Tiernan Ray Talks 2Q Earnings, Tech Stock Volatility & Financing AI Buildout

Welcome to Market Matters. I'm Marly Caden. Each week we're stripping away the noise to focus on the big picture economic themes driving the market. Today we are tracking the heart of tech, a raging bull market in an AI infrastructure that's battling deep disruptions in enterprise software, crowded chip markets, and a reevaluation of tech's most speculative frontiers. While benchmarks like the Nasdaq are up double digits this year driven by massive strength in semiconductor capital equipment and memory, we're also seeing legacy software stocks get slammed 20% or more on unpredictable AI costs, recent IPOs like SpaceX plummeting below offer price, and quantum computing names facing a brutal reality check. With memory pricing skyrocketing, custom silicon sparking a new CPU arms race, and hyperscalers touting hundred billion-dollar budgets, the market is forcing investors to separate long-term cycle winners from accidental passengers. >> All of the spending that they are promising, uh, and I sit here saying I believe in these companies, I know their products are excellent, I know their management's excellent. Um, how sustainable are the promises, how real are the promises of spending to which their fortunes are pegged? >> Joining us today is Tiernan Ray, veteran tech journalist, editor, and publisher of the Technology Letter, and senior contributing writer at ZDNet. Together we'll dive deep into the supply-demand dynamics of the AI build-out, why software budgets are getting eaten alive, and where the real fundamental conviction lies in tech today. That conversation starts right now. Tiernan, thanks so much for being with us today. Now, you have a very interesting background. You spent more than three decades covering everything from uh the dot-com boom to the smartphone revolution to cloud computing, now generative AI. As you have now been a part of and tracked multiple tech cycles, what are you seeing in this current AI cycle that feels unprecedented versus what elements seem like classic tech cycle deja vu for you? >> We are Marley, first of all, thank you for having me. It's a pleasure to be here. We are, I like to say, in the midst of a renaissance in computer chips, which is at least going on 15 years now. It goes back to 2010 or so. Um, and what that means is that chip technology had for a long time um kind of been stale because you had Intel, the Wintel duopoly, you had PCs. Um, and you got a little bit of innovation when Apple started making its own chips for its phones, for the iPhone. Um, that was a big deal uh starting with the first iPhones in 2007. But, what we're seeing now is just a kind of a Cambrian explosion of these chips. And that's the thing that um is a dramatic multi-decade change. And it's showing up in lots of places. And the thing that's sort of most unprecedented now is these companies like Micron Technology that make the DRAM had sort of been pretty predictable producers of a commodity product for decades. Uh, you always had more memory in your PC. Um, but now they're being flooded with requests for chips because everyone's building these data centers that are so controversial for more and more AI. And you're now seeing the stock valuations of Micron Technology, SanDisk is another one, Samsung Electronics, the biggest memory chip maker, and SK Hynix is the South Korean company that um is number two behind Samsung that just introduced a sponsored American Depository Receipt, ADRs, right? So, you could if you're an American uh investor, you can trade directly on Nasdaq these shares. And so, this is this huge um development where these companies that were reliable but low-valued like Micron are now suddenly superstars cuz everyone needs more memory. And that's the thing within this whole chip renaissance that's kind of most startling to me. >> We have so much to dive into in terms of the specific companies and memories and chips. But before we get to all of that, I want to talk about what got you here. I mean, you spent a decade covering tech over at Barron's. Now you have the technology letter, which is your own publication. What made you want to go from being a working journalist for a publication to having a publication of your own? >> We, Marlee, you and I had not a lot to do during 2020, if you recall. There were lockdowns, and so we all kind of like took up an eight musical instrument or a hobby of some kind to fill the time. And I had some extra time between filing articles from the home office under lockdown. I said, "Okay, why don't I try and do something because there are people who are trying to put together new styles of journalism where they're like an aggregator." And I talked with them, and Substack was growing in popularity. And I said to myself, "You know, I think I could do a Substack just on my own." And I've done this for decades at Barron's when I was there for 13 years starting in 2005. I was, you know, filing sometimes 15 blog posts a day. So, just churning out copy. So, I said, "I I could do this on my own." And what I noticed, Marlee, was that people would talk to me about investing in tech and individual investors. And they were not clear as to quite what was out there in terms of stocks. And they were sort of striking out in the dark trying to come up with guesses. And to me, it was an issue that the mainstream media, including Barron's, including Financial Times, Wall Street Journal, um, all the major publications that have some kind of investing thing had abandoned what was traditional kind of enterprise reporting. Um, and what I mean Marley is, you know, the Wall Street Journal used to dive