How serious is the chip shortage? What do you believe are the most important factors of the global chip shortage? Some experts attribute several factors (U.S.-China Trade War, Taiwan’s droughts, etc), while others put most of the blame the COVID-19 pandemic. Are there any other underlying factors that the mainstream media fails to attribute?
If you look at the losses and disruptions in production of various firms, it's pretty serious. It's a combination of factors but the most important factor is decades in the making with the emphasis on having very lean supply chains and low inventory. But when supply gets disrupted, you're in trouble. Once you have this lean inventory, global supply chains, and then you get hit with COVID. COVID keeps coming back as a disrupter by doing things like disrupting port facilities, particularly in China, where they can be quite draconian. That becomes quite problematic. The shortage is also dependent on the supply chain. For the relatively low end of the supply chain, you saw that auto manufacturers cut back because they figured people weren't going to buy cars, which was a correct assumption. But when places started to recover from COVID, people were wanting to buy cars, and manufactures didn’t have any inventory. During the lockdown in various places, or quasi loc k down, people were buying other items, such as freezers and washing machines because people were spending a lot of time at home. Those often use a lot of the same controllers, or at least the same kind of mature manufacturing processes for chips, that the controllers that go into automobiles also use, so all of that capacity was taken up by white goods. Now we have a huge backlog on the auto side. But really, many people would argue you wouldn't have that if there wasn't such an emphasis on manufacturing where there is very little room for error. The U.S.-China trade war is pretty marginal. The biggest front ordering was done by Huawei and that was for the more advanced processes. Once they stopped being able to use TSMC, that actually created a lot of room for other suppliers, so I would put that further back. Taiwan's drought. I mean, that is a constraint but it's not like that fire at the Japanese plant with a dramatic shutdown of production. I would say without COVID and all the rational reactions by corporations to COVID, you wouldn't have this ongoing rolling supply chain issue that these other these other problems aren't sufficient to explain.
Can you specifically discuss the repercussion of the U.S.-China tech war and its impact on the chip shortage? The semi-conductor industry is capacity-constrained and global. Any disruption due to geopolitical shocks could cause serious imbalances. How did America’s tech sanctions and China’s reaction affect the supply chains?
I've already started discussing the U.S.-China tech war. It seems like the Biden administration is taking a pretty moderate course on export controls to China. I don't think this is a major issue for the chip shortage. Particularly since the controls that were in place were aiming for more high-end manufacturing and a lot of the supply disruptions are of lower, more mature production nodes. There are a lot of Chinese firms on the entity list, but the vast majority of them aren't large consumers of chips. Huawei is the big exception, so it's a huge constraint on Huawei, but it may even have the opposite impact on the rest of the companies that want seven nanometer and below technologies. China would obviously like to de-Americanize its production chains for chips, but it's a very long-term process. And in the short term, they still want to build up their capacities and the U.S, controls are not going to affect most of them, so most of that is going to be for not cutting-edge EUV lithography capacity, so China can still build that up. You also see this among all the players with some capacity in semiconductors. Besides China, Taiwan, Korea, Japan, the U.S., and EU are competing and committing to more capacity build up, so we're going to see maybe overcapacity in the not-too-distant future. Looking at the chip shortage, although it is very important for a lot of companies right now, isn’t a helpful way to understand these other issues like the U.S.-China tech war. It seems to the extent that new controls have been rolled out against China, they typically are aimed at firms that are close to the Chinese military industrial complex. And for Huawei, there has been some licenses that have been issued, other licenses that have ended, but in certain areas Huawei is still able to purchase chips, just not in 5G infrastructure or equipment.
The semiconductor industry is currently dominated by large companies like TSMC, Samsung, Intel, etc. Is this type of oligopolistic structure responsible for the shortage? What are the prospects of significantly changing this structure?
