The Revolutions and Repercussions of the Semiconductor Industry

In this episode of Advantest Talks Semi we demystify the complex interplay of semiconductors and the automotive sector. Buckle up as we navigate this intricate landscape with Malcolm Penn, the CEO of Future Horizons. 

From deep-diving into the cyclical nature of the semiconductor industry to early signs of industry recovery, Malcolm illuminates the market's recent movements. He spotlights the explosive growth of China's EV market and its strategic grip on the low-end market - a move that's not to be underestimated by the established auto industries globally. 

In the second half, Malcolm walks us through the intersection of the EV market and the rising demand for advanced semiconductors. We examine the reverberating effects of sanctions on the industry while underlining the growing importance of decentralized computing. 

Toward the end of this podcast, catch Malcolm's take on the current OpenAI media buzz and how AI chipsets are leaving an indelible impression on memory. We wrap the discussion with a look at how Avantest is riding the wave of innovation with its open solutions ecosystem Advantest ACS. 

Download this episode now for your deep dive into the dynamic world of semiconductors.

Thanks for tuning in to "Advantest Talks Semi"!

If you enjoyed this episode, we'd love to hear from you! Please take a moment to leave a rating on Apple Podcast. Your feedback helps us improve and reach new listeners.

Don't forget to subscribe and share with your friends. We appreciate your support!

Transcript

Keith: 0:29

Hello everyone and welcome back to Advantest Talks Semi. In today's episode, we're fortunate to have Malcolm Penn from Future Horizons back on the podcast. For those of you who tuned in earlier this year, Malcolm shared his insights on the cyclical nature of the semiconductor market and how we were entering a period of deceleration and contraction. Fast forward to now, eight months later, we're revisiting this conversation to see where we stand and what changes have occurred. Malcolm, it's great to have you back.

Malcolm: 1:04

It's my pleasure.

Keith: 1:05

Could you give us an update on where the semiconductor industry was eight months ago, what transpired during this time and where we find ourselves now?

Malcolm: 1:14

I think eight months ago, in January, we were just beginning to go into the downturn, which we advised was going to happen. And it was then showing us first signs that we were in a retracting type mode. And once this cycle downturn starts, it typically takes four quarters to navigate through before you make the adjustments that have to be done, which tends to be certain things that we're in control of. Like the shortages created and excess sales, because we had to build out inventory and lead times were extending, so we had to overcompensate for that. You have to rain back capacity, you have to manage the average selling price, all of those practical issues which cause, when they accumulate together, the market contraction.

Malcolm: 2:09

So, we were in full flood back in January there and now roll ahead eight months' time. We're just about coming through that part of the cycle. So the latest data, which was published at the beginning of this month, actually shows that the industry has now reached the bottom and showing signs of starting to readdress itself and the beginnings of the turning up of the next cycle, which is quite normal. It was right on cue almost to the month. So, I think we are in a classical cycle here. It's continued along that vein and fortunately now, eight months further on, we now are starting to see those early green shoots of the recovery.

Keith: 2:51

Building on that a bit, how did the current trends compare to the previous ones? If we're in this classical cyclical state, are the established patterns still holding steady, or are ther any noticeable shifts? What should we be paying attention to?

Malcolm: 3:07

I think at the macro level the patterns are just rolling out exactly as you would have expected them to, in synchronization with previous downturns and cycles. There's been no surprises. It's really been a textbook example of what the adjustment to the shortage period, that we had a couple of years ago, and the recovery from that, the overinvestment, it's pretty much running as you would have expected. The only surprises come in the detail, which is quite normal as well. Because not every sector, not every market, not every product behaves in exactly the same way, and that's because the industry is so diverse that there is always opportunity for sectors which are better protected, better insulated. Maybe they're high growth sectors which are just starting. And their growth trajectory, and anything on the very latest technologies where they're not in oversupplied, they're always in shortage when you introduce new technologies. So, there are always these exceptions, and they vary from cycle to cycle, but in general the pattern is very much aligned with what what we would have expected.

