Mar 29, 2021

The Semiconductor Shortage: How It Will Cost Billions to Solve

Show Notes

In the last episode, or Blitz Talk 31, we talked about how the semiconductor Shortage happened and how important it is to society.

Why can’t we just scale up production and meet demand? Well, we think it might be a supply chain problem. There are only a few players in the manufacturing of chips which means that the supply chains are NOT diverse.

As one use-case, Apple has one supplier for its semiconductor consumption and that is TSMC, or Taiwan Semiconductor Manufacturing Company. It is the worlds largest semiconductor foundry.

With the disruption that the pandemic has provided, we are realizing how important these chips really are. Countries around the world are taking it more seriously and investing more in creating their own supply chains, like China, and recently the United States.

Join us as we parse through the supply problem that chip makers and chip consumers and why the solution might be public-private partnerships.


News [00:00:01] The global computer chip shortage is impacting industries from medical devices to military equipment in autos, General Motors saying could take a two billion dollar hit in 2021 thanks to the semiconductor shortage. It announced three temporary plant shutdowns as assembly lines wait for more chips 

Jed Tabernero [00:00:20] last episode or Blitz Talk 31. We talked about how the semiconductor shortage happened and how important they are in our modern day society. 

News [00:00:30] You know, we certainly see GM shedding more production lines down this week. But if we can't get critical chips, we need for fighter jets, military applications, artificial intelligence applications that the economy is increasingly running on. It's a big problem. We are transitioning to a digital economy. And if we don't have these chips, we're in trouble. 

Jed Tabernero [00:00:48] But why can't we just scale up production and meet demand? Well, we think it might be a supply chain problem. There are only a few players in the space of manufacturing chips, namely Intel and TSMC. This means that supply chains are not diverse 

News [00:01:11] and these companies have a lot to lose. Apple relies almost exclusively on semiconductors made in one spot in Taiwan. And TSMC is an incredible company. But if something were to happen there, there's effectively we could possibly go years without Apple being able to ship iPhones or iPads. 

Jed Tabernero [00:01:28] And so with the disruption that the pandemic has provided, we're realizing how important these chips really are. Join us as we pass through the supply problem the chip makers and chip consumers are going through and why the solution might be public private partnerships. Welcome to TLC, where we unpack the ever changing technology economy 

Adrian Grobelny [00:02:03] hangout with Jed Shikher and Adrian as we tackle the industries of tomorrow. 

Shikher Bhandary [00:02:09] This is things have changed. So last time we really covered a whole lot about chips and semiconductors and why it's important how it runs our lives, how it's in everything we do and interface with in our world today. Right. And specifically, we kind of called out, though, the shortage that car manufacturers. Had to face and what happened was corporate hits. Car demand really low car manufacturers tell the chip companies, hey, we don't need your chips. They cut the amount that was that they would need to source from these chip manufacturers. Now, when they do that, there's since so many people went online, these chip companies were like, you know, rubbing their hands and being like, yes, you know what? We can give this this the amount that the car manufacturers said they don't need. We can give it to someone else, like an Apple or a Sony or whatever. Now, six months into covid, suddenly people want to get out of their houses, they want to travel, they want they don't want to use public transportation, they like, you know what? I need a car. And car manufacturers are like, OK, we need the chips to make the cars because the cars are so advanced these days, everything is controlled by an electronic system from your braking to your acceleration, steering, audio, stereo or whatever. And so they were in this tricky situation where they had canceled orders, but now they need orders like right away. So what happened? Like, just a recent like this this was like just three days ago where GM, you know, there are there were reports that they shut production for three weeks, four weeks, cost to them three billion dollars and above and cost a lot of people jobs, which is terrible. You know what happened? And now they come out and say that because they do not have the chips, the upcoming GM trucks do not are not as fuel efficient, they'll pollute more because they don't have a specific fuel module. That depends on the chips. So my question to you guys is, what else is missing in those GM trucks, 

Adrian Grobelny [00:04:55] all the smog sensors for pollution? 

Shikher Bhandary [00:04:58] So what, if you, like, put the indicator left and you just you're right. Imagine that. Geez, we spoke about the Demand-side last time. And yes, the manufacturers did not plan that well, though. There's a lot of management planning, sourcing, supply chain, all of that that gets impacted here. 

