Hello, and welcome to a free preview of Sharp Tech. Hello, and welcome back to another episode of Sharp Tech. I'm Andrew Sharp. Joining me in the same room today, Ben Thompson. Ben, how you doing? Pretty flustered. Apparently, I'd forgotten. I don't know that I ever knew that your podcast voice is a good, what, 15 decibels louder than your regular? A little bit. I was just lectured as I did my hello. I'm trying to speak in hushed tones here as we record in person because I was allegedly too loud. I listened back. It sounded normal. But here we are, you know, sharing your basement together. Yeah, I don't know what to say. It's not quieter at all. He really is. He really is flustered. No, I, it's been a day of fluster to be totally honest. Hilariously, we plan for you and some other folks in this checker universe to come visit this weekend, which is going to be like the coldest weekend in literally like 45 years or something like that. I'm trying to talk loud on the podcast today because this is actually going to be the last podcast that I'm ever able to record. 20 below is the low tomorrow in Madison. I'm glad that I'm actually experiencing real Wisconsin winter, though, with my first trip. I don't think you're glad because you showed up in loafers and no socks. It was one of the most unbelievable sights I've ever seen in my life. Andrew Sharp coming down the airport stairs, utterly unprepared for what he's about to encounter. And immediately being roasted by Ben for about 10 minutes straight for the loafers. But look, one of my rules for adulthood is you have to dress well when you're traveling. So I had a nice pair of loafers, nice pair of slacks, a nice little Henley here. Casual. Don't do too much when you're flying. But I felt good. Look good. Feel good. Feel good. Play good. Or podcast good. I asked a lot, like, hold your hand walking to the car to make sure you didn't slip. Yeah, loafers with no socks. You're going to have frostbitten ankles. That's my best case scenario for tomorrow's weather. In any event, here we are. We have a lot to get through today. And we're going to begin with one of the most important tech companies of the century. And a company that we actually don't talk about very much on the podcast. So I'm going to read a note from Sam. That's a new problem, not a B problem. I was thinking about it. I was like, it's probably my fault that we never discussed TSMC. You certainly write about TSMC plenty. Well before it was cool, for the record. That's true. You were early. Early on everything. That's why people subscribe to Stratechery. Sam says, Ben and Andrew, what is stopping TSMC from charging more insane margins? They have clearly increased their margins after the AI boom, but they're the only option in town. And especially during the duration of this bubble, nobody can go anywhere else. Don't they have such insane leverage in the current moment that Apple and Nvidia would pay basically whatever they charge? Why have their margins not increased more? So Ben, we get this question every couple months. We actually answered it about two and a half years ago. But the same question came to mind with me. I was reading about the H200s and NVIDIA having to go to TSMC and negotiate for more capacity in order to serve the H200 orders. And it was just a reminder that basically anyone who's doing anything in AI is ultimately going through TSMC. you wrote about them earlier this week. Do you have an answer for Sam? Why aren't they able to just extract crazier and crazier margins from everybody? Well, before we get to that, I do want to compliment you, especially since I revealed your sartorial choices to the world and how inappropriate they were, that at least you think about TSMC and the fact that all the AI chips come from there. That is a big improvement over... Everyone. Well, you had Dario Amadei at Davos this week talking about giving chips to China is like giving nuclear weapons to North Korea. And what did I do as I do with every article or every comment from him about this issue? I do a control F. I search the document that I'm reading. I type in Taiwan and there's no results. Like, can we actually, so a little bit of a rant before we get to the margin question, Can we think about the – if these are the same as giving nuclear weapons to North Korea, what does it mean if the nuclear weapons plant is 80 miles off the coast of China? Like can we think – it is – it's unbelievable to me. Like just the – can we think through the totality of this issue? Anyhow, that ran aside. TSMC margins. So this is actually an interesting time to revisit this question. In broad strokes, TSMC – I don't know how long do you want me to go on this. There really is a cultural aspect to this. You have to remember TSMC starts 40-some years ago, and they have nothing. They get some process technology from Philips. there's actually an interesting universe of Phillips spinoffs. ASML is also a Phillips spinoff. TSMC is one. There's several other ones sort of in this sort of ecosystem. But what they can offer is the fact that they have nothing to offer. And what I mean is if you're someone who has an idea to make a chip, you can go to them and, number one, you can get guaranteed capacity. You're not going to get crowded out because at that point, the alternative is you go to texas instruments you go to intel maybe intel you like beg