Malcolm: 4:23

I think the big surprise in this down cycle is Europe has been the star. I mean you don't often hear me say that, but Europe has really been the star in this downturn and has shown almost no contraction at all. I think that that's probably because it's very strongly aligned to the automotive market, to the industrial market, to the analog kind of sectors. They don't feel the pain as much as the computing segment and the consumer segment and the mobile segment. They are the ones that always take it very heavily on the chin, whereas those other sectors are a little bit more resilient and Europe's kind of wedded to those and that's been a benefit for it. So, the European companies have actually done very well and Europe as a region has done very well.

Keith: 5:18

Speaking of the high growth automotive segment, then let's shift over to that for a bit. In fact, in our latest podcast episode, Electrifying the Automotive Future, Fabio Marino delved deep into the role of power-simming conductors in this expanding EV movement. EV is all in the news now. Notably, he highlighted the rapid growth of China's EV market. And in the Washington Post back in June they reported BYD, which is China's electric vehicle giant, has surpassed Tesla with unit volumes of 548,000 in Q1 of 2023, Tesla had 422,000. There's obviously a lot going on in the EV market both in the US and in China. What's your perspective on the automotive sector, especially concerning EVs?

Malcolm: 6:11

Yes, EVs gets all the headlines these days, and rightly so, because it's being driven for the right reasons, but it's still very much in its early stages. The penetration of EV is still relatively low compared with the overall automotive market, so you would expect to see some fairly eye-watering growth numbers coming from that sector, which is simply because it is at the beginning of that hockey stick curve. The big issues in EV is China. You mentioned they are determined to dominate the global EV market, everywhere, from battery production all the way through to end vehicle production. And I think that's the real scary thing, because they're obviously well on the way to achieving that kind of dominance and they're attacking it, I think, in a very dangerous for the established car companies way. They're not just focusing on the high-end vehicles, which is what tends to happen in the West, either with Tesla. And their cars are not cheap, they're expensive, so they're really looking at that top 10% of the sector, of the market. It's very easy to make a profit in that end of the sector, but China is addressing the full value chain, including very low-cost entry cars, and if you learn to make a profit at the low to mid-end of the market, then you will just kill the high end of the market as well. So, I think that's a big danger. From EV overall

Malcolm: 7:38

I think the big danger is that once you go electric like that, then, I know we're simplifying, and I don't want to be too dramatic, but it's just a tin box on wheels with a chip and a battery in there and a motor, so anybody can get into that market. There are 550 "car firms, in brackets, which have announced EV plans. Now that is ridiculous. Most will fail or be acquired, but it is exaggerating the market right now because they're placing orders. You've got the classic scenario of 12 firms all planning to get 10% market share. So, there's a lot of froth in that market right now and there's going to be a shakeout. It's very much early stage, it's very much in its opening stage of market development, but there's going to be some real rocky roads ahead when that market does shake itself out. So, it's one where one has to be really careful about in terms of trying to achieve dominance in that factor. And China, and also to a certain extent India, are determined that they will be dominant in that sector.

Keith: 8:44

So that's interesting. Fabio mentioned something very similar with regards to the barrier to entry for competing in the EV market is much lower than in the gasoline industry, which aligns with what you just said that so many new entrants are coming in. I guess maybe one upside for the consumer is that will create more competition, so it should drive prices down. But can you comment a little bit on why do you think China is focusing more on the low end? Is that more because the high population areas where it just fits better to their end markets, or is it something more strategic?

Malcolm: 9:25

I think it's both.

Malcolm: 9:26

I mean, I think obviously they got a huge demand for a low to mid-range car and I think therefore there's a ready market for that there locally, and, from there, they can spring that out into international markets.