into companies that were merely 20 billion in value and tell you about what do they do, you know, and it'd be how you would learn as an individual investor. And they stopped doing that. And the reason they stopped doing that, of course, is because the biggest companies, Microsoft Meta Alphabet Amazon Tesla had become the entire focus. The entire focus was on seven or eight companies. And so I said, okay, you know, there's probably, not only can I do this stuff cuz I did it, but um, and I have extra time on my hands cuz it's the pandemic, but people need to know about, you know, something that they can invest in that is like below a trillion dollars in market capitalization that might actually be a terrific investment cuz they're terrific companies. And there's dozens of them. There's stuff to write about all the time. So it was really, you know, a combination of I could do this on my own and we all as journalists, I think we have an eye to where is the industry going? Are there going to be jobs in reporting? And then also there's a massive kind of absence of the kinds of reporting that investors need about other kinds of companies. >> And it's, you know, one of the things I really like that sort of distinguishes your letter from other newsletter writers or other market commentators or other people on Substack, you make a point of not owning or trading the individual stocks that you cover. You've maintained your journalistic independence. So, how does that then help you analyze balance sheets and capital cycles or inform your reporting compared to consensus Wall Street commentary because I know I look at it through a similar lens, but you're arguably sort of standing over the line right now. You've got one foot on each side. >> They are always uh long long only investors, Marli, who believe in this company, and short sellers who found the you know, the thing that's going to sink this stock. They are always fans of a certain angle, and I think that's a trap. I think if you're going to be sort of logical about the investment, you should be able to say on any given day, "Hm, this doesn't work." Or, "Hm, maybe this company's a wonderful company. Maybe they're simply not going to attract investment." Or conversely, you know, I had a whole notion about this particular company and its stock, and now I'm wrong. And you should be able to have a certain detachment, and that came out of you know, being in journalism for 32 years, cuz when you're at Dow Jones, which produces Barron's, uh and before that at Smart Money Magazine, you always have an actual institutional rule. You actually have to report on this. Uh News Corp, Rupert Murdoch's News Corp bought Dow Jones, right? Um in 2007, and so there were instituted very specific rules, and you would be audited to make sure you don't own the things you're reporting on it, cuz it's a conflict of interest. And so I had just adopted that from my professional background, and it makes sense to me. I It feels like you should be intellectually detached from worrying every single day about your own portfolio, so that you can have some kind of um way to not be a promoter or a fan, and not be defensive about what's the story for this company, but actually looking at you know, what are the relative strengths and weaknesses. >> And so was that a part of the creation of the technology letter? Because I have read quite a few of them. You very much focus on the fundamentals. You separate out all of the hype and the story or the narratives that people are trying to drive that that often emotionally tied to it being a holding for them, and you really just look to fundamental tech cycles and what certain things in those fundamental tech cycles might indicate. How important was that to you when you were creating the technology letter? >> I love the technology, Mårten, and I am a student of history, and I love how the technology changes, and I walked backwards into investment analysis because I started out fascinated. In my early days of reporting, I was covering private companies, and there was no investment angle up front. You were really writing about the tech. And when I went to Smart Money, and then I went to Bloomberg, and then I went to Barron's, I had to really learn to think like a fund manager. What is the cash flow? What is the stock's valuation? So, I I came backwards into the investment analysis from just loving the technology. And so, every single time I look at these companies, what's first and foremost in my mind is what is the story in broad strokes that is going on to the earlier point about chip technology. I'm fascinated by semiconductors. There's so much information. You really can understand what they're doing, and you can see patterns over decades that evolve. Such as, you know, suddenly there's a break, right, in the memory chip makers like Micron. There's something new. And I love to see those things develop, and I think there's an important vantage where people who only know the value of a stock or the price of a stock, people who are traders or fund managers, um they can be very astute, but they're sometimes missing what what is a profound change, a historical change, that's going on in a market such as we're seeing with your data centers, memory chips, the semiconductor proliferation. Some of the fund managers manage to do both, um, but a lot of times if you just look at what's the price to earnings ratio of a stock, um, you're trapped in all you can see is a is some of those financial data points and you're missing the larger story. So, I usually come at it from, you know, a a background of having first been fascinated and love with the technology and then how does that work