Adrian Grobelny [00:05:18] You know, you would think that throughout this whole technological expansion, you know, we're living in a software era where companies are fully taking advantage of software development and the the things the new things that you can use software to do. I don't think that when you look at the last 20 years and how much investment and capital expenditure has been put into chip manufacturing, it's been lagging when you compare it to the early 90s and 80s when really that's when they're pushing for better hardware, faster computers, trying to get smaller and smaller, and looking at the capital expenditure of these chip manufacturers. It's been lackluster from I think it was 20 or 20 tents. And that kind of created this issue of they're not able to scale, they're not able to have the infrastructure to create these chips. And also with just a huge jump in demand, all of their capacities are maximized. And these factories, they can't be built overnight. It takes so much, so many years to build the infrastructure. And so the combination of that not being in place has GM cutting back on the chips that they're able to make and put in their cars. 

Jed Tabernero [00:06:40] I mean, you know, if we think about the demand for these chips, for certain products, right. For example, like iPhone, you know, Sony's PlayStation and other game consoles and whatnot, some of them will buy exclusively from literally one company, like we'll buy all their chips from one company. Their supply chains are not diverse. Like, it was interesting because I was reading a paper called or a paper from a publication called Asian Commentary. And it's this dude from Taiwan that studies supply chain in Southeast Asia. And obviously his main topic since he's in Taiwan is semiconductors. Right. It's so important for Taiwan as a country. Right. The the boost that they've experienced from the 50s to become a tiger economy was all reliant on manufacturing chips. Right. So it's interesting to see some companies like the iPhone will exclusively by semiconductors from one manufacturer. Now, think about how crazy that is. Right. One manufacturer can produce, you know, semiconductors and whatnot and and fulfill the demand of these companies. But if you think about the iPhone, the iPhone, it makes billions of dollars of sales a year. You know, like that's a ridiculous product in itself. And the chips come from one spot, right. TSMC now, you know that that supply chain risk is immense. And it's fascinating for me to see that these massive companies would really rely on supply chains. Obviously, there are positive sides to having one supplier for your products. But at the same time, you know, the supply chain risk of non diversity. When we were talking to our supply chain expert here to see, you know, that was one of the important factors of of having a safe supply chain is having a diverse not having come from one place. 

Shikher Bhandary [00:08:28] There are two issues here. The amount of capital needed to build something like this and the complexity. And you can add in time, too, which is relevant for GM. So maybe we can start with capital and time it costs to build a fab just off and having it functional. Right. It takes. About four billion dollars to set it up the way these chips, they're so. Immensely complex that it takes a lot of time to get the fab to produce chips that can be put into cars. It takes a lot of time like these trips are now smaller than the width of your one hair. Right. You're talking about like seven nanometers, 10 nanometers, five, nine. 

Jed Tabernero [00:09:19] I don't have any hair, so I don't know what the context is here. What is one? One. 

Shikher Bhandary [00:09:24] There you have it on here on your chin. You have a decent interest anyway or the other. Yeah. So, I mean, say you you pay four billion dollars to set up a fab. Right. You take it takes two years to make the chips need it. Like to start producing good enough chips. It takes one more to get capacity up there and it takes a year or two to get the yield up to a point where you can fulfill demand, you can provide to many automakers. So you're looking at five billion up front, which scales up to 20 billion dollars. And five years. Duration for you to go from no fab to fab that produces chips that manufacture our car companies need. It's five years and GM needs it today. 

Jed Tabernero [00:10:22] They did it with vaccines, right? Typically takes a couple of years to make vaccines. They sped up that process. Can't we have an operation? Warp speed for semiconductors, bro? 

Shikher Bhandary [00:10:31] You have 40 billion, 60 billion. Hundred billion. Given Apple, Amazon can't do that. 

Jed Tabernero [00:10:39] But the US government just spent a one point nine trillion dollars on a big machine. There's something in there for semiconductors, which is, by the way, a part of his administration, our new president's administration. 

Shikher Bhandary [00:10:50] Exactly. Because they're like, OK, we are severely lagging over how embarrassing it is for a company like GM to say, hey, we have a module that helps your mileage, but we can't put it because we didn't get the chips for it. Like, that's actually quite embarrassing for for a billion dollar Fortune 100 company. 

Adrian Grobelny [00:11:14] I mean, you look at all the recalls that they have, though, they come out with recalls all the time. But you know, something I wanted to touch on with why it takes so long for these fabs to be built. I mean, just think about how reliable chips have become. You know, we used to have chips and those Pedders, those early BlackBerry phones, just early consumer technology that we would use. And, you know, last year, the batteries would die out really fast and he'd kind of recycle it or whatever. But chips nowadays, I mean, sugar, you can talk better on this, but nowadays they they can last 50 years like the amount of reliability and the robustness of them. It's unparalleled. It's never been this good. And they're so reliable. Like like you were saying in our conversation earlier, if there's just one little imperfection on the on the waffle waffle, the wafer awful on the wafer, just one little imperfection, it's scratched. So the amount of technology and building these manufacturing lines to develop this has to be perfect. Like, I would assume that these these fabrications are literally like those, you know, white glove factories where, you know, there's no dust, everything's ventilated. 

