Odd Lots - This Is How the World Ended up with a Shortage of Semiconductors

Episode Date: February 22, 2021

The world is facing a chip shortage. Numerous companies, including the auto sector, are facing an inability to get semiconductors, hampering their ability to manufacture their goods and generate sales. Part of this is an acute crisis, related to the virus. But there's also a long-term structural issue, with so few companies able to manufacture at scale. On this episode, we speak with Stacy Rasgon of Bernstein Research, who helped kick off our semiconductor series last fall, with a discussion about the current problem, and how it will get fixed.See omnystudio.com/listener for privacy information.

Thanks for listening to Odd Lots. Follow the show on Amazon Music for more future episodes or just ask, Alexa, play the Odd Lots podcast on Amazon Music. Hello and welcome to another episode of the Odd Lots podcast. I'm Joe Wisenthal. And I'm Tracy Allaway. So Tracy, you know, a long time ago now, it feels like we had the joke about, well, should we just turn this into a semiconductor podcast? Yes. Yes. And you've really run with that joke. Okay. Well, the thing is, is we can't escape it. Like, we keep, like, you know, we, like, first started talking about it. It's like, oh, this is, like, an interesting topic for us.

But it turns out little did we know when we first started covering this story on the podcast, which I think was last October, last November, that actually it would blow up into this huge issue, semiconductor manufacturing that became, like, essentially a topic of national news far outside the sort of like. the niche audience. Right. Semiconductors secretly rule all our lives. And I'm joking, obviously, but nowadays, everything is so high tech that there are a lot of appliances that you wouldn't necessarily think of that have chips in them. So smartphones, computers, things like that, obviously, but also lots of cars. And I saw one headline float by. I haven't had a chance to read it yet, but something about aluminum producers warning of a downturn because of a chip shortage as well. So chips are everywhere, and I think we're really starting to realize how important they are and also how important chip makers are. And of course, as we've been discussing, there's a limited

number of those. Right. And so anyone who sort of listened to our series, we started talking about the decline of intel. We talked about why U.S. manufacturing in general is sort of gone away and we talked, of course, about the dominant role of Taiwan semi. And it's like almost like, again, it was not intentional, but now there's like this huge thing and everyone is waking up to how dependent we all are on Taiwan Semiconductor and a few other major fabs. And it's become a legit U.S. national security question. We know the Biden administration is looking at it and, you know, looking at different ways to reduce U.S. dependents. So we really can't get away from this story.

And obviously we're going to be talking about it again today. And I bet it won't even be the last time we talk about it. I think you might be right on that one. So, well, I'm really excited because actually we are going to be going back to our very first guest who gave a sort of great overview of some of the best clear English descriptions of the challenges of chip manufacturing. And it was the first one. Everyone should go back and listen to that one. We were then talking about the decline of Intel.

But we're going to zoom out a little bit and look at the acute shortage that the world is facing right now. Why are so many companies struggling with their ability to source chips? And then the longer term issue of this is scarce and capacity is scarce. And even if we get through this current phase, there is going to be still this sort of perhaps a dangerous overreliance on a few manufacturers that are capacity. Limited. So very excited. We're going to be speaking again second time on the show. Stacey Razgan, his managing director, senior analyst U.S. Semiconductors at Bernstein Research. Stacey, thank you so much for coming back on OutLod. That's fantastic to be back. Thank you for having me again. What's it like? I mean, you know, you're like a star now because at one point, semiconductors were just like

this thing that maybe investors mostly cared about. But it really does feel like, you know, me if I'm wrong, but it really does feel like in the last few months, everybody is now obsessed with this story. Oh, you know, I think I may have mentioned this last time I was on, but I, there's one reason I love this space. It's literally ground zero for everything that's been going on. And it's, you know, it's not just the last few months. It was all the trade and the tariffs and then the burgeoning geopolitics and now, obviously,

the shortages and everything else. Like, it's, and like, you have to remember, I mean, like, the global electronics, like, enterprise. I mean, it's like a $4 or $5 trillion industry like worldwide. If you add up all the PCs and all of the consumer electronics and then all the services and software and everything that goes with it, trillions and trillions of dollars. And it all rests on semiconductors.

Semis are the fundament of all of that. Like we don't have any of that without semiconductors. And so I think it's a phenomenal place to spend time. And it's job security for me too. So that's a... Yeah, that's the most important thing. that helps for sure um so i guess just to begin maybe you could describe how bad things are at the moment in terms of the chip shortage um it depends on on the market obviously automotive is getting

the most press and that is where the shortages are most are most acute although it is they are starting to bleed into other areas um slowly but they are and i think the driver is a little bit different let's talk a little bit about automotive great and automotive was was sort of unique um most end markets in semis through COVID actually proved to be fairly remarkably resilient. Automotive, however, had some fairly sizable shocks. And if you sort of look at like end market production for cars, in terms of like year of year growth in the trough quarter of COVID, it was actually worse than what we saw in the finance, in the trough quarter of the financial crisis.