Malcolm: 9:40

But I think it's also very strategic. If you learn, as I said, to make a low to mid-value car profitable, then you won't have any trouble making a high value car profitable. Whereas if you only manage to make a profit by selling a high- end car, you're going to struggle, you're going to preclude yourself from a large part of the market or you're going to not be able to make much money out of it. So, I think there's a lot of strategic movement going on. They want to be able to show that they can capture that low- end market and use that as a springboard to then dominate the whole car value chain. And the facts that Foxcom is offering on automotive platform tells you how easy it is to get into that market, because they are building essentially a platform, and that platform you just simply put your own box around it. It's going to be quite exciting.

Keith: 10:35

Speaking of the semiconductor chips going into automotive for self-driving, let's shift over and talk a little bit about AI. During our conversation earlier this year, OpenAI had just started to capture the media's attention, and now particularly large language models dominate the headlines and there's a trove of new YouTube channels that focus just on this topic. So, given the surge in AI, there's this heightened demand for sophisticated semiconductors as well, and how do you perceive the dynamics changing between these titans of the industry and a lot of the newcomers?

Malcolm: 11:18

Yes, that's a good question. The whole concept of Chat GPT was like the star which took everybody by surprise, and we all kind of embraced it quite dramatically and it was interesting because it was a different approach, and it was kind of interesting and it was cool. So, it opened the door to AI. Now, whether it is real AI is perhaps another question, but I think it is opening the minds to what can be done, and it certainly is creating a disruption in the chip supply market. So, for example, the winners are the NVIDIAs of this world as opposed to the Intel AMDs of this world with the more conventional architectures in their computer chips. So, I think it certainly is a very disruptive happening and people are wanting these very high performance intelligent computer chips.

Malcolm: 12:17

I think it is disruptive. It is causing a certainly a spurt in the market. It is one of the drivers for the recovery in terms of what's making the chip market recover. These chipsets, particularly from companies like NVIDIA, are hugely expensive and they're selling like hotcakes. So, that's all helping the market grow, but it's only really doing what the market does. The whole purpose of the chip market is to create new markets, ones that were previously impossible with the current, then level of technology. So, this has opened up the next door in the semiconductor market. So, it really is a question of the chip industry is just doing what it does best, creating new markets that were impossible beforehand.

Keith: 13:05

So, let's talk about the memory implications. So, a lot of this new technology, especially the AI, uses HBM, or high bandwidth memories, or 3D stacking and putting all this together to get the highest performance but also to try to minimize the power, and HBM stands out as the one of the pivotal elements for AI chipsets. So, given that we follow the memory market and its previous contraction, how is that downturn currently faring?

Malcolm: 13:35

When the memory market collapses, and that's the one which tends to always collapse first, then price is pretty much full to cost very quickly, and so then the contraction is really following the cost profile, which reduces over time and still is reducing, but that's a much slower decline than beforehand. Memory still has a big inventory overhang problem. They're gradually working their way through that, and Samsung was a little bit sort of gung-ho in the beginning with their strategy of not cutting back production. That was probably a little bit ambitious. But now they have cut back production and that's rippling through the whole value chain.

Malcolm: 14:22

So, it's getting under control and you're seeing, yes, this shift towards the high-density memory with Hynix I guess leading that way, the SK Hynix kind of the forerunner of that doing very well. The market's there for it. So, even within memory, it's not all over capacity and over shortage. There are sections there which are growing very fast, and so, as an overall situation, I think the memory market is getting better and the demand is always there for faster, lower power and more bandwidth. I mean, those are the three things which drive that market, and these new applications are, as you'd expect, pushing it.

Keith: 15:03

So, let's tie that to what's going on with the geopolitical landscape. So, I read recently that Toshiba is trying to go private, and a lot of memory manufacturers are investing more in South Korea. We know that the semiconductor industry is at the intersection of technology and geopolitics, especially with countries striving for self-reliance. How do you see this geopolitical landscape influencing the production?

Malcolm: 15:34

When you look at what it takes to be successful in the semiconductor market, it is this constant reinvestment in new capacity and in new R&D. And that's probably easier to do if you're in a more private environment than if you're under the spotlight of the financial community and Wall Street and places like that. So, I can see why there would be this desire to want to have a little bit of flexibility and control over how much investment you can put in and how quickly you can put that in to be able to compete with your competitors, which are in a slightly more private kind of environment. So, I think that's one aspect of the industry. If you're going to be successful in memory, you've got to invest an awful lot consistently, but it pays off in the long term.