Shikher Bhandary [00:12:30] It is. I wish I could take you in because it is sort of like I remember like so I came from Abbott, which is a medical device space. And, you know, you have technicians building the pacemaker. So they are taking off with their hand. They're putting stuff in. And, you know, I see that. And then I come here and it's like there are like robots picking stuff up and zapping. And I like 

Adrian Grobelny [00:12:53] everyone's wearing like 

Shikher Bhandary [00:12:55] there's like body suits and it's like 

Adrian Grobelny [00:12:58] gloves, booties on their shoes. No contamination, nothing. 

Jed Tabernero [00:13:02] Shit is real, bro. Like they have an air filter for you when you come in and an air shower. I was like, what the hell's an air shower, dude? There's an air shower just to go in the chips. 

Shikher Bhandary [00:13:12] They, they come they made the first raw material is silicone so that you can think of like sand. Right. You're taking sand and giving it electrical properties at the end of it. So how many steps does it go through? Like it's around 270. So 270 processes that takes sand, a layer of sand and makes it into a chip that powers the global economy. Right. Right. Now, each of that process. Even if each of the process had 99 percent efficiencies, each process at the end, you still get 65 percent yield because 99 percent or scaled across 270 processors just gives you a 66 percent at the end. So you just saw less margin of error that it's so complex that, you know, you really need to have, like, your stars aligned. You need to have a lot of things going your way to ensure the technology reaches the public. You cannot just get five people in a room like how you can build for a software company, build an app and get, you know, if your stars align. Right, like Instagram with 13 people, 20 people, you've made a billion dollar exit to Facebook. You know, that does not exist in something like the space because you need to prove that your yields can scale, like you can pump a million chips and it's going to have good performance. So you can't really do it with just a small team. So you need a lot of investment. You need a lot of. Inherent knowledge. And that's the tough one, you're as well where companies the reason why there are only two or three manufacturers of chips I'm not talking about design companies goes like your AMD, your Apple, Nvidia, all those are design. They design the chips and then they send it out to TSMC, Samsung, Intel to manufacture it. There's only a few that can manufacture it because that knowledge is like gained for like 60 odd years. 

Jed Tabernero [00:15:34] Dude, I mean, the issue there I think is how much attention that goes into just, you know, even the field of lithography and material science. Right. Like, it's still a niche field. And at the same time, like you look up, you know, courses for LITHOGRAPHER or even material science, like it's very limited to certain areas, material science. There's a lot more things to do with Michelle science. But lithography seems to be a very profitable place. So maybe like a brain drain is occurring in which like there's a very small amount of people studying lithography. And then we can think about how many of them are working on semiconductors, you know, and how many of them go straight into TSMC or, you know, ASML. Right. Which is another company that provides the ITVS for TSMC and whatnot. But, you know, this type of technology, you can see there is a monopoly of sorts in this area. You know, it seems that there's only a couple of companies that lead in the field of creating the semiconductor start to finish. Maybe it's education as well for the future that we need to focus on. 

Adrian Grobelny [00:16:37] So what are we going to see in the next coming years? Are we going to see more diversification in the supply side of companies and where they get their chips? Are we going to see more research and development into the materials and the composition of how these chips are made, how to better make them efficient, more efficient means of, you know, cutting down the 270 processes that need to be done? And to that, what can we expect to see from electronic and chip intensive products that are really struggling right now? 

Jed Tabernero [00:17:11] So before Chika goes on with what the future is here, because he's really is in the ground with this space, let me just share something that I recently heard of on Bloomberg. And it was the assembly's CEO, right? He was being asked a couple of days ago what's what has changed during the span of the past 12 months, you know, in the semiconductor space for him for for a small company. Right. And, you know, they were saying like we're seeing the shortage happening in different markets for semiconductors. What are you know, are you there to fill that demand? And he said, you know, nothing's changed much for me. My orders are going. It's the demands. He is excited. He's literally like they asked him, is there upside risk? He's like upside. He's like there's upside to this. He's like the necessary risk. 

Shikher Bhandary [00:18:02] That's not a risk. 