I think in COVID, in the in, in, was a Q2 or whenever it was or whatever month. It was like I remember. It was down like 40% or something like that in terms of production. It was a lot. And then it's obviously production snap back faster than the financial crisis. But it was a very big hit. Part of that was obviously because of supply. So we have a lot of auto plants that are obviously in places like China,

but also in like the U.S. and Europe. And they were all shut down with the early lockdowns. And then there was a demand shock as well. And people were trapped in their houses and they weren't driving anywhere. And autos were a pretty big purchase. And so people weren't like rushing out to go by cars. And because of that, the auto OEMs, I mean, they kind of in some sense did this to themselves. they went and they canceled all the orders for all the chips and everything else that goes into

this stuff because demand did collapse. And the problem is when they canceled those orders, for companies like, you know, an NXP or an Infinion that have their own factories, and they reduced their factory loadings. They didn't make the parts. And, you know, for parts that were being outsourced to companies like TSM, you know, those orders, those auto orders went away. But like other demand for other things was so strong that that capacity immediately got backfilled. Right. Now we go forward a little bit and, you know, some parts of the world, even in the U.S. and Europe, like the auto plan started to open up and then parts of the world like China,

who actually came out of COVID, you know, I guess good for them, much better than we have in the U.S. here. Demand started to go off the charts because, you know, people, they didn't want to take public transportation. They wanted to drive. And so there was a lot of pent-up demand and everything. And so you have this massive demand snapback. And so they all went and to recover the orders. But the problem is you can't turn a semi-com. factory on and off like a light switch.

Like if I take my factory loadings down, if I reduce that, I mean, if I start a wafer, a raw wafer today, I don't get chips out the other end of the factory for three, four, five months. And this gets the last issue of automotive. They don't run with any inventory. The auto supply chain has tended to run much more on like a just in time kind of inventory management, not just for semis, but like across the board. So there's not a lot of slack in the system.

And so now they're coming back like hat in hand. they want seven. They're just no parts. It's just going to take time to build them. And so this is problematic. And this is why we're seeing the effects to the degree that we are in automotive. Just to be clear, when you say the effects that we're seeing to the degree that they are, I know I think these stories have really emerged over the last month and a half or so.

What do you put into context? How big of a problem this is for the industry? Well, I mean, IHS just came out with their forecast. They think the Q1 were going to lose close to a million cars in terms of production. levels. Wow. Right, which is what I mean, you figure the industry, I mean, what's a typical, like, sometimes you'll hear me say SAR. The SAR is, uh, is, uh, people refer to that. It's an acronym, but it basically means like how many cars are they going to sell in a year, seasonally adjusted annual rate, I think is what it stands for for, for car sales. Um, you know, normal SAR is a lot

of ballpark, 90 million cars a year get built something like that. Um, and so we've got a million in one quarter. I mean, it's, it's a decent amount of cars that are missing. Hopefully those get those will get built later on in the year. You know, hopefully you can argue that that demand will stick around. But for now, that's the kind of impact. And you can, I mean, make a price assumption for what a car sells for. And that sort of tells you from an auto OEM standpoint how much revenue they're losing. I, you know, GM and Ford have put out some numbers.

And it's in, you know, it's in the multiple billions of dollars in terms of many billions of dollars of lost sales, at least for now, that's, that's impacting in the auto supply chain. Now, in terms of the auto semi-suppliers, they're actually a really good place because if you sort of look at these dynamics, it also suggests that the auto recovery from a semi-standpoint has some legs to it. The reason, and there's a number of reasons. One is obviously there's just selling everything that they can possibly make right now. The channel is bone dry. There are no parts. And so everything that the semi-guides can possibly make is getting shipped and it's getting put into a car.

And they're ramping their production levels up. And so that will carry on for a while. And you're not just like going to eventually to go into a car. You also need to, to refill the channel somewhat. And then longer term, it is highly likely, in my opinion, that the auto supply chain may be rethinking this whole just-in-time inventory management. And so if they need to run with higher levels of buffer inventory, that will be an additional layer on top of everything else that will need to be shipped to build that level of parts up since that's good.

And then if you just look at the overall level of the recovery, we're still early in the unit recovery in generally. If I look at like the IHS forecast for autos, I think the Autosar was down 16% in 2020. I think the current forecast, at least like from IHS, have it up like 14% year over year in 2021, but at levels that are still below pre-COVID levels. And I can make the argument that post-COVID,

maybe auto demand in general will be higher. You know, again, maybe people will want to drive more and everything else. And so I can make the argument from the semiconductor side of things that this has some legs. But we have to see how this works out. We're going to get more. disruption. Obviously, these guys were all trying to build this up. But you have to remember the other

problem with cars is there's a ton of semiconductors in them. And they come from all over the place. And you need all of them. You need the full kit before you can build the cars. I'll make it up. But if you're building a $50,000 automobile, if I'm short the one $2 microcontroller that goes into the driver's seat with the motor to make the seat go back and forth, if I can't get that one part, I can't build the car. I need everything. And so you'll have some companies that sort of start to correct like faster than others maybe, but in general, everybody, the whole supply chain kind of has to get back into balance before things can really be smooth again. That's, it's going to take a while. But it's probably a good thing to be an auto semi-vender right now.