Malcolm: 16:23

In terms of the CHIP Act and the geo-diversification of manufacturing, we've seen some signs of that. We were seeing the factories being built in the US and the announcement recently of the factory scheme plans to be built in Europe, in Germany in particular, from TSMC and Intel. Most of these are still plans. They haven't actually put the shovel in the ground and started the investment, and I suspect that they'll be tapered by the normal investment cycle. Hopefully they are, because the last thing you want to do is to build more factories that you can cope with at any point in time.

Keith: 17:05

Speaking of diversification challenges, the industry is facing a shortage of engineering talent across the globe. Unfortunately, just less people are going into STEM, and now companies are finding it exceedingly difficult in recruiting new talent, especially from universities. Just to cite a few numbers, the Bureau of Labor and Statistics has STEM jobs growing at about 11% per year, and they also show that the median salary is about twice or better than non-STEM salaries. Yet here we have less and less of the kids signing up. What is going on?

Malcolm: 17:53

I can understand why they're finding it hard to recruit, in the same way as Europe and America find, it hard to recruit people in this, because they spent the last two decades outsourcing their chip industry or putting their chip industry into decline. So, what student is going to want to go into an industry that they can see is in decline? So, we're kind of paying the penalty for those past decisions that business has made over the last two decades.

Keith: 18:21

The conventional thinking was that offshoring led to cheaper electronics and that benefited consumers. And it also allowed US and European companies to remain competitive in a global economy. From a technological perspective, while, for example, the US could remain a leader in semiconductor design, the offshoring also meant it would lag in some areas of advanced manufacturing. Which now has come back to bite us as countries are concerned about long-term technological leadership, hindsight is always 20/20. Offshoring had benefits in the short term, but the long-term implications concerning national security and technological leadership are questioning the wisdom of those decisions. But I don't know, maybe it wasn't foreseeable.

Malcolm: 19:12

Well, I mean, I have a different view on that. I think they could have seen it. I think the manufacturing has always been important. It was always considered to be important until the financial community told people it's not because we don't want to invest in factories. I think it's curious to see the logic behind that kind of statement. That was no longer a strategic issue. A lot of companies always felt it was strategic. I certainly supported that view and I think we're paying the price for that now. We could have seen it coming, but we took the easy route and that made Taiwan the powerhouse that it is today. But now, of course, we're suffering the geopolitical consequences of that hugely well-executed success story.

Malcolm: 19:56

I mean, you just got to look at the basic numbers. You know, if you end up making the parts yourself, doing the foundry work, you get $5 a square centimeter. If you just do the design, you get 10 cents per square centimeter. For every square centimeter of silicon produced the revenue value of that is $5. A very crude calculation, but by God, it's eye-opening in terms of the relative value of the food chain. If you take a company like ARM, for example, and look at their total sales of $2 billion or whereabouts versus the $550 billion chip market and ARM's 40% of the IP market. So, you're looking at $5 billion versus $550 billion. I mean when people say you know we're going to focus on design because that's where the intellectual value is: Absolutely, but it's not where the GDP is, it's not where the value, the wealth creation is.

Keith: 21:01

Malcolm, you said it succinctly, with strategic importance. That continues to be an underlying theme to all of these topics. Let's shift focus now over to mobile phones. So, the smartphone volumes have been waning over the last year or so and, with the expected market recovery, what might we expect to see with mobile phones?

Malcolm: 21:29

Well, I think I mean the mobile phone is saturated saturated. There are more phones than people on the planet right now. Who are you going to sell the volume to anymore? It really is a very replacement- intensive market and that means you've got to have a strong desire to want to replace your mobile phone, and that's been a little bit elusive for the last two or three, maybe more, generations of phone. It'll be interesting to see when Apple launches the next iPhone next month as to what exactly will be on there that's going to make me rush to go to the store and upgrade my phone. Because phones are so reliable these days and so good, but why do you want to upgrade it? A two or three generation old phone is still a pretty groundbreaking machine. So, I think that's the challenge in the mobile market is what is the next must-have application on there that's going to make people rush for their wallet. So, you're looking at a new application.