for extra capacity which by the way if they suddenly have more sales than they need you get booted out right and by the way so when you say what they can offer you're referring to tsmc back when they were trying to take market share in the beginning right this is literally all they have is number one we can guarantee you capacity we're not going to crowd out your order and number two because we're not making our own chips we'll work with you or whatever you want to make no we won't take your ip oh yeah like like which is a very sort of real and legitimate concern um there's aspects of intel is still not trusted uh samson ran into that issue as well right um samsung has gotten uh it's a little more complicated because obviously apple used to be on samsung i think samsung's done pretty well in terms of having a wall between their foundry business and the rest of their business. But there is a fundamental conflict of interest. If you're Apple making your chips at Samsung and Samsung is competing with you. Right. I recall reading something about that tension in the Apple in China book, but I don't know exactly what the circumstances were. But either way, in broad strokes, the conflict of interest just does not exist at TSMC. That was literally their selling point. Yeah. It's like, we're not going to crowd you out and we're not going because, by the way, our process is like five processes behind the leading edge. Like, so you start out and you're making your basic chips, right? The things that, that, you know, we don't even think about. We didn't think about until COVID when we ran out of these super basic chips and like nothing could be made. So they come and then the other thing is it's going to be super cheap, as cheap as we can offer it. What we're going to do is we're going to build these fabs and they will depreciate over five years or whatever the number is. Yeah. But we're going to run those fabs forever. Now, the earliest fabs have long since it's been closed down, but they still have fabs from the late 90s. Like 98, 99, I think is maybe the oldest fab that they still have in operation, which is making these ancient chips that are fine for what they are. And there are certain products, long-lived products that have just been standardized on this chip for ages and ages. Sure. And they sell these chips for pennies. Money on top, though. Right. They've already invested in the fab. Yeah, exactly. And so you have, in general, this very customer-centric, customer-first mindset because that was literally the only thing they could sell. Combined with this low-cost mentality, that's what we have to offer. We don't have the fastest chips. We don't you know, we don't have the leading edge processes, but we were a reliable partner for you. So culturally is it a struggle then to be extracting crazy margins because of the way they began yes okay that basically it so it it a bit a really difficult transition to go from that to being the leading premium and it's been this is one of the things that i've written about a fair bit i think would have made a lot of great podcast material but unfortunately you didn't care um but it's no but So one thing that was really, really interesting a few years ago was – so they had this model. You build a fab once. You run it forever. They incurred extra costs. I think this was two or three years ago because what they said is we're going to basically rework some of our five-nanometer fabs to be three-nanometer. Okay. And what's interesting about that is one of the challenges they have is they have these seven nanometer fabs that are still included in their advanced sort of fab numbers. But they're a little bit stranded because basically there was a turning point somewhere around 14, 15, where the chips just definitely got more expensive. And if you didn't need it, if you didn't need the speed, if you needed the efficiency, you could just sort of stop. Actually, the biggest stopping point was probably 28 nanometers. There's just a lot of and that's where China is really like invested a ton. And the chips are good enough and it's not worth the price premium to go to sort of the next step down. I wonder about that on a more general basis when people talk about Taiwan invasion scenarios. Like if everybody had to run on iPhone 13 level chips, how many people would actually notice the difference between what we have now and what we had then? Well, no, you have actually the problem is the opposite. the problem is and i wrote about this a few years ago um when i think the chip ban went down which is my point there was people are thinking too much about the leading edge right real issue is chips there's all these legacy chips that the u.s has no capacity global foundries might have some something or other but the reason they don't have it why intel doesn't have it this is where where i'm to go back to the five nanometer or seven nanometer story intel because intel was always on the cutting edge. That was their differentiation. And they internalized all the gains from that by we could sell the fastest chips because we have the best manufacturing. But what they would do is when they went to the next generation, they would dismantle the old generation or they try to reuse as much stuff as they could going forward. They didn't keep fabs going on forever. But what that meant was if you have this five-year