So, I mean, what's coming through from those conversations that cars are clearly the most affected? And this brings me to a question that I've been wanting to ask for a while. But, But given we have an overwhelming demand for chips versus available supply, how do chip makers actually decide to allocate their orders? So I imagine if they get an order from, say, Apple for a million chips, that's probably going to be treated differently to an order from Ford or something like that. Yeah, and it depends. And so, again, there's two mechanisms.

We may have talked about this a little bit last time, two mechanisms that the industry uses to make chips. So one is, like I said, it's called an IDM model, integrated device manufacturer. These are companies that make their own chips. And then there's companies that outsource production to what's called a foundry. This is a company like TSM in Taiwan, right? And some companies split the difference. Like, you know, NXP, they make a bunch of stuff in-house and then they also outsource,

depending on what it is. So in some cases, they've got a mixed model. I guess what you're referring to primarily is probably the, the, the, the Foundry model because they're like, TSM is a company that makes products for everybody across every end market, right? This is another problem, at least with Otto. Auto is a very small piece of TSM's business. Even in a normal year, it's like 4% of the revenues.

And so it went down to like 2% in 2020 because like I said, when they canceled those orders, like other people backfilled them. And like everybody went bananas obviously over this. We're seeing, you know, like Germany and like all these other companies, they were trying to put pressure on TSM. But I mean, at the end of the day, like auto is. not a very important market for TSM. It's actually very small. So, I mean, it's quite obvious they're going to obviously prioritize like higher volume stuff, especially if it's there.

We're actually hearing now that they may be starting to push some of these auto orders, like through the fab faster now to try to make up for this. And they can do that. But you, you disrupt others. You got to push other guys aside. And so it costs money, typically, cost extra. So we'll see what that means. But yeah, I mean, in general, I mean, you order, at least at the foundry, like you put your orders and you order a certain amount of wafers, a certain amount of capacity. It depends on the relationship. If it's a big order and you've got a long relationship,

yeah, you'll probably get prioritized. If you're small, and again, you know, if a lot of this stuff, maybe the volumes aren't that high. I mean, that becomes problematic, especially if you give up those capacity slots. And that's what these guys did, right? They gave them off. Well, can you explain that a little bit further?

Like, how commoditized, I guess, is the production capacity such that, okay, the automakers canceled their orders in the spring when it looked like the industry was collapsing, or at least there was this lull. How translatable is one type of chip capacity to another such that they could just fill the slot with a totally different kind of customer, I guess.

I mean, at TSM, it's fungible in a process-by-process basis. So, I mean, we're talking like, you know, I'll make it up. If it's, you know, if there's something leading edge, it was some sort of advanced processor that goes into the infotainment in the car or something. Maybe that's being made at TSM on like their 7 nanometer. like that gets canceled up.

I mean like media tackle snap that up in a heartbeat, right, for the smartphone. It's the same process, different chip, but same process. Okay. On a more lagging edge thing, you know, analog, I mean, because the other thing is there's been a very strong like in general like industrial demand. And so even on mature node stuff, if that capacity opens up, it's available. And like a TSM, I mean, you're basically booking wafer is at a certain process, mostly.

And, you know, it's in some sense it is fungible. It depends on the product. There are some products that are, you know, use more special. specialized manufacturing, more specialized materials or structures. And that may be less so. But by in large, I mean, especially for the stuff that's being outsourced, yeah, that capacity is available. That's that's TSM's business model. That's what they do. Hello, I'm Michelle Hussein. And for more than 20 years, I was at the BBC. But all the time I was delivering the headlines, I wanted to go further than the news of the

day, to spend more time with the people shaping our world. And that's what we're going. what I'm doing here on this podcast. Speaking to people from Nigel Farage, to love you trying ever so hard, to listen to love you trying ever so hard. To tech journalist Karaswisher and the tech industry is running wild. You know, they've gotten what they wanted and they've seen a huge run-up in their stock prices. This will be a place where every weekend you can count on one essential conversation to help make sense of the world. So please join me, listen and subscribe to the Michelle Hussein show. show from Bloomberg weekend, wherever you get your podcast.

You certainly ask interesting questions. So Stacey, you mentioned the idea that the chip shortage is bad enough that maybe it's going to encourage car companies to rethink the way they're actually ordering and stop piling all these chips. I guess the other big question is, will it be enough to encourage the factories themselves to try to expand supply in a meaning? way or can they actually do it? And also will it encourage efforts by governments to build their own sort of independent supply of chips? Yeah. Well, that's your second question that's already happening.

We can address that. There's a whole thing around trying to bring more to diversify the manufacturing supply chain and bring more back on. And you guys have talked about this in other semiconductor odd lots, the whole, you know, full dependence on Taiwan and everything. So that's going on anyways. we've actually had some talk. I mean, like I know there was a letter that went, I think, from the SIA, signed by a whole bunch of semiconductor CEOs recently. And I know people have been to the Biden administration. And I know people have been pushing them to try to do something to alleviate the capacity

contract. There's not a whole lot they can do in the near term. I mean, even if they decided tomorrow to fund a bunch of capacity expansion, you know, it takes years to build this stuff out, right? So nothing they're doing will affect the issues right now. The only thing you can do right now is you can encourage the guys that do their own manufacturing to ramp capacity, which they're doing. And I'll get to the other part of your question in a minute on that. And encourage TSM, you know, and other foundries to try to accelerate the auto waferes through the fabs, which again, they seem to be doing in some sense.

