for someone to blame if things go south and also an excuse to push out their IPOs because increasingly we're learning that their financials are really, really messy.

15:26Right. So now OpenAI is officially not going public this year. They pushed it to next year. And Atherapic pushed it from October to November as far as their IPOs goes. And so, you know, when we put this all together, yeah, the big picture is becoming more concerning. But it goes back to this idea that it just took Muse and Agenda Gay Eye headlines for the market to lean into the greed again. So that kind of push and pull between greed and fear that I think we talked about a couple podcasts back is still very much alive. What else have you been taking away from the earnings that you've seen and or heard?

15:57And what are you looking forward to in the upcoming earnings? So I think the earnings that we're looking forward to the most is the one that's coming up next week, which is Micron. And so we have Micron reporting on September 30th, and they're going to really kick off Q4. They're the first to report. And there's a lot of pressure on Micron into print because we've had a switch, I would say, in the memory sentiment intra-quarter. In Q2, everyone was obsessed with the memory trade. In Q3, we had a lot of excitement behind the memory trade, but I think some of the data points that started coming out began to be actually priced in by the market and recognized.

16:33So one of these data points, for example, is we had TrendForce come out and talk about actually, you know, a deceleration or cooling in the acceleration of ASP for DRAM, right? So TrendForce is now talking about DRAM ASP only going up 13 to 18 percent in Q3 versus I think it's closer to 50 to 60 percent in Q2. And so we're seeing, you know, we're seeing these data points come out. We're seeing more Chinese supply come on. And that's something that we kind of talked about in a lot of detail last episode. but just to kind of drive home how much Chinese supply is coming home, we have CXMT, which is, you know, the main Chinese DRAM guy who's entering the market, expected to end the year within 20 ,000 wafer starts range as Micron's own monthly output.

17:20So they're coming in with, you know, they're coming in aggressively with a lot of supply. And so since, you know, the demand supply dynamics are what resulted in this shortage, right? Too much demand from memory, not enough supply, so prices go up. If we get more supply on the table, the logic would only say that you're going to have pricing cool off a bit. And at the same time, aside from supply coming on, you also have this weak end demand that we're talking about outside of the realm of AI. So PCs weak, smartphones weak. So you don't really have end demand to drive. You don't have strong end demand to help the demand side of that equation.

17:56So in our opinion, we could go very quickly from a shortage to a glut, and that's where the risk becomes. Now, looking at Micron specifically for next week, what we're most interested about is really the gross margins. That's what we're worried about. What will Micron's gross margin guide be? Because last quarter, they guided for 86 % gross margins after printing 84.6. And so if we're in the 80s range, and that means that Micron right now has the highest gross margins in the entire sector, even higher than NVIDIA's own peak margins, which is crazy to think, right? Memory is still cyclical. It's crazy to think that, you know, margins are just going to go up endlessly from here.

18:34And so our concern is about what the outlook is going to look like, what the guide is going to look like for margins. And we don't think the market is going to be very forgiving if Micron's gross margins aren't guided to expand again into the first quarter. Any other companies that you would add to the conversation? Yeah. So I think as far as earnings go, the first couple of companies, I think they're good kind of indicators. We have ASML is going to follow and TSM. I think the two of those don't give a huge gauge for what's happening in the near term because they're further out indicators with either the foundry side with TSM or lithography side with ASML.

19:15But what we're interested more than, and we're bullish on both of those names, what we're interested more to see is how market is going to react to Q4 earnings. So if you recall in Q3, when we started off, we had Micron, market reacted positively, stock stored, and then a day later, it began to trade downward for a while from there. And then we had ASML and TSM, and while both of them reported great results, both sold off after print. And that sentiment of, you know, earnings acting more as a test than a catalyst played out through the rest of the quarter because expectations have gotten so high after Q2.

19:49So what we're interested in, what we're watching as we kick off this quarter is how is the market looking at it? Are earnings going to be treated as a test or as a catalyst this quarter? Are they going to be looked at positively against lower expectations after the Q3 reset? or are we going to see it kind of differentiate now depending on the sector and that broader AI theme? Is it going to do that heavy lifting that it once did? So I think that's what we're looking at more broadly. And a big part of how we're looking at this specifically when we think about, you know, the hyperscaler side of the conversation is what we're going to learn in terms of CapEx, specifically in terms of commentary for next year.

20:28Specifically, as we take that CapEx spend into context of what's happened this year, as far as, you know, AI becoming a lot more cost conscious with, you know, the penetration of cheaper Chinese alternatives. So, you know, when you look at it from that point and you take it into context of everything that's happening, I think a big part of why we're seeing, you know, this quote unquote slowdown, I know I keep going back there, but Anthropoc and OpenAI are too big of this not to come back there. You know, the two of them across each other, I think they have, you know, north of 1.2 trillion commitments across the semi-space through 2030.

21:02So they really are at the core of this is the idea about pricing and pressure from these Chinese alternatives. So since that slowdown conversation two weeks ago, Anthropic released cheaper models, Opus 5.5, which is 40 % less to run on typical workloads than Opus 5. And OpenAI actually released two cheaper models and that basically slashed the API developer pricing by 50%. Even Microsoft is announcing tiered volume discounts, 30 to 50 % for large enterprise customers and smaller businesses. So we're seeing a lot of changes as far as pricing goes. And I'm excited to see how that relates in terms of the CapEx conversations that we're going to be listening to.

21:41Because we know that the, you know, OpenAI is struggling internally from its financials. We know that there's some fishy business happening in its realm. And I can expand on why I say those two statements if you'd like. Absolutely. All right. Yeah. So just briefly, you know, OpenAI, I think part of that push out, part of why we're seeing, the kind of conversations that we're seeing today from Sam Altman is because OpenAI's revenue to cash burn ratio is actually getting worse rather than better. So right now, they are projecting to burn almost $280 billion by 2030. So spending is estimated to be $856 billion on compute and infrastructure, while revenue is projected to grow tenfold from$36 billion to$350 billion.

22:22It still can't keep up with the amount of spend. And so this is becoming an increasingly big conversation. Are we going to be able, you know, it's not just about when are we going to see the return on the investment, but it's about how long is this going to take, right? Because it looks like increasingly matters are getting worse before they're going to get better. And I think that's something that Wall Street's now wrapping its head around alongside this idea that why we need so much circular financing. And so, you know, that's on OpenAI. And if you look at Anthropik, Anthropik, you know, part of their justification for pushing out the IPO is that they want to have the Q3 results out because, you know, they're claiming that they're going to be profitable for the second quarter straight.

23:00But they're getting creative, I would say, in how they're defining profitability because their profitability excludes Amazon's cuts, excludes stock-based compensation, and it excludes training costs. So how do we think about that in terms of, you know, if you're profitable once you take out your biggest costs, are you really profitable? And so if you take that alongside the news we got from Oracle yesterday about the widening CDS spread, all the hyperscaler debt, and Apollo warning about riskier hyperscaler debt. I think all of this big picture, again, it becomes this idea about how can we keep spending if we don't know that there's this return that's coming when we have a case about parallel in China where we're seeing cheaper alternatives, gain share and gain use, and not need to spend as much.

23:50And the numbers are really proving it. So when we think about, you know, the events of this entire month, we follow the money. And so when you follow the money, you see, okay, Anthropik and OpenAI and Microsoft are trying to slash pricing. You see, you know, the Kappa's conversation, we're going to get some color on 2027, but expectations are for even higher spend. And at the same time, the entire trade is getting riskier in terms of needing to spend more before actually getting any revenue recognition, at least for the two main center stage guys in this AI trade, which are Anthropik and OpenAI.