within day-to-day? Are you going to put money into these stocks? >> All right, so then let's look at that through the lens of this earning season. I mean, we saw tech results that have been met with extreme market reactions. As I just look broadly across the entire tech landscape right now, how would you characterize market sentiment here? I mean, you mentioned there sometimes people missing historical changes. Are you looking at the reactions we're seeing as healthy consolidation? Is this the beginning of a broader regime shift or is it something else? >> There's two things. You're right, Marlie. There is a regime shift. Um, and there's also uncertainty about the financing. So, the first part is semiconductors, to my earlier point, chip, uh, development is a profound change, the proliferation of who can make chips. Google is one of the largest chip makers in the world, um, right? It was an advertising web company, search company, and now it's one of the largest producers of chips, the custom TPU chip. And this is proliferating. Anthropic's making its own chip. Open AI's making its own chip. So, this proliferation of chips is a profound historical shift where there are more There's more of everything. There's more diversity in the chip market and that's just going to increase, uh, uh, over time because the whole infrastructure with Taiwan Semiconductor to produce chips for everyone who wants to design a chip is happening. You and I will produce our custom AI chips at some point. So, that's something that is going going over decades that is the kind of thing you can plug into and say, "Yeah, that's that's real." Um and it persists through ups and downs of market cycles. At the same time, there's this thing that you and I have been writing about and seeing the headlines for a long time now, which is um Nvidia's throwing money at everyone who's one of its customers. They're offering backstops to fund data center development. Um Elon Musk is uh projecting a trillion dollars in revenue in 2030, annual revenue for SpaceX. And to do it, he is uh increasing the outlook for spending, for capital investment. So, there's tons of money coming in and I sort of sit there, you know, as a person who's part of reporter, part of stock analyst, struggling with I know the chip trend that I spoke about of proliferation is a reliable decades-long evolution of the technology. You can just see it and you can count on it and it comes intersecting with this massive spending that like past uh eras, like the railroad era in the 1800s in America, um like tulip mania, right? Um will eventually reach some point where the the financing cannot hold. Some parties will back out. Some promised spending will not be delivered, will not be realized. Um people will contract their budgets and and that's the whole part that is kind of a cloud of uncertainty um that is intermixed with this AI thing that, you know, I'm not an economist. I try to understand where financing comes and goes and it's it adds a huge element of uncertainty to the first part that is reliable. >> All right. So, a lot to unpack there. We are going to get to SpaceX. We are going to talk about some of the things Elon Musk has said in recent days. Uh but I want to talk about chips really quick because you mentioned it feels like everyone is making their own chip. Everyone's making an in-house chip. We've got custom AI chips. There was a point in time where when we were talking about CPUs, it was basically AMD versus Intel. And now it feels like that is a very crowded space. Chips are becoming more crowded. You mentioned Nvidia tossing money at every customer it can. How crowded is the space as you look at it because I know you also study the trends that you're seeing. And are the gains, the massive gains that we've seen from some of these companies sustainable with how much competition there now is? >> I think there's two parts of this, Marley. I think the CPU market where there was Intel and AMD, as you said, a structure that's very stable duopoly, right? It's great to have chocolate and vanilla in a market. Somebody always wants one of the other. And by being two players, they both vendors, AMD and Intel, avoid the appearance of having a monopoly. They avoid regulatory scrutiny. It's a very stable structure. It's We have Uber and Lyft. You have two options. There's always two options in a market that is stable. Now we have multiple parties entering the CPU market. Nvidia has its own uh CPU. So every Nvidia AI chip thing is actually Grace Blackwell. Blackwell's the GPU chip that does the AI stuff. Grace is the CPU offering that kind of can replace what AMD and Intel do. Now we have Vera Rubin is the next flavor from Nvidia. Rubin is the new GPU. Vera is the new CPU. So Nvidia's throwing multiple offerings of CPUs at the market because richest company in the world, they can afford to get into that market and develop their own and try to control more of the platform. We also now have ARM Holdings, which is a company that has been the vendor of intellectual property for decades. For years, everyone would come to ARM and they would get a little bit of the design, the floor plan for a chip that would help them get a a head start, and they would pay a royalty fee to ARM. Now, ARM is going to sell, make and sell its own CPUs in competition with Intel, AMD, Nvidia. So, this is becoming a crowded market for CPUs, and that's a little bit perilous for the traditional two, AMD and Intel. It's not as clean and and simple a duopoly as it was. On the other hand, the part about AI chips, I think will sustain lots of parties, um and it's basically, Marli, it's the it's the iPhone model. Apple surprised