But it'll take time. A lot of the guys that do their own manufacturing actually are increasing capacity. And, you know, NXP talked about this little bit. They're ramping. It just takes time to do it. Like none of this stuff is instant. Right. You have to buy the tools.

You have to install them. You have to qualify them. With automotive, in particular, the qualification, there's what's called a qualification process to make sure that the part itself works. And you have to qualify the part typically made on a specific process at a specific location. And the reason is within automotive, the reliability levels need to be really, really high. Like you can't, like, have, recall thousands of $50,000 automobiles because a $2 microcontroller doesn't work.

Right? So there's a whole qualification. process you have to go through too as you're bringing capacity up. So it just takes time. As far as TSM goes, I'll be honest, like, auto's not a big piece of their business. I don't know how eager they are to bring on more automotive capacity in general. But it'll get to the other point of like where the rest of these constraints are coming. TSM actually is adding a ton of capacity this year. Right.

Because demand is off the charts. So that was something that came out well after our last discussion. And I think it was even after when we did our TSMC. specific episode. It might have even been like a couple of days later. What they announced was like something like they were going to do $28 billion in fresh capital expenditure. 25 to 28 billion this year, up from like 17 last year. Yeah, I was just going to say like, what is that, put that into context, because that number came out, TSM, tell us how they're going to do 28 billion in capital expenditures. How, what does that mean? How big is that? It's like 50% of the revenues. It's big.

Intel's average capital expenditure given your ballpark $15 billion. Wow. It's big. And frankly, some of it actually may even be Intel because Intel will be doing more outsourcing, although my guess is, and when they saw that, when people saw that guide, actually, people really thought Intel was to be doing a ton of outsourcing. Although when Gelsinger talked on Intel's earnings call recently, he basically said, we think we fixed seven, like we'll see.

We think we fixed it. And the majority of our products, like in 2020, we'll be. be in-house, although they will be increasing their use of outsourcing. So, but this is a, this is a 2020 guide. Like, I think a lot of the, just in general, demand overall is off the charts. So you can look at, and this gets to your other, the question you asked like first, which is, what about the other supply constraints outside of auto, right? So they're actually being driven more by, by demand versus, like, massive whipsaws in, in the supply situation. Like, we didn't really see all that much. And by and large, like I said, Sundays in general through COVID,

have been pretty resilient. PCs have been all. off the charts, like anything that was work from home, study from home, play from home, PCs, 5G smartphones, game consoles, GPUs, AI data center stuff, industrial in general, especially in areas like China, it's all been just really, really strong. And like PCs, for example, and we had 300 million PCs in 2020. We haven't seen numbers like that in years and years. Notebooks grew like 35% year over year. Total PCs grew like 16 or something.

To give you some context there, the price. kind of normalized run rate for the industry, it was like 250 million units and it had been falling for a decade. All of a sudden, we went to 300 million. I mean, like demand is off the charts. Smartphones, smartphones have been kind of weak, but 5G has been incredibly strong

and 5G has much higher semiconductor content. Game consoles, go try to buy a PlayStation or a switch, like you can't. They're sold out. Same thing with a GPU, right? Like that demand for leading edge capacity, TSM is sold out. I mean, they've been sold out.

And then on the more mature node stuff, industrial has been very strong. And again, we can talk about some of the risks here, sustainability and everything else, but areas like in China that were covered very rapidly. I mean, China's GDP grew this year, right, in 2020. Industrial has been very strong, and it just sucks up all that capacity. And TSM's been tight forever, and they're even tighter now. And so they're adding, right?

And I think they've talked about kind of multi-year growth. You know, I can't remember 20% a year, I think, for the next five years or something, I think is what they said. But that's what's causing most of the other shortages. It's less of a, you know, the, the, the, the, supply went offline and then you have to come back. Just demand's been off the charts and just there's only so much supply that's out there in the first place. So this feeds into something else I wanted to ask, which is how normal is it in the chip industry

to have capacity constraints after a downturn? Like, does it always take some time for supply to balance out with demand, given it's hard to sort of estimate, you know, when people are going to start buying computers and things like that again? and how exceptional is this current situation versus what we've seen before? Yeah, I'll need to step you back, but first I've got to tell you,

2020 for semis was not a downturn. Semis, semis grew 6.5% year-over-year in 2020. 2019 was a downturn. 2017 and 28 were very, in 2018 were very strong. 2020 was a good year in 2021. I mean, if you just roll up like normal seasonality off of where you are right now, I mean, you could be up double digits,

like assuming we don't hit a wall. We can talk about that. But 2020 was not a downturn. And that's part of the thing. I think early in 2020, people thought it would be a downturn, right? And so you sort of planned accordingly, and then it wasn't. You sort of asked, like, how does this compare versus history? The cyclical nature of the industry has actually changed quite a bit.