depreciation, they needed to pay for the fab in that five years. Yeah. So but they could do that because they could charge very high margin. They're taking margin. So they would charge very high prices and pay for this. So what happened to TSMC is you have this overall market issue where like the 28 nanometer fab you built back in the day was the fastest of its time. But then it just became a great line that you could run for 40 years. Yeah. And so that old model sort of worked when you got down in this. You really saw this happen with 7 nanometer, where it was expensive because that was the first one to be using EUV. And it just wasn't worth it for most customers. For most customers. Yeah. And so it kind of is a little bit of a stranded node to an extent. It's still used. That makes sense because all the people who would pay to be on 7 nanometer would then pay to be on 5 nanometer. They'd already moved on, right. And so what was so interesting about this announcement, oh, we're going to incur more costs to transition, is that they had to become like Intel. They had to start thinking about actually these incredibly expensive fabs are not just cost way more, they're also shorter lived. We're not necessarily gonna be able to run them forever and sort of get the money from it, which means we have to raise, so it was almost more of a bottoms up, we need to raise prices and make more margin. Otherwise we're gonna be underwater on this seven nanometer investment or five nanometer, any of these notes. Right. I mean, they're not underwater, to be clear. Right. But the problem is money you don't make is you don't get it back. You don't get to go back to Apple and say, three years ago, we probably should have charged you more. And these are just like tremendously expensive investments. Oh, yeah. The latest ones are like well in the $30 billion. It's probably like those 7 nanometer ones are probably like $15, $20 or something like that. I'm just pointing out a thin air. So if you're putting $30 million or $30 billion down. Yeah. Was I saying them? Yeah. A million would be great. Big difference in them. So what's interesting about this is – remember this is also combined with the overall – and this is a huge strength of TSMC. Very customer-centric. They are a customer service organization. This is by far the hardest thing for Intel. It will continue to be the hardest thing for them to figure out is being customer first. But as part of that, they've had a natural disinclination to jack up the prices. Like they would talk about this. Yes, of course, we need to raise prices, but we also are cognizant of a long-term relationship. And there's a bit about this where that's actually a good thing because they keep loyal customers. It's a big – you don't just decide, I'm going to go with Intel this time. It's like a multi-year commitment to even go to another foundry. But this sort of locks them in even more, right? But what happened was, and this is all, just to be clear, this part is all, what's the word I'm looking for? Ben theory? Ben theory. Maybe a tiny bit of scuttlebutt that sort of infuses this. What happened with this 7-nanometer bit of stranding and then when 3-nanometer launched is they, you have this time period where they're by far in the lead. There are no alternatives. Their old model isn't quite working anymore. They need to switch to this model. And they did not raise prices nearly enough. And I think that might be what happened to the previous CEO who became the chairman of the board and suddenly retired. Who was that? Mark Lill. Okay. Again, this is just sort of my theory. It was sort of a weird transition and thing that happened. what I think happened is TSMC insufficiently raised prices. It cost them a lot of money. That's sort of gone forever. And it came from this sort of inherent customer-centric, but also conservative, also bottoms-up mindset. And this sort of bottoms-up, cost-plus sort of thinking is very endemic to Taiwanese business culture. Yeah. And so whereas what TSMC, to this emailer's point, What they need to do, you don't have to go like they need to become more like Intel. They were forced into that in terms of how long they can depreciate their assets and having to learn how to reuse stuff. But that means you need to do it from a pricing perspective, too. People hated Intel, not just because they were arrogant, because Intel knew they were the best. Putting the screws to them. And absolutely captured their value, right? I'm glad that we don't have to lay the failure to raise prices at Morris Chang's feet because I love the story of him coming out of retirement to lead the smartphone era of TSMC. It's one of my I consider that is part of the story. I'm glad you brought that up because what happened there? You had TSMC was in an increasingly strong position, not fully caught up to Intel, but caught up to Samsung in terms of the making arm chips and sort of being the third-party foundry. And the great financial crisis happens. The world goes into recession, and you have this conservative instinct to pull back happening. Morris Chang comes in, fires everyone, and is like, this is the biggest opportunity. this is like where this company's ever seen we're investing into this and that's how this whole last 15 years happened now is morris chang involved uh i feel like if the once ceo then chairman of the board ceremony