everyone by saying, "If we sell the phone with our own chip, we can make better profit margins, and we also can make the phone do stuff that we like that we couldn't get off the shelf." So, going from off-the-shelf to bespoke turned out to be a great model, what people sometimes call vertically integrated. You know, it's one of these awful jargon terms. Build everything. We build the casing, we build we design the chip, and we build the software, the iOS. And everyone who builds a data center now, including Google and Meta and Microsoft and Amazon, just take the big the big four here, uh and you can throw in Oracle, too. They all now have deals to make their own chips. And what has partly enabled that, Marli, is that once upon a time, if you and I wanted to make a chip, we had to first have a huge market um in order to be able to guarantee volumes of production to fill the factories. Now, Taiwan Semiconductor, the biggest contract maker in the world, has figured out a scheme where you and I can go and sort of bunch in our designs with other people's designs. And so, everyone, uh down to small volume, can have a chip produced by TSMC. And Broadcom is the other one. Uh they are working with everyone to help them design. So, Broadcom is sort of the middleman. You and I make a chip design for AI. We hand it to Broadcom. Broadcom goes to TSMC and and tells TSMC, "Okay, here's the plans. Here's how to um to set it up." And TSMC runs the factory. And so, there's now a supply chain for anyone who doesn't own factories to design an AI chip. And I think that will just keep going. I think that's a a reliable trend for decades is the specialization and running kind of it's kind of you know, it's um it's whiskey that's produced as kind of a small batch, you know, cuz it has special properties. And everyone loves the custom thing like that. And there's it will finally be supported by the supply chain. >> But does everyone need their own chip? I mean, is this just everybody diving into a trend here? I mean, I on a near daily basis find myself hunting for squishies with my daughter. And it feels a little bit like chips are sort of the the adult finance version of that right now. Like get your get your own chip. Everybody wants their own chip. A small batch chip even better because it's special. At what point might I mean, you said it's reliable, but are we going to see a point where this dies off because it's not actually efficient for everyone to have their own in-house chip? >> It will be efficient for those who have access to capital. And that certainly includes the top four or five, you know, hyperscale companies, Google, Meta, Alphabet, Amazon, Microsoft, Oracle. So, if as long as they have money to spend, it's affordable for them. And actually, I don't know if it's true for other articles like scrunchies, but you can actually get benefits if you have a custom chip running your software. So, the the combination of what you want to do and and a special chip actually can provably yield efficiencies. It could be more power efficient. It can have features that you wouldn't get off the shelf. So, there's a reason to do it, and if you have cash, if you have the balance sheet, you can support it. Now, at some point um does everyone rush to do this, and they haven't actually gotten a market in which they can sell the what that chip can do? Yeah, that that is there's certainly uh potential for a huge bubble in everyone spending and spending in advance of their balance sheets um to produce volume, and you you make a good point, which is ultimately, there's a finite market for anything. So, if you and I and everyone makes chips we all kind of think we're going to have X percentage of this finite market, and probably some of us will get, you know, um maybe a couple of us will get 30% each, maybe some of us will get 10%. Some of us will get none of the market cuz we designed the chip that's not as good, and we'll be losers. And so, yeah, there can you can have a rush to do this for a finite market, and some people end up not getting any share. So, all the money they put into custom AI chips is a waste. >> And if we had had this conversation 12 months ago, chips would have been the half hour of this podcast, but we have broadened. There is a lot more in the AI trade in the tech trade right now. So, I I want to talk about your TL20. This is your your basket that sort of gives a barometer of the broader tech trade. AI winners still dramatically outperforming the broader market. Now, despite that sharp correction that we've had over the past month or so, your basket's up about 55% this year. That's outpacing the Nasdaq's 13% gain. After this rebound from the July pullback, are you looking at these AI winners as still offering as strong of a risk reward? Have the earnings that we've gotten actually strengthened the AI bull case, or did it expose where expectations have simply become impossible because I can recall at least three or four instances where I found myself wanting to scratch my head as I was looking at reactions to certain reports. >> Yeah, I think Marley, thank you first of all for the kind and gentle plug for the TLT 20. I appreciate that. Uh TLT 20's been is 4 years in running now. I started in July 2022, which was a great time to pick stocks. It's always a good time to pick stocks when the market's plunging cuz everything's cheaper. So, in that time, the TLT 20's up 500% through this week. Um and some valuations have gotten pricier and you have to say, uh okay, when is this stuff going to correct because it in April, May, June, it just went up and up and up. Now, it took a breather in July, tiny correction, right? And now it's back um