And so you're right. It used to be very cyclical. And you would get big swings in industry utilization and big mismatches between supply and demand. And the reason is, I mean, you can go back to like two. thousand, for example. This was like, this was the tuck bubble, but it was sort of an extreme example of a supply-driven cycle. What used to happen is demand would be strong, supply would be tight. Companies would build out incremental capacity because you'd be leaving money on the table, right? Pricing would go off because like end customers would be fighting to fill up that capacity

and then to get the products. It would take, you know, 18 to 20, 18 months or whatever it was to sort bring that capacity back online, to bring the new capacity on. Usually by then the business cycle would be turning. All of a sudden, and I have the charts, like the CAPEX historically would come on right when demand was falling off. All of a sudden, you go from full fabs to empty fabs. You would lower price now because, like, once you have the acid in the ground, it's always better to fill it with something rather than nothing as long as you're covering your cash cost. And so you get, in the industry units in pricing moving together, and you get really big swings. And I mean, it wasn't uncommon. We used to see like normal revenues swings year-over-year in semiconductors. It could be 20, 30 percent in either

direction. Like it was very, very common. And like I said, 2000, 2001, the tech bubble was sort of an extreme example of this kind of supply driven cycle. We don't really get these kinds of cycles anymore. We get them in memory. But for the rest of the industry, supply and demand has actually been much better matched over the last like five to 10 years. Yeah. I remember like when I first started, like I briefly had a job in like 2004, 2005 and it involved some equity research. And like the book to bill ratio. Isn't that like a thing? Didn't that used to be a big thing in semis? Maybe it still is, but yeah, it depends on the company. It still can be. But like that was like a big indicator in terms of like where we were at any given time for the industry in that cycle that you just described.

Yeah, yeah. It's less important now. I mean, we've had just like I said, we've had better, you know, kind of channel management. We've had some companies like Texas EnSonsor, for example, that do a lot of stuff on consignment where the book to bill by definition is one anyways. So it depends on the company. For some of these large diversified guides that sell a lot through the channel book to bill can still be important. For the semi-capped guys, book to Bill

in some sense can still be important. But it's probably a less meaningful metric than it was, you know, like in the early 2000s, which is when you're talking about. Yeah. Like I said, these kind of supply cycles, we do get them in memory. And memory is a supply demand-driven business.

And we can have probably another whole call just on memory at some point if you wanted to. Great. But in general, for the broader industry, we don't really see massive supply cycles anymore. What we mostly see are inventory cycles. Like we'll see, we haven't, we've seen constraints before, but we've never really had constraints that were in some part driven by a by a potential global catastrophe. That's we we've had we've had natural disasters and things that have caused problems. And I want to talk about that.

Because I remember like 1999 and this is something that actually I want to bring up because I remember like I know 1999 that was an earthquake in Taiwan and briefly not. Taiwan semi's production off with the world so dependent on Taiwan and Taiwan is on a fault line, what are the risks of a big earthquake that seriously disrupts the semiconductor supply chain, which now, as we understand is the supply chain of everything for a sustained period of time? And how much of that is an issue? Yeah. Look, so, TSM, I mean, they build pretty well in Taiwan and they're prepared for earthquakes.

But if we had like an eight and a half or something in Taiwan, yeah, it would be really problematic. Sure. This is one reason that they, in general, people have pushed for diversification. Like you take global founders, for example, I mean, the global in their name is not an accident. Like part of their value proposition was we have fabs everywhere, right? They're in Singapore and they're in Europe and they're in the U.S. Right? Obviously, they don't do leading edge anymore.

It's other stuff. But that was part of the theory their value proposition was geographical diversification. I mean, you can get disasters anywhere, obviously. I mean, even just recently we had the. earthquake in Japan. You know, we've, we've got the storms in Texas right now. We've got semiconductor fabs, a number of them in Austin that were shut down because of that. But yeah, I would say like if we had some like massive like natural disaster in Taiwan, it would be problematic. Yeah. So I have a million dollar question for you, but, you know, is, is the

shortage going to be intense enough or bad enough to help Intel? Intel's a little different, right? So they've had some shortage of less a year or so, but not so much because, I mean, they brought an additional supply. It's just PC demand has been off the charts, right? So it's not like if you're asking, like, because of the shortages, are they going to pick up incremental customers? No. They do not have a custom foundry business, right?

And this gets part of the argument of like building new factories in the U.S. And we've written about that. We may have even talked about this a bit last time. I think I told you last time, like, the $20 or $25 billion over like five years, which is I think what some of the numbers we're hearing about in the U.S. It's rounding error. I think I use the phrase, we need an Apollo moment, right?