unceremoniously retires at when he i don't think was ready to retire that might have more fingerprints on it um yes uh again this is pure conjecture that's what we're looking for okay this is been been conjecture a tiny bit of like scuttlebutt there might be something here but nothing i mean it's not reporting this is this is conjecture um regardless about this time frame i was hammering on this in the daily update i'm like tsmc is screwing up their pricing and it's a problem for all it's not just that to the emailer's point they have the opportunity It that the fundamental structure of their business is changing They becoming like Intel whether they want to be or not Intel in a positive sense where they on the leading edge They have to capture a margin much more upfront. They can't count on this trailing in the back end. They need to capture their value, and they need to capture their value. The reason why it was so damaging, that five nanometer, three nanometer, where the two nodes I think were underpriced, that was the time there was no alternative. Right. so fast forward to today that well okay that was going to be my question because when we had a conversation along these lines like two and a half years ago you talked about nvidia wanting to second source or at least have another player that they could pit against tsmc which they did which they always did like their previous generation or a few generations ago uh was the ampere generation like they did like the gaming chips with samsung and then like the server chips tsmc and that's how Nvidia was able to keep TSMC from extracting too much margin in the value chain. Is there anybody who can play that role today? Or is it basically just TSMC that's able to serve the AI demand right now? Well, this is where it gets really interesting. And this is what I was writing about this week. So the issue is TSMC is the best, still the best, probably will continue to be the best. there's certainly like rumblings about in about intel about 14a maybe being good but even there it take like to commit to intel today is not going like you're not going to have chips come off the line for like three or four years yeah like it's it's maybe now again people might have already committed it's not announced so that this sort of could happen sooner um or same thing with sort of sam it's not a trivial thing to sort of map a chip on onto a new process um and again if TSMC is there and willing and the best, why would you want to go somewhere else? And take that risk three or four years out. Exactly. The issue, however, is that I think it's fair to say, it's clear now, that the sort of conservatism that Morris Chang fired everyone for in 2008 and the conservatism that led to them underpricing four or five years ago. And perhaps in early years. It has manifested itself in their CapEx. Okay. And what I mean is after ChatGPT happened, there was a choice to make, which is how much demand do we think there's going to... So this is what, 2022? How much demand do we think there's going to be in 2025, in 2026? and TSMC has been, was relatively conservative. And the net result is because these decisions, this is the whole, what's so hard about semiconductors because decisions are made years ahead of time. The issue is that there just isn't nearly enough capacity at TSMC for all the demand. And everyone's sort of stuck because to go somewhere else is going to take a few years. but at the same time but but tsmc is not there and also they're the other ones aren't good enough and so everyone you had this whole thing last year of like sam altman visiting taiwan and jensen's here and everyone's like they're basically coming to tsmc saying please invest more and the invest but the tricky thing is the invest more isn't about 2026 the invest more is about 2028 and 2029 Yeah. So TSMC comes out. The reason why this is a big deal for earnings is they announced their CapEx plans. And they announced, I think, between $52 and $56 billion they'll spend on CapEx this year. A lot of money. This year they spent $41 billion. I think last year they spent $30 billion, which was way too low. Like the last three years was – probably the 2022 to 2024 eras were particularly too low. And this raises a really, really interesting question for all of the TSMC customers. Is $52 to $56 billion in a context of prices up in general, all the equipment is – so it's not a linear increase in capacity. It's like the curve bends in the wrong way because just stuff in general is sort of more expensive. Is that enough for the demand in 2028? in 2029 doubling it is that enough yeah well but they're not doubling it it's like a fifth it's like from last year it's like a 25 increase okay so i thought it was 30 billion and now it's no it's 41 41 in 2025 um 30 billion and like around the 30 billion mark in the years before that so even from there it's like a 67 increase yeah and the reason and the problem is that tsmc the reason why TSMC is nervous is think about this timeline. It's reasonable to be nervous. Right. So if you show up, if there's the bubble burst, say we have, 2029 is three years away. I keep reading it's 2026. The bubble burst in 2028. You've spent all that money and you spent money next year and after that and suddenly you have fabs with all this equipment you spent that no one needs or wants the chips from. Yeah. And that is like- That's