to uh going up on the way up again for the TLT 20. And so, the issue becomes the the business for all of these companies, these 20 companies, which includes Micron, includes Nvidia, includes Broadcom, uh lots of of uh AI All of them are AI linked in some way. The issue becomes um their businesses are excellent. Their management is excellent. What it what their success is predicated on is the spending of uh SpaceX, Elon Musk spending hundreds of billions, uh raising debt, Coreweave, uh right? The Neo Cloud, which has gone into massive debt, Oracle, which has a huge backlog and is taking on debt after having been a fairly efficient software vendor. All of these companies that have trillion-dollar valuations that are private, Anthropic, OpenAI, all of the spending that they are promising, uh and I sit here saying, I believe in these companies. I know their products are excellent, I know their management's excellent. Um how sustainable are the promises, how real are the promises of spending to which their fortunes are pegged? And uh I cannot tell you authoritatively when some of that financing may dry up. The best sense I have is, you know, as long as it's the free cash flow of say a Google or a Microsoft that is sustaining the chip market, uh they have the latitude to do that cuz they have gold-plated balance sheets. When outside financing comes in, um private capital markets come in to finance some of the chip purchases, um then you get the real possibility that the investors are not strategic, they're financial, they're speculators, and that money is less reliable uh as on the demand side in, you know, 2027, maybe we could see some of those bets start to fall apart from the outside financing. See private capital markets uh being less robust, uh and that starts to call into question what the payoffs going to be for all these wonderful chip companies. >> And what surprised you the most about the results that we got this quarter? Or perhaps it wasn't even the results, maybe it was a reaction to some of the results. >> Uh yeah, so specifically software, Marley, was a big surprise this quarter because um I think you and I have talked about this a bit. We've had excellent companies like Twilio that I mentioned with you on air before, Datadog, uh Dynatrace, Inno data, um Digital Ocean. All these companies did pretty great this past week or so when they were reporting results. Um the thing that was interesting is we saw an even sharper separation between the AI winners in software and the AI losers. So, on Thursday, for example, HubSpot, which creates um marketing and sales tools, sort of front office software, very established multi-billion-dollar company, uh saw its stock plunge. And the CEO, Yamini Rangan, told the street that the companies she sells into are having a harder time um managing their budgets. They are not prioritizing spending with her company. They are spending on what? AI tokens. Um and so what that means is traditional sort of system of record software, such as HubSpot, such as salesforce.com, you know, the poster child, biggest of the application software vendors for customer relationship management, this thing that's been around for decades. Uh Workday, which makes general ledger software, manage your finances. All these companies are stuck in a situation where the CIO or line of business managers to whom they traditionally sold are so completely preoccupied with AI and spending so much on AI, some of their budgets are out of control, they cannot reliably, at the same cadence as they once did, upgrade, expand their software contracts with HubSpot, Workday, Salesforce, and that's really called into question the viability of those companies because their whole deal was every year they predictably got renewals of software, you know, just like you or I would might um true up our subscription to a publication, right? Or we bought more uh Photoshop. And on the other hand, these companies that you and I have talked about, DataDog, Dynatrace, Inno data, Digital Ocean, Twilio, are doing great. And it seems to be the case that what we're seeing is the infrastructure software, which is a term in the in the industry for stuff you use to make things, tools that you use to make other software. That stuff is becoming a hit in AI because as much as you and I could vibe code an app, you know, I don't I don't vibe code a lot. I don't know if you do, but um as much as we could build an app, we still need these tools to try to um, sort of guide, provide some structure to using AI to build So, that stuff's suddenly newly relevant. The other stuff, the application software, HubSpot Workday Salesforce struggling. And so, we that divide was very sharp this quarter, and that was uh I did not know it'd be quite so clear in one quarter in 2 weeks. >> I I do not vibe code. I will uh confess. I >> Not yet. Not yet. >> Not not yet. Um, it's >> [laughter] >> not my But, I can knit you a sweater. So, you know, I have I have old I have old school >> I value that more. >> I have old school skills. Uh, but let's let's talk about software because you mentioned HubSpot there. They the the predominant memory for me is this unpredictable AI cost, that language that they'll never escape using that term. But, then you mentioned Digital Ocean Dynatrace Datadog great numbers. Datadog also had uh its worst 1-day drop following those great numbers. So, Tynan, what is going to fundamentally separate the the winners, the companies that are going to maintain their pricing power versus those that are going to be forced into these downgrade cycles? >> You know, it's not a great situation for winners nor for losers. Obviously, if you're a HubSpot or