But you could imagine, like, what could happen? Let's imagine a scenario where, like, Intel was anointed and national champion and we were going to help them build, like, tons of new fabs in the U.S. They don't have a third-party business, so either they'd have to find some way to attract a bunch of third-party customers, which we can talk about why they'd had problems with that in past, or, like, that money may be better off going to, like, a non-U.S.

company, like, a TSM or Samsung that has customers. So, no, I don't think that, like, because of shortages at T- SMC or something that like Intel is going to pick up a bunch of incremental revenue from that. Where Intel is benefiting right now is that PC demand is off the charts. Right. And Intel, to their credit, had a very good year last year. I wonder, and it's so horrible to say it, but I wonder like what they would have looked like if it wasn't for COVID.

My guess is PCs would not have been nearly as strong as they would. And we'll see what happens, you know, going forward. I mean, just my personally, like, I live in Los Angeles and I've got four kids and like they've all been home from school since April. I bought four notebooks last year. Wow. I'm probably not buying any notebook computers for a while. What's that like buying a notebook as a semiconductor analyst?

You know, it's like I go on every, I buy a new computer every few years. I like click a few buttons. I don't really care because all I do is tweet and record this podcast with Tracy. But like, you know, what's the computer buying process like for someone who knows as much as you? I've learned a lesson, which is to try to buy better. And I got to give Intel credit for him. I bought cheap PCs in the past. They never last.

So I bought good ones on the hope that they can use them for a while. But that's got and that's what I bought. So if you had to choose like one lasting impact of the chip shortage, like one big thing it's going to change in the industry, what would it be? Like what is the most likely outcome? Well, I think there's a few things. I mean, look, we've already talked about like what's auto going to do with inventories. I think they're going to have to hold higher inventories. I think that's fine.

The other is like, you know, again, I wonder about TSM. So, like, they've got this, like, massive Cappex guide. Is it ever going to go structurally lower? I don't know. Right? I mean, again, they were doing you, whatever it was, you know, 15 billion, 17 billion. Like, is that Cappex number ever going to not have a two in front of it? Maybe not.

Like, maybe it'll stay up here. So I think that's quite possible that Cappex level. I mean, there are reasons to think that. that semiconductor cap-x should be structurally higher going forward anyways. But, I mean, like, that is one. We're actually seeing a lot more demand. And by the way, especially if you think that, like, coming out of COVID, that semiconductors are going to be more important than they were.

And I've kind of been coming around to this. You know, I was of the belief for a while, for a long time, frankly, that semis were mature. We even did a big piece of what Bernstein calls a black book, a few years back called Playbook for Maturing Industry. and the idea was that maybe the industry has gone X growth and like in that environment like you want to invest in certain ways and I'm no longer convinced that that's necessary.

We've had a number of good growth years since then and coming out of COVID, I'll put the near term cyclical concerns aside for a minute. But if I'm looking long term, I've got to say I'm more positive on semi's longer term than I think I have been in a long time. I think we are seeing real, not the whole little bit,

but real growth, growth applications that are people driving real businesses off of, whether it's AI. you know, or whether it's, you know, again, auto, I've been an auto semi-bow for a long time. That that content story is really coming through, an autonomous driving and even 5G and IOT and all this stuff. I also think COVID has really pulled digitization forward by years.

And then we're seeing like both from a functional as well as a strategic standpoint, just given all of like the sovereign interest, how important this industry actually is, you know, to the global economy. Yeah, I'm more important. I'm more positive long term on. on semis, I think, well, then maybe I have been in a long time. And so maybe that's the third change that could be coming out of all this. And maybe not having you necessarily just with the shortages.

But I mean, we're just seeing in general, like, how important this stuff really is. And when we have a lack of them, like, how much it hurts. On April 4th, 2023, around two in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political fire. Firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16.

What separates good leaders from transformational ones? I'm Jessica Chen, and in season two of Leading By Example, we'll sit down with executives like Grace Chen of Bertie Gray to find out. It's important to understand where you spike, but also really acknowledge. where you don't and find people who can fill those gaps. Listen to leading by example, executives making an impact on the IHeart radio app, Apple podcast, or wherever you get your podcasts. Could you actually, I feel like we brushed over a little bit, the impact of the shortage on the sort of the non-auto areas. I mean, like, so like I assume there's semiconductors in refrigerators. We know there's big shortages of those, and I think there's various reasons.

I know demand is off the charts. Tracy and I recently did an episode about shipping. And so I think just getting like a, you know, washing machine or a refrigerator shipped is probably a pretty big bottleneck right now for that industry. But how much is the sourcing of chips even an issue for like things that we don't really think of as tech per se? There's chips and everything. And frankly, like, even things that didn't use to have chips are getting chips in them now. even your refrigerator, which by the way, you could argue maybe is not the right way to go. I don't really need, like, internet on my fridge, but you can buy one.