the risk of their business is it's so capital intensive that if the demand ends along the way- You're either extremely profitable or you're just not profitable at all, right? It's a big- That's what I mentioned with them being underwater. You can really struggle in follow years. But the issue is, but the problem, So TSMC is trying to reduce that risk. Risk doesn't disappear from the system. That risk is being offloaded to TSMC's customers. How so? Whether it be NVIDIA or Microsoft or Google or whoever it might be. Because what does every single CEO say on their earnings call? I don't know. I don't listen to nearly as many earnings calls as you do. They say, if we had more capacity, we would have sold more. The risk that TSMC is offsetting is foregone revenue for all these companies. And all these companies are realizing that risk right now. Every earnings call, when a CEO is on there saying demand vastly exceeds supply, what that is, is foregone revenue. That revenue is gone forever. And that revenue is downstream from TSMC not having enough capacity right well and so is tsmc i mean by themselves are they gating the ai infrastructure build out the bubbly conditions no i called it the tsmc break like i came up with that last year um i should have wrote a big article because i think it is it's like they are the break on a on a bubble right on on sort of ai sort of generally well no i when you wrote it last year, it didn't fully register with me. And then I read it earlier this week and I was like, oh, he's making a play on accelerationists and TSMC. They're the ones that are the brakes in the middle of all this. And it really is fascinating when you step back and look at the ecosystem and think about how much crazier the numbers could be if TSMC could serve the capacity in the middle of all of it. That's exactly right. So so all these so the issue is that all these folks who they think they're de-risking by sticking with TSMC because it's the known entity. They have good customer service. They have confidence in a work. There's a ceiling on how much they can produce. They have been on how much you can make. They actually have been loading on fab risk to themselves. And that risk they're loading onto themselves is foregone revenue, which is being realized right now. And you could imagine if AI progresses like people thinks that it will, in 2028 and 2029, this mismatch between supply and demand, 52 billion is not nearly enough. Interesting. And so you're actually loading up on risk. I'm glad you said that because reading your update earlier in the week, I was reading it early in the morning and I was like, does Ben, it seems like Ben thinks that they should be investing a lot more in CapEx build out than they actually are. And it sounds like that was an accurate reading of the subtext of your analysis. I think that TSMC is doing what is right for TSMC Okay And they able to do what is right for TSMC because they don have any competition There no competition yeah What I think behooves everyone in AI to do is they have to get Samsung and Intel on board. So Intel's not just a charity case in this scenario. Well, I mean, this has been, but I think making this clearer than ever, I should have made this a poker call. Maybe I'll write another one. Like they need to – because what happens if Intel is a credible alternative or Samsung is a credible alternative? What happens is what becomes TSMC's greater fear? Is it five years from now we might have this overhang or is it we're losing business to Intel and to Samsung? And we know once someone switches, it's hard to get them back because of this long sort of cycle. And right now, the risk to TSMC of under-investing in CapEx build-out is, okay, we're going to make less profit in three or four years than we might have otherwise. But we're offsetting that risk against tremendous downside risks if the bubble bursts. So we're happy splitting the difference with a 67% increase as opposed to 167% increase. That's their logic and the way they view this. Is that right? That's right. And the way and so in but that what they're doing is changing risk risk onto their customers. And there's a much bigger risk if they get five or six years down the line and half of their customer base is working with other competitors. Well, what you want if you're a Google or a Microsoft or an Amazon or an OpenAI or an Anthropic, you want cheap chips. The way you get cheap chips is by there being too much capacity and they're having to sell it at very low prices. Yeah. The way you get more capacity is you have everyone overbuilt. You want overbuilding. like you you want tsmc facing it's why sam flew to taiwan no sam's totally wrong oh i interviewed sam in the fall and pushed him on this like you need to be exploring intel and he doesn't want to do that because everyone's a little scared of tsmc because tsmc has limited supply they have limited capacity so they can play favorites that's right yeah that's right but the problem is that The issue is you're getting – you're enabling the break. It's like the BR-A-K-E, the slowdown. And the issue – like what needs to happen is people need to – Intel and Samsung are not going to get there without customers actually going for it with them. They need this customer base. And the reason to do it is it shifts the risk back to the foundries. You want the foundries taking