Salesforce, and you can't get the attention of CIOs to to to sell your product, that's pretty bad. However, if you're Datadog or Dynatrace, you have another problem. And that problem is you've kind of got a got a tiger by the tail. And you suddenly have new found opportunity for your product to be relevant. You're selling to a lot of companies that are AI natives. These are venture-backed startups um, that have just built some new thing, and they need those tools to help with their program their AI programming. They still need traditional tools. Uh, Um, so you have that that cohort you're selling to and you're also in the case of DataDog you're selling a lot uh, to a handful of giant companies, Anthropic, OpenAI, who also need those tools. And so it's precarious for the winners because it's great to have uh, suddenly a tiger by the tail and be newly relevant, be selling a lot more and be perceived as an AI winner, but think about the fact that the AI native cohort, those are venture-backed companies. Marley, we know most of those companies go away over time. Most AI startup most I'm sorry, venture-backed startups fail. Right? That's the rule of the of the venture game. So some of your customers will just cease to exist or there will be radically downsizing their budget maybe in a year's time. We saw this in 2023 when Silicon Valley Bank collapsed and a lot of the startups in the valley were told by their investors, "You got to stop spending so much. You got to get real about budgeting because the market doesn't look as good for you." So AI native cohort is always vulnerable. It's always a customer risk for a company like DataDog. Uh, on the other hand, the giant companies uh, these giant buyers like Anthropic and OpenAI, well, it's nice to have their business, uh, you also have a huge exposure to how much are they going to spend if they make up 40% of your revenue per quarter, that's also a huge risk. So it's great to be a winner, but it brings all kinds of uh, challenges for uh, the successful company to keep um, a uh, budgeting intelligently, keep hiring intelligently based on an expectation of the market that could be wrong if suddenly the smaller the AI native customers flame out and you lose a whole bunch of customers, if a giant customer like Anthropic suddenly decides to cut their spending in half and that could be a a huge chunk of your revenue. So, it's it's just it's a really rough thing for software, whether you're a flat-out loser like HubSpot was on Thursday, or you're a big winner like Datadog, and even when you deliver great numbers, the investors are saying, "I don't know. Will it be okay again next quarter?" And we don't know. >> All right. So, you gave me a perfect transition there. I appreciate you. It's great to be a winner, precarious to be a winner, which makes me think of memory, specifically SanDisk getting some pretty significant punishment over what seems to be a hang-up on this 2/3 higher pricing, 1/3 higher volume setup in terms of their sequential revenue growth. How are you looking at memory right now and the sustainability of this significant move to the upside that we've seen in these memory manufacturers? >> Yeah, Marley, the the plunge in SanDisk this week is doesn't matter over the long term because what happened was they have contracts to lock in pricing, and what that means is they will leave some money on the table. You know, it's like you do a you go and enter into a contract, and you agree to take a certain amount of money, it gives you stability, but then if the price is going to go up in the market, you're going to leave some money on the table. So, that's okay because in a commodity business like NAND flash, which is what SanDisk makes, you would rather leave the table if you know that you're going to lock in some business, especially with they're receiving in some cases huge upfront payments by their customers saying, "We want supply. We We're concerned about securing supply, so we're going to pay you in advance, years in advance." So, this week is, you know, not a to me a particularly significant thing. The issue for SanDisk, and also for Micron Technology, which makes both DRAM and NAND, also for Hynix and Samsung, the two giant uh biggest companies in memory chips, the issue is in a supply and demand situation, Marley, we know what demand is because we hear about Elon Musk and Anthropic and Google and everyone talking about how much they're going to buy. Every day they tell us how much more money, how much more capital they're going to put into building data centers. So, we know we have a picture of demand. We can wonder about how much of it will come true, as they said. But, we see a picture of demand that's pretty clear. What we don't know about is supply. You can have an oversupply situation fast in memory chips. Micron and SanDisk can simply build too much. And when they do, then you get a situation where the customers who signed contracts might say, "It might be worth it to us to simply walk away from the contract we signed with Micron and forfeit the money or get sued if we can buy it cheaper cuz now prices are coming down cuz there's too much supply." So, the supply situation is always the issue for memory chips in this commodity market. And Marley, unlike OPEC, unlike the oil market where there's pretty good information to make the market relatively transparent about how much is in the ground, how much is being brought up, how much is being refined, total global production capacity and supply of oil, of crude, in the case of memory chips, we don't know. It's not a transparent market. Inside of the four biggest companies, SanDisk, Micron, Samsung, Hynix, they know what