Keith: 22:38

Considering the need for a new must-have application to drive mobile phone upgrades: How does the approach of certain leading smartphone manufacturers to chip design and manufacturing partnerships play into their ability to introduce new features and capabilities in their products?

Malcolm: 22:54

It's a very neat little system here whereby we know every September the next iPhone is launched, which is a process, either a new process or a variation of the previous process. And that kind of goes in a tick-tocky kind of cycle whereby TSMC comes out with a process and then they tweak it, then they come out with the next process, and they tweak that.

Malcolm: 23:19

So, you get these kind of half-nodey kind of steps in there. But it's always done with the iPhone chip as the launch vehicle, with Apple as the launch customer and Apple gets the first bite of that cherry, and it is because it's an iPhone. It fills the fab. But a nice part about that is that TSMC has a single chip design to really debug and run a brand new, untested, untrained, unused process against a single entity. And then once that's worked out and the yields have got up and they got it up to a program that could be released to the next tier of customers the NVIDIAs and the MediaTek and the people like that. They're still tier one, but they're tier one minus Apple. Apple is the Godfather and then these are the next ranking down. And then the next year it gets released to the rest of the great-arm washed. So, it's a very stable business model to bring new nodes on in a seamless way.

Keith: 24:22

Shifting gears a bit. It appears that certain companies are making a significant comeback in the mobile phone market, particularly in the face of sanctions. In light of what we discussed about the challenges and the opportunities in the mobile phone industry, what are your thoughts regarding the impact of sanctions?

Malcolm: 24:39

It's like a lot of these issues here. When you try to put sanctions on someone, you tend to accelerate their inventiveness to get around them in some way or other. A lot of these kinds of sanctions often backfire. They have an immediate impact, for sure, but the longer term impact is not quite what people perhaps wanted. If they can buy the parts from the West, then fine, that's what you'll do. If you can't buy the parts from the West, then you've got to make it yourself.

Keith: 25:10

Yes, as the old saying goes, necessity is the mother of invention. Speaking of, in the semiconductor industry, there is a necessity for decentralized computing. This is being addressed Advantest by the invention of ACS, which is an open solutions ecosystem in collaboration with key partners. ACS Edge, for example, is a component of ACS. It's a decentralized real-time data infrastructure. The reason is the capability needs to be able to process the data next to the source, whether that's in the FAB or along the testing lines, like wafer testing, package testing, burn-in or system level. They're doing all of this because of the manufacturing benefits to quality and yield and cost efficiency and speed to market. All that's well-known and all of that's being improved. I want to talk about beyond these tangible benefits. I sense this growing wave of innovation and creativity that companies want to unlock. I want to get your perspective. Have you observed a similar trend? Are you hearing similar things from your customers or your interactions?

Malcolm: 26:32

Oh, yes, to a certain extent. That's what makes the chip industry tick. We invent something new, but it's never quite good enough. We always want the next thing. We're never satisfied. It's just really a continuum of what the chip industry does. When you go back to the mainframes of the 1970s and the microcomputers and the mini-computers and the workstations, there were tremendous improvements at that time, but we outgrow them very quickly. Now, we just want more and we can use more until we end up not being able to use it. That treadmill is just going to keep this industry rolling, and that's fantastic. It's still got so far to go and it's a 70-year-old business, and yet it's still nowhere near scratch the surface.

Keith: 27:25

Malcolm, we've reached the end of our time today and I'd like to express my gratitude for joining us on Advantest Talks Semi and sharing your valuable insights into the ever-changing semiconductor landscape.

Malcolm: 27:38

Thank you

Keith: 27:38

Keith, I look forward to having you back around 2024. Malcolm: Thanks for inviting me back. It's been my pleasure.