Again, I would probably say no. Much of the shortages are in products that Intel, and Samsung, but even TSMC, aren’t that interested in producing because they're more mature products. The more advanced the process node, the smaller this oligopoly. The more mature, there a lot more players involved. That's not really the problem with this shortage. Is this oligopoly going to change? Maybe at the most advanced nodes, since there are now basically only two mass producers. There is going to be at least three with entry of Intel, so it will be slightly more competitive. Other things are happening as well in the industry, which has arguably been maturing for some time. How this plays out over the next 10 years is a little bit unclear. If this industry becomes a stereotypical mature industry with slow technical change, you will typically see then more emphasis on operational efficiency when that happens. All you will need is financial capital and efficient operations. So, you might just have a small oligopoly. However, I mean, maybe traditional Moore's law is dying or dead, but they are thinking of other ways to innovate going forward, at the chip level innovation or stacking of the chips. And there you will have lots of tech innovation so you're going to need very significant resources, both financial and human resources, so you could still see a fair amount of fabrication or fabrication-plus-packaging innovation among these larger firms. The barriers to entry knowledge wise are going to be steep, so I don't see too much disruption at the top. That’s why even if there weren't export controls in China, it would be pretty hard for China to compete. SMIC has always been at best a fast follower, so even if they were allowed to import seven nanometer and mythography equipment, they might find it hard to compete. They’ve never been as efficient as TSMC, and frankly there is a lot more state interference in the running of SMIC. All these factors don't bode too well. You also have to remember, outside of China, other firms are also lagging. Global Foundries and UMC in Taiwan also are lagging further and further behind, and that's probably going to continue. There's probably going to be a bifurcation where you have these firms concentrating on older process technologies that look more and more like a traditional, mature tech industry where it's just about capital investment and managing it well, and then you have the cutting-edge ones. Maybe Global Foundries and UMC can invest enough to catch up, but that's unclear.
China has launched an ambitious program to achieve semiconductor self-sufficiency. What are the odds of success? If China is half as successful as it hopes, would that structurally alter the supply-demand equation in the chip industry?
Self-sufficient? Nobody is self-sufficient in this industry. If any of the major players just cut themselves off from the world they would not be self-sufficient immediately. If they really wanted to do this, the countries with the knowledge base to do it would probably be the U.S. and Japan, but it would be so costly. Even if they do not pay attention to IP and try to reverse engineer anything, they would find it pretty steep going. The cost involved would be huge. China doesn't really have much of a semiconductor capital equipment base. There's a lot of hype coming out of China about how they're building that up, but that's really hard to do. For the last 40 years, this has basically been dominated by the U.S. and Japan, with the exception of ASML in the Netherlands. Taiwan has tried to do this, Korea has tried to do this, but they haven't really succeeded. China’s argument is we're bigger than those places we can do it. Even if they were to succeed, this is not measured in years but more like decades. It’s important to remember in theory, you can make advanced chips without a EUV lithography, using e-beam technology, but that's very small batch. You're just not going to have any efficiency. If there's some structural break, China would be able to make a small amount of advanced chips. It just won't be efficient. But if you're, let's say involved in war or something like that, maybe you don't care that much about efficiency, but it's not viable from any commercial perspective. Even China's hopes to curb this chip deficit, they can curb it a bit, but I don't think they'll bring it to zero, especially when they're counting on all this chip consumption that's basically assembly platforms to reexport because just China consumes so many chips if you count the chips that go back into exports. That's a big hurdle. It's getting worse for them to find other countries which will accept relative capacity share loss and basically every new country that's entered this market, somebody had to adjust. Korea forced Japan to adjust because Japan is so big in memory chips and Korea’s entry fell quite heavily on Japan and on the U.S. Taiwan’s entry affected a lot of U.S. firms, but they didn't care. They decided to go fab-light or get rid of their fabs, so from the firm level perspective, it didn't matter to them. It might have not been great for the U.S. in terms of its own overall capacity, but that was the story there. China has built up ten times its original capacity, at least during the 21st century, so it's grown from a new player to a very large player quite quickly. Going forward, there's basically no appetite to allow further adjustment and that's why you see the U.S., Europe, Japan, deciding that they're going invest in capacity. Taiwan and Korea, particularly Taiwan, it's such an important industry for them, there's no way they're going just stop investment. Unfortunately, given the behavior of other countries, you can't be a viable player in this game if you don't subsidize production. When the U.S., Japan, and Europe were not proactive in subsidizing, their capacity shrank vis-a-vis these other players. Going forward, there's going to be less of that. Who does China become attractive for? Basically, domestic ones. You're probably not going to see Samsung, TSMC, or other firms investing more capacity in China because if they can get subsidies in their homelands or they can get subsidies to go to these other places. Frankly, China is politically risky, not just because of U.S. decoupling but for a bunch of reasons. The THAAD spat with Korea has really soured Korean views on China. Taiwan has always been an issue, so I actually believe you're not going to see the same speed of relative gains by China going forward compared to the last 20 years, just in terms of global capacity gains. But having said that, there’s tensions here because there's just more and more demand for chips and more in certain sectors like automobiles, so there's an argument that actually all this new capacity will eventually be justified. But there are worries. This is a sector where there have been, despite consumption going upwards and onwards most of the time periods, periods of just trying to work through overcapacity. And China's other issue is a human resources issue, particularly in memory. They don't really have experience in this. They have a lot more experience in logic fabrication. So where are they going to get this experience? It’s been pretty hard to headhunt people from Korea. They've had more success in Japan and some extent Taiwan, but Taiwan is not that great of a memory player. But those people are also expensive. So the ramping up of human resources is also not necessarily going to be easy for them.