Jed Tabernero [00:18:02] Well, the thing is, like at this point, the upside risk here and this is interesting because I looked it up after hearing that comment, I was like, what the hell? Why would there be an upside risk? Well, if you can't provide that demand with your products, other supply chains will be built. For example, TSMC, that company that manufactures, you know, most of everybody's freaking semiconductors on this planet. They will look elsewhere to look for ways to create fabs or any sort of, you know, whatever the next wave of technologies will be to create these semiconductors with different suppliers. That could be a chance that they miss the opportunity to manufacture this for the companies. Now, I don't know what the intricacies are of their their non competes with each other. They need to fulfill that demand. I think one thing that we can expect from the future is that there will be new supply chains built from this process because, number one, when the government gets in on stuff like this, usually the way that they'll get into it is to fund research and development, number one. Number two is they fund those companies to start off manufacturing, whether that be in a form of a loan or literally money from the government that will come for free to create these companies. So fulfilling that demand will happen in the next few years. And I don't see that demand going anywhere else 

Shikher Bhandary [00:19:18] in the future. You know, they will have advancements. But if today Apple decides, OK, we need a fab, they're not building it, they're going to give 100 billion to TSMC to go build a up. That's the only way to get around this, because the chip manufacturers, they want like they are not going to buy a build one like a fifty billion dollar fab and then not see any demand. Right. They need to be selling. So that's one thing. And what's happened is it's become such a geopolitical we didn't want to get kind of get into it, but now it starts playing right. Because TSMC is Taiwan and it's right next to China. And so they are the center of the global race for getting chips right to small country. So there's a lot of factors that now play into this that, OK, we need to be good to them, we need to provide them or we're to maybe set up fabs in the US, which they actually just got greenlit. TSMC got greenlit to build a fab in Arizona, very close to Intel's, because they were, I guess, want to board some of our folks, but builds build something around here. And but even then they got greenlit. But that project is going to complete in twenty, twenty eight. 

Adrian Grobelny [00:20:46] And you're mentioning how they're considering building these Fab Four. Fifty billion dollars. Like just imagine they build this 50 billion dollar fact fabrication. It's ready in 2020 and then a new technology just comes out with all this. 

Shikher Bhandary [00:21:00] So that's the big problem because you never know when that breakthrough happens. And then you've spent all that money setting up tools that are on a previous technology and then you need to start from ground zero. So you need to know what Apple AMD wants. You need to get that in writing. You will purchase from us and only then you can go. That's why, though, the supply side hasn't expanded a whole lot, because that demand hasn't really expanded. It's more or less met now. Suddenly when demand spikes, it's like, bro, we can't build, we can build. But in six years, 

Jed Tabernero [00:21:36] I think if there requires a lot of capex and supply and demand shocks happen by nature, these industries will get disrupted in a way that will not see in other low capex industries in which they'll have a hard time deciding whether the supply chain should be met right now or wait for the next big technology. Right. And so the scaling piece of this is very difficult to find. But again, if we have more minds thinking about this, you know, more people in the process of creating a semiconductor and maybe if it becomes more open to society and people understand what its contribution to society is, maybe there will be more people who wake up and go to college thinking, oh, my God, I want to be a lithographer or whatever the fuck they're called. Right. So that might be the next step. 

Shikher Bhandary [00:22:23] It's good that it's forcing the US government to kind of act because it's like, OK, this is proprietary technology that you have to support. The big problem with technologies like this is if you don't keep improving, you are more or less dead in the sense if you don't keep putting out a better product, that is moving society further as you are feeling. So there needs to be like support. Government support, tax subsidies or something like that that I think they just announced Biden came out and spoke about how the shortages are affecting everything. So maybe there's more legislation that comes forward that helps these chip manufacturers to kind of take some of the costs of take some of the risk off the table because you're building a 50 billion dollar farm and next year there's a new technology like you need to be able to protect some of that. So I think that is a good direction to go. But with regards to technology, like I have an iPhone six, OK, you can laugh. You can Mark. But I got mugged last night, so that's why I'm saying it's a five year old iPhone, but even today that iPhone six has more computing power than the computer that landed Apollo 11 on the moon. So that's how crazy things are, right? So. It's every all of us have technologies that were unthinkable, like 20, 30 years ago. So what happens in the next 20, 30 years? Don't know. Maybe a genius wakes up and thinks we can. Get past Moore's Law and Moore's Law is basically Gordon Moore, who's the founder of Intel, came up with this thing that this principle where he's like your power of computing doubles every so many years, every two years. So from 1960 to 2000, our computing power and ability has doubled every two to four years. And that's why the phone that I have iPhone success has more computing than what landed the rocket on the moon. But at a point we mentioned this, where you run into a physics problem, where the chip is producing so much power that the heat created by the chip is burning the chip while it's giving you the power. So that's the physics like. I'm really simplifying it. But you run into physics. So how do you, like, get past it? What do you do? What is the what is the next move? And that's the next trillion dollar idea. Guess. Hey, thanks so much for listening to our show this week. You could subscribe to us and if you're feeling generous, well, you could even leave us a review. Trust me, it goes a long, long way. You could also follow @THC_POD on Twitter and LinkedIn. This is things have changed.