Right. But no, but I mean, semis in general. I mean, like, but is that part of like when all these, when all these customers are complaining about shortages and I can't get this and et cetera, like I don't know if it's as big of a deal as it is for cars, but is that part of the story here that people need to understand? Look, it's not as, like I said before, it's not as as much of an. issue as it is with cars because I don't think you had the supplies chain disruption that we had with cars. And I think automotive has some of these unique characteristics that made it particularly vulnerable to this. But we are seeing shortages in general just, again, just because demand is off the charts and capacity is what it is. It'll take TSM time and other foundries

time to bring more capacity online. But yeah, I mean, anything that requires like semiconductors in general is going to be tight because you've got lots of, and some of these end markets that are fighting for capacity are very big, right? And they're going to suck, suck up more. We just went through an Apple cycle. Like Apple takes up a ton of capacity. You've got the data center growth and everything else. Everything else that's going to take up wafer volume. There's only just so much to go around. So yeah, I mean, and look, over time, it will normalize, right? By definition, like over a long period of time, supply has to exactly equal demand. I mean, if we take it out to infinity, like eventually, like that, that's how it will play out.

So it will normalize. By there's other things you've got to worry about too, because in some end markets, if I don't ship the parts now, I can ship them later. But in other end markets, if I don't ship it now, that demand just vanishes.

PCs is a good example. We're seeing strong PCs right now, but in a normal year, there are product cycles. Like, if you miss back to school, like you don't ship it, right? I mean, it just doesn't happen. People don't buy two PCs later

to make up for the PC that they didn't buy, right? I think this is, you know, these are things that people are going to have to worry about. So here's a really basic question, but do prices of chips just go up or become prohibitively expensive at some point, at least for, I guess, non-essential uses? Like maybe we get fewer fridges with chips and internet capacity and things like that. You know, it really depends on the chip. Like in some markets, like we're seeing this. So, for example, some of the auto guys. they're taking up price.

By the way, they're not exactly benefit because this isn't the only shortage we're seeing. Like the semiconductor manufacturers themselves are seeing costs go up because like raw wafers are getting a shortage. So the costs are going up. And then for some of these guys that are using foundries, some of the founders are taking their quotes up. Right. And so we're seeing them.

So they are like taking a price where they need to sometimes. But in other markets, you may have like long term agreements already that stop you from doing this. Over time, though, I mean pricing adjust. there's like continued strong demand and no supply. Like, we'll see what happens. But pricing tends not, again, it depends on the end market, but like you're not going to see it like right away.

All right. I have one more question. It's very important. So you mentioned maybe we'll, we should do a memory episode. So at some point in 2021, maybe in a couple months again, we'll have you back and we'll talk memory. What is the next episode besides that?

People tell us that we should do an ASML episode. episode. Is that where we should go next with our semiconductor journey? Well, maybe. I mean, like, if you're arguing, like, what are the most important, strategically most important companies in the world? One, I think, is certainly TSM. The other you could argue was ASML, because especially leading edge manufacturing, like, it requires advanced what's called lithography. This is the process that's used to imprint the circuitry patterns on the chip. ASML is the only game in town. They're the only ones that make the tools that can actually make features that small for for leading edge logic.

And so, yeah, and it is a, it is a phenomenally interesting company. And they do, like, if there's any, like, I always, I think I said before, I think semiconductor manufacturing general can be, it's got tones of black magic to it. And those guys are the, are the ones that are, you know, those are the wizards, right? Asm, they're the ones that enable it. Like, without ASML, there's no TSM. Right, like, that's like a big part of their KAPEX. is basically of check that they write to ASML?

Yeah, I mean, you've got to think about these tools. Have you ever seen like a leading edge lithography tool? Not in person. I mean, they're, they're into, so it's this big thing. It's like the size of a room. Like the EUV tools can cost like $200 million a piece. It's like the price of like a jumbo jet, right? I mean, they're incredibly complicated machines.

Again, we could talk about like how they work and, and, and, but again, I go back to that, I'm amazed it works at all. Like, it shouldn't work. it's unbelievable, the level of engineering and science that goes into these things, just to get the tool to be able to do what it needs to do to make the chips to do to do to do what they need to do. So, yeah. I have one more question, actually, which is I've seen a number of commenters now talking

about chips as the new oil. And you can kind of see the analogy there. I wonder, like, how far does that analogy go? is it the right way to be thinking about chips, or do you have your own sort of conceptual commodities framework for how you're thinking about semiconductors right now? Yeah, I mean, look, so semis is the new oil. I mean, look, if you figure like a lot of the industry ran on oil

and shortages caused problems, I mean, semis are important, right? I mean, they're critical components of everything that's around us. And I guess in that sense, you could argue that they are the new oil. I guess one of the differences, I mean, like, I don't know how far you want to draw the analogy. Oil is a pure commodity, right? And so, like, the price of the oil is, like, is set completely by the supply demand situation and, frankly, the cost structures of the various producers.