they can produce today, tomorrow, next year, two years out. They probably know what each of their competitors produce cuz they have that kind of intelligence. We on the outside have to guess. And so, we just don't know when they're going to build too much. The good news is my conversations with various parties in this industry suggest to me that um to try to ramp up production will take several years. And so, it's going to be a long time before they produce too much. Uh but again, I don't I'm not inside these companies. It's not a transparent market. I kind of have to believe that that's the case. Um but oversupply building typically building too much uh running factories too hot and building new factories too fast is the biggest issue for all four of these memory companies. >> And let's talk about Hynix here because its ADRs are down double digits since making the Nasdaq debut in July against a broader surge in the space. So, what is driving that divergence, Tirnan? Is this a company-specific issue or is this a warning flare that the broader sector should be paying attention to? >> I think that it is investors anticipating that oversupply is right around the corner and being concerned about that happening. I don't think that's the case. You know, as I was just telling you, I don't think that oversupply is right around the corner. I think it takes a lot longer. But, um since it's not a transparent market, kind of who knows? And so, um for the second biggest memory company company in the world, SK Hynix, second behind Samsung, uh people are worried they're going to quickly um use proceeds of their of their equity offerings to ramp up production and will be in an oversupply situation in you know, 6 months uh or that we should start discounting the stocks now as if we will be. And uh I don't believe that that's the case. But, that is the way stock market works. It starts to try to discount what might happen in 6 months, 12 months, 18 months. >> And you mentioned Elon Musk in your previous answer. And one of the things he said on the SpaceX call really stood out to me. I'm sure it probably perked your ears, too. He said DRAM demand is growing 200% annually while supply grows just 20%. Now, we've already discussed that we don't have any insider knowledge on how accurate those figures are, but as you look at it, is this a a realistic framework for investors to latch on to? Is this the right way to think about the memory bottleneck or is this a severe oversimplification? >> I think Elon is a great salesman right for a point of view at any point in time. I don't think he's meant to be necessarily held to the letter of it. He's probably right about the basic asymmetry. You know, again, it takes time to produce factories to produce chips. So, DRAM capacity grows uh not quite as fast as demand is growing. The 200% I don't know that that's an accurate figure, but the basic asymmetry is right that demand has shot up and production's trying to catch up. They're they're selling everything they can make at Micron when they make DRAM and NAND and same with SanDisk. They can't meet current demand. Um but within that again, you have to wonder both what is the stability of the demand that we keep hearing is increasing based on the financing for it. Is it reliable or does it at some point uh does some do some parts of that go away because the buyers suddenly have capital issues? And on the other side, what is the the nature of supply which again we don't know in DRAM and NAND. Does it um quickly ramp up? I think it takes a while. Um but because it's not a transparent market, you and I are kind of guessing about how fast let's say Elon's 20% production per annum increase in DRAM is right. How quickly does it get to 40% production? 50%? We don't know. >> And so I say Elon Musk and we have to talk about SpaceX. I we can't move beyond him without a conversation about what we've seen from SpaceX. We got their first earnings report. We saw, I think, close to a 15% drop-off on those numbers. It's double digits well off of its opening IPO price, but we also have this Elon Musk halo effect where a lot of people seem to be hanging in there because he has proved to be an exceptional innovator in other spaces. How were you looking at SpaceX and the reaction to the report and the price action we've seen since it went public? >> Did investors just realize they got way out over their skis on this one? Are we looking at a valuation reset? Does it have something to do with all of these shares in the lockup expiry that came into the open market? >> Yeah. It's definitely, at the moment, morally, you're right. It's the It's the last of those items. The lockup expiration has taken effect this week. It takes effect in stages. So, all the people who bought into the company before it went public get a chance, suddenly, to go out and sell some portion of their shares. And this happens in stages through the rest of the year. Elon Musk, as I understand it, who owns the majority of these holdings, has to wait until middle next year. The point is people can start to cash out, and that always has a depressive effect on a stock. And I think that is the main issue for the shares at the moment. It's just it's predictable. Some people going to cash out. There'll be more supply shares. And so, you can wait and watch the price go lower and lower. The longer-term issue for this company is, um, we know this week that spending expectations by SpaceX are going to go way up. That is what investors think that this company will spend will go up. We don't know from Elon and team how much because they're not specific about it. They're not making forecasts for their spending, uh, out