risk. You want them building for a huge explosion. The worst case scenario is you get super cheap chips because they built too much capacity. That's a good situation. And you're not going to get there. All these companies need to sack up and stop – like they need to think about – And the clear-eyed recognition would be TSMC is just not going to go that direction unless they're forced to – It's kind of pathetic. You're flying to TSMC and begging them to put their business a little bit more at risk for your sake. No. What you need to do is you need to go out and you need to empower and enable an actual competitor for TSMC. That's how you get more TSMC volume. And by the way, that competitor comes online, there's more volume for everyone. That means lower prices. That means you can actually start like like create chips and your capital costs are lower and you can actually not have these insane bills. Yeah. That all these that all these companies do. This is great stuff. You really should have made that update public. Bad job by you. Well, I have one question on TSMC big picture and like potential risks that they incur in the midst of all this. how much of their revenue comes from like three customers at this point like it's nvidia apple uh amd okay amd and i mean like in the ai space well i mean everyone in the world fabs at at tsmc their issue right now is is not that they have limited customers it's that there's too many customers and people are by no choice having to at least consider alternatives because there's not enough that you know that's so there's a two-part thing number one they don't want to raise to prices too high because people could leave but they also need to have enough capacity so that people don't leave because they can't get leading edge chips what they need right so um but the issue is that you don't it's like all these folks who claim to be big capitalists are not are like they're being chickens they're not taking the risk of trying to get an intel off going to get a samsung going because so they're they're thinking about their short-term risk of what if you know might be expensive what if it doesn't work what if there's a delay what if tsmc gets upset and starts playing with our supply what they're not thinking about is the long term like there is a scenario where we no one does this we get to 2028 2029 and this entire opportunity is actually what actually ends up bursting the bubble is there not enough chip capacity like this entire opportunity is sort of uh killed off because no one was everyone was being chickens in 2025 yeah or there's a war that's another reason another reason to uh build up alternatives yes yeah um well any final thoughts on tsmc before we shift to netflix here like if you can't get morris chang to give them a kick in the rear end and take the risk uh you get sort of mr market like that that that's the answer um and we'll see what but this is this is the time now the time now is to make decisions for 28 29 and i think there's a bit about tech companies are uncomfortable thinking that far in the future even though like software you obviously it takes a long time to build but things like this like meaningful capex investments it's almost like a new muscle for silicon valley just assume you've always been able to assume the hardware is there right right and especially with the rise of the cloud and sort of being able to rent um but there needs to be this development and this increased appetite for risk because you're actually risking more by not thinking about it than you might realize. Risking foregoing profits. It's an interesting corner of the space though, because even TSMC, it's hard to chart a course for where all of this is going to be and whether the bubble will or won't burst and whether the demand will or won't be there right but that four years but you want to put that risk on the foundries the way you force well if you're one of the ai companies of course you do but it's a fascinating aspect of the whole discussion because they're the ones that really do have to think about the end of the decade as they're making decisions today uh tsmc that is in addition but if you if you're an ai company you need to think about the end of the decade too because you're going to show up at the end of the decade and not be making nearly as much revenue as you could i mean sam ohman in his The interview with you is making all these investments today for what demand will look like in 2028. It's like winding again over and over again. We don't have enough compute. We don't have enough compute. We don't have enough compute. Well, then. Or we want to be ready when there's even more demand than there is today. Right. They're not taking the steps they need to take. To CC Way is. Short-sighted. Yeah. Well, it's. Yeah. All right. Well, on that note, we can shift gears and go to Netflix because you interviewed co-CEO, Netflix co-CEO, Greg Peters this week. All right. And that is the end of the free preview. If you'd like to hear more from Ben and I, there are links to subscribe in the show notes, or you can also go to sharp tech.fm. Either option will get you access to a personalized feed that has all the shows we do every week. Plus lots more great content from Stratechery and the Stratechery plus bundle. Check it out. And if you've got feedback, please email us at email at sharp tech.fm. you