You know, semis are probably a little more differentiated, I guess, from that standpoint. So the dynamics aren't entirely similar. But, I mean, the idea that it's like a single, if you want to call semiconductors a single hobo genius, like sort of thing, which they're not, but if you wanted to sort of look at it, at the industry that way, it is a very large, sort of very strategically important thing that right now is getting concentrated in some areas of the world that are getting increasingly risky. I think that analogy kind of holds. This is great. I love, you know, I love talking chips with you. Tracy, anything more from you? I think, Stacey, you've made it so clear that

we've run out of questions, so thank you. No, that was great, Stacey. It was awesome. It was awesome. Really appreciate you joining us. I probably will think of some questions. You know what I've always wanted to do? I've always wanted to do a two-part podcast where we, interview someone and then publish it and then let a bunch of people ask, like, send us follow-up questions to do another one. Maybe we'll do that one day with you because I bet- Do it. Look, anytime you want me to come on, I'm happy to come on. Yes, if you want to like put like a request out there for Q&A, we can just do a free-form Q&A. We'll do that in a few months. Stacey, thank you so much for coming on back on Audla. My pleasure. My pleasure. Thanks,

thanks, guys. Thanks, Stacey. That was great. I love this topic. We're going to do more. I don't know. Do you want to do more? I really want to do more. I feel like we have to do at least two more. I can tell you're gunning for that ASML episode. So that's fine. I do think it's going to come up again and again, as it has just in the past six months. So yeah, I'm in for more Chips episodes. I do think we should do a sort of Q&A with Stacey. Maybe we should get a clubhouse. Oh, we could do a clubhouse with them. Well, you've said you hate Clubhouse, but I think this is a legitimate use. just a hater, but you know what? Or what do we do like a, I think we should do, we should do like one of our streaming episodes with Stacey and then take people's questions. But you know, the other, the thing is that I really like, I mean, obviously, he does a good job

of explaining how there's like two proximate causes for the acute shortage. And one is the sort of distinct situation that happened with autos where they canceled a bunch of orders and then scrambled when they realized that auto demand was off the charts. And then just the fact that. The demand for everything else was off the charts. I mean, hearing him describe buying four netbooks or tablets or whatever was for his kids and everyone buying, you know, new monitors and new computers at home and 5G and AI and video games and PS5s and all that stuff.

So you can really see there is essentially like two things that came together at once. They're just totally overloading the system right now. Yeah. I mean, this, I think, gets back to the uniqueness of the downturn that we just saw, right? Everyone was sort of treating it like a normal recession. Yeah. But it really isn't a normal recession. And we saw consumer demand in many ways was much, much stronger than people might have expected.

Well, you know something I was thinking about early on, and this should have been my clue that we were going to get this sort of more robust recovery than we realized. I thought to myself, it's like, man, there's never been an economic downturn. that was so expensive. Like, because everyone had just had to go out and buy a bunch of stuff, right? So you go out a few episodes. I talked about how I was like, went out and bought a bunch of sardines, people bought toilet paper, people bought, you know, kiddie pools for their backyard so that their kids wouldn't go crazy.

Computers, like, we would never have recessions if every recession required people to spend a bunch of money. But this one, it was like we had this huge downturn. It was very expensive. And so it's like the stabilizing mechanism, of more spending kicked in in a way automatically, of course, the fiscal stimulus helped too. But obviously the tech industry and chips in particular, one of the huge beneficiaries of that sort of for spending that lots of people had to do right away.

Yeah, that's a good way of looking at it. I guess you had like a lot of displaced services that you could no longer go to, right? So you couldn't go to restaurants, you couldn't go to schools, you couldn't go to your workplace. And so you kind of had to recreate a lot of those at your home if you had to have. have the means to do so obviously. And basically you had a sort of like double demand for goods that would enable you to do that. Pretty extraordinary. And then of course, you know, obviously beyond the acute thing, it really does seem like the sort of supply constraints as judged by that huge Taiwan semi-CAPEX forecast for the coming year. Like that part

isn't going away. So even if, okay, maybe in a couple of quarters or whatever. whatever, the car cycle is back and everything. And maybe some of this one-off demand, you know, I don't think Stacey's going to be buying for new netbooks for his kids this year. So some of that one-off demand will fade. It really doesn't seem like that's going, the broader story of like how much demand there is for chips overall, it is going to go away anytime soon. And so we still are going to be left with as issues of the world extremely reliant on a

couple of manufacturers in both a physical and a geopolitical hotspot. That's your way of saying that we're going to be recording semiconductor episodes for the next 10 years. Yeah, that is basically that. Okay. All right. Shall we leave it there? Yeah, I'll leave it there.

This has been another episode of the All Thoughts podcast. I'm Tracy Allo. You can follow me on Twitter at Tracy Allo. And I'm Joe Wisenthal. You can follow me on Twitter at the stalwart. Follow Stacey on Twitter at S. Razgan. Follow our producer Laura Carlson. She's at Laura M. Carlson.

Follow the Bloomberg head of podcast, Francesca Levy, at Francesca Today. And check out all of our podcasts at Bloomberg under the handle at podcasts. Thanks for listening. A lot of short daily news podcasts focus on just one story. But right now, you probably need more. On Up First from NPR, we bring you three of the world's top headlines every day in under 15 minutes because no one's story can capture all that's happening in this big, crazy world of ours on any given morning. Listen now to the Up First podcast from NPR.

What separates good leaders from transformational ones? I'm Jessica Chen and in season two of leading by example, we'll sit down with executives like Grace Chen of Bertie Gray to find out. It's important to understand where you spike, but also really, really, acknowledge where you don't and find people who can fill those gaps. Listen to leading by example executives making an impact on the IHeart radio app, Apple Podcast, or wherever you get your podcasts.