How are businesses (especially in the automotive business) coping with disruptions in the chips supply chains?
I would say they're not coping. You have automobile plants that you don't want to sit idle. You're still churning out automobiles that are missing a couple of key chip components. I mean, that's one solution. Is it a great solution? How well do these automobiles actually keep sitting out on lots? I don't fully know the answer to that. There’s just everyone's trying to game everyone else. They're trying to over order, so they can get more access, even though that's socially a bad thing to do, because everyone else loses. The solution is to not have such lean supply chains. While that's true, it doesn't help anybody now. And once the pandemic is really over, firms will revert to these lean supply chains. There probably won’t be any lessons learned. There are too many incentives for that, keeping inventory is expensive. The whole lesson of the Japanese lean manufacturing model was to have very lean supply chains. Of course, initially, those were also very geographically clustered, so maybe less prone to disruption, unless there were some specific local events. But now, you know, that's no longer the case. But I don't think there's great coping for this. Before white goods companies were benefiting. Now they're probably also suffering since other people are putting in orders.
When do you estimate the chip shortage will end? What needs to happen to end the supply crunch in the short term?
When will the chip shortage end? It's hard to end in the short term because if you're supplying the auto manufacturers, you're probably in the short term, not supplying white goods or some other industry that wants to use the same manufacturing technology sharing technologies and, in some cases, the same chips. If you already have an empty shell of fab, meaning you just need to bring in equipment, and then you need to test it, a super-fast time would be, nine months to a year. And that's assuming you can actually get hold of the equipment you need. On the high-end, there's a huge backlog at ASML, for example. Then again, I see most of the chip shortage not at the high-end. So, it's not like chip manufacturers can immediately react to shortages and then supply them. That's why this is going on. Remember once you have your fab up and running, then you need another two months to manufacture the chip, then it has to be packaged. That's why there's not going to be a short-term solution. And frankly, there's a lot of packaging in China. If you have port shutting down in China, or elsewhere, such as in LA or anywhere else, it's easy to keep these shortages going for one reason or another. Upstream material inputs for chips are likely to be shipped whereas high-end chips likely to be flown]. Also, if you're making these mature technology chips, you have to think, well, I have to pay off this fab over five or ten years, how long is this shortage going last? Right? Is it really rational for me to invest in a new fab? Rather than just running the fabs I have at maximum capacity? Probably in many instances, it's more rational to run them in that maximum capacity, rather than investing in new fabs. Those are the issues. In the past you’d have the design firms invest with the fabs to make new facilities, new fabs, and then they get a share of the production. That's never seems to be something that's lasted over the long term. Maybe if you're very large supplier, a consumer of auto chips, let's say Toyota, it might be rational to do something like that, but I haven't seen new bold moves to do that thus far. Also, this recovery has been bumpy. If you're looking at the U.S. market, every time COVID surges, the auto market is going to go down. Until we get really a full handle on this, it’s not really clear whether addressing these short-term catastrophes makes longer term sense is the issue for a lot of these firms. You have to put this in the context that there's a lot of non-chip shortages going on in these economies for COVID-related reasons, so it's not a chip-only shortage. Chips are what are affecting a few prominent industries that are consumer-facing, but there's all sorts of shortages going on basically because these global supply chains are so disrupted, and people don't have inventories.
info2072, CC BY-SA 3.0 <https://creativecommons.org/licenses/by-sa/3.0>, via Wikimedia Commons