years in the future. So, investors just have to guess. But the things that they say are qualitatively concerning. Uh Elon Musk said next year his company SpaceX will suddenly be the fourth large national national nationwide cellular operator in the US in competition with T-Mobile, Verizon, AT&T. Well, what does that mean? Some investors assume, oh okay, he's going to start to put tens of billions into acquiring spectrum leases from the government cuz that's what you do. You go to the FCC, go into an auction, and you get, you know, electromagnetic spectrum to operate a cellular network. Then he's going to have to build out towers. And all that costs tremendous amount of money. Musk on the call sort of tried to avert that concern saying, oh well, you know, we have ground stations, we have dishes for um Starlink, and we'll just put up something in those same spaces, and it'll cost less. It's all very vague, and it's all very uncertain. And this week what it meant to uh investors is more capital intensity for this business, but not a clear trajectory, not a clear final dollar amount for what that could be. It's just they have the option, and they'll keep spending. And so, you sort of look at that as an investment, you say, well, I I don't know what the cash flows of this company look like a year from now, a year and a half from now. It's just I can't predict what the spending is. It seems to be whatever Elon decides tomorrow might be a good vector to spend on. >> We have come to the point of the podcast where we do this thing called the sprint. It's basically rapid-fire questions. Whatever comes to mind when I ask you the question can be your answer. Can be one word, a sentence, whatever you want to say. Um but I'm just going to fire away at you. Five-year winner, Nvidia or Broadcom? >> Broadcom. >> Name one AI stock that has gotten too expensive. >> Wow. Uh ooh Palantir. >> One that is still underappreciated by Wall Street. >> BE Semiconductor. >> Okay. Uh, 12-month horizon on this one, software or semiconductors? >> Oh, semiconductors without a doubt. >> Memory or networking? >> horizon. I'm sorry, what was the next one? >> Memory or networking? >> Memory. >> Are we in the early, middle, or late innings of the AI infrastructure buildout? >> Oh my god, we're in the middle, the early to middle, and we're in the middle in the sense that we're in a muddle. >> In the muddle, I like that. I may borrow that. I'll give I'll give you credit though if I borrow it. >> in the muddle. >> We're stuck in the muddle a little, yeah. Most misunderstood AI stock right now? >> Mm, most misunderstood is still Nvidia. >> Why? >> Because because it's hiding in plain sight, and the company's the company's understanding of the technology and of its market tends to be glossed over simply because it's kind of a no-brainer stock, and it it is probably well owned in some portfolios. So, it's kind of hiding in plain sight. >> All right. Biggest surprise from this earning season? >> Uh, that software again has had such a shakeout between the haves and have-nots. >> And the company you're most excited to hear from next quarter? >> Oh my god, Micron is always the company I'm most excited to hear from every single day. >> And I turn and you're a New Yorker, right? >> Mhm. >> What was it like to go to school in New Jersey? >> Oh, it was hell. It was terrible. It was like a four-year sentence in a a minimum security recreational facility, I like to say. >> I hope you can feel my New Yorker in that question, cuz clearly there was something motivating it. >> Oh my god, it's just you go from a dense urban environment to a place that's relatively suburban. Um and there's a lot of Porsches and BMWs, there's a lot of money. Um it's just, you know, and things close at 6:00 and the most exciting thing in in Princeton uh town itself is the sweater shop. It's just It's not It For a city kid, it was really uh a culture shock and really hard. And so um I kind of came home a lot of weekends on the train just to get back to New York and it was like shh. It's hard. >> Well, at least at least it was close. You could hop on the train and get back home. But Tiran, it has been so great having you on the podcast. I'm so glad you were able to join us and have a longer form chat than we're allowed to do on my shows. >> Let's do it again next time. >> Please, thank you. Thanks again to Tiranin for being with us on this episode of Market Matters. There are several things that stick out to me from our chat, but the first one was his thoughts on how you have to maintain a level of detachment from your investments, similarly to how journalists need to remove themselves from a story. You have to be logical and not wedded to a thesis that can change at any point. That directly ties into his thoughts about traders and fund managers sometimes being so deep in the weeds that they miss a profound change in the markets, essentially highlighting this idea of not seeing the forest for the trees. Everyone needs to take a step back and look at the broader market to be sure that this isn't happening to them without realizing. Lastly, how he explained this new chip supply chain and how anyone who doesn't own a factory can have their own chips. He says as long as they make chips that are effective and offer efficiencies, in-house chips will be successful and this could be a reliable trend for decades. Those are just a few of the things that stuck out to me, but that is going to do it for us today on Market Matters. Really appreciate you joining us. You can find us wherever you get your podcast. Please make sure to rate, review, and subscribe. I'm Marley Caden. I hope you'll join us again next week.