Navigating the Semiconductor Slowdown: Insights on the Industry in 2023

Get the inside scoop on the slowdown in the semiconductor industry and what to expect in 2023 with our distinguished guest, Malcolm Penn, CEO of Future Horizons, and one of the most experienced experts in the semiconductor industry. 

In this episode of ‘Advantest Talks Semi’ we dive into the forecasted contraction of the semiconductor industry, the impact of Covid-19 and the shift towards electric vehicles. We talk about the forecast for the memory market and the investments being made in the name of national security.  

We will answer the question of why the memory market will give an indication of the overall market's recovery and that the market’s decline is expected to be double-digit. Questions about Europe and China will be answered as you listen in.  

Tune in to our latest episode of 'Advantest Talks Semi' to stay ahead of the game in the ever-evolving 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: Hello, and welcome to Advantest Talks Semi. I’m your host Keith Schaub, Vice President of Technology and Strategy for Advantest. Today we have the distinct pleasure to be talking with Malcolm Penn, the Founder, and CEO of Future Horizons. Since 1989 Future Horizons has been analyzing and making sense of the global semiconductor industry's tea leaves. And Malcolm has kindly agreed to share some of the future horizons of the semiconductor industry. Hello Malcolm, welcome to Advantest Talks Semi and happy new year 2023. 

MALCOLM: Hello Keith, thank you, and yes: Happy new year back to you as well. It is my pleasure and thanks for the invite. It is always a great pleasure for me to talk about an industry that I’ve lived in all my life. 

KEITH: Great! So, Malcolm, to get us started, tell us a little bit more about yourself and Future Horizons. 

MALCOLM: I’ve been in the semiconductor industry really since the mid-1960s, so my career started way back in the beginning when transistors dominated the industry, and I've lived for a lot of experiences in that time there in various aspects. I have been a designer, I have been in marketing, I’ve manufactured, I’ve run a wafer fab, and a whole bunch of things in that industry sector there. Then I decided to become an analyst when a company called “Data Quest”, who is now a part of Gartner; they got me into the market research business.  

I've been analyzing the market and in ‘89 we decided to spin out and do our own market research. So, we focus much more on providing analysis. We’ve always been independent. We cherish that independence and that impartiality. And we think that helps us, gives us a much more sort of a broader perspective on how the industry works. I kind of like in this industry to be a “rock 'n' roller” for nerds, it's got all the excitement that you get in that music, but it is technical and it's lovely! It's just a very exciting business to be in. 

KEITH: As you know the semiconductor revenue has enjoyed really unprecedented growth over the past two years. Advantest and other companies have been breaking all sorts of revenue records and it has been dominating headlines since 2020. And during that time most of the experts have been forecasting a decade of growth out to 2030 and for the industry to reach 1 trillion dollars. And this year originally, I think it was supposed to be around $620 billion dollars, but late 2022 things started to slow down a bit, and now analysts are revising their forecast downwards. And some analysts are even forecasting a slight 3 to 4% contraction in the overall market. Can you walk us through the last couple of years and why the shift and what is causing the pullback?  

MALCOLM: Yes, sure I think there are several factors at play here. First of all, the industry has never ever grown on a consistent level. It has always been highly cyclical. And you go through periods of accelerated growth, and then a correction often a contraction, and then it starts over again and starts to grow again. So, when you look at the growth rate history over the last 50 or 60 years or so, you do see these huge spikes and the variations are massive. I mean, one year you can have a growth of 30, even 40%, and the next year you can have a contraction of 20%. So, it's a real roller coaster ride. And that’s been typical of the industry in the various systemic factors around that economically driven, which is what causes it to happen. And we can touch on that a little bit later Keith.  

The real problem that the industry has had is that since the 2008 financial crisis, the whole industry has been in a very benign state of growth, so we haven't seen these huge cyclical changes in there. And that kind of let everybody to believe that the industry had kind of matured. And that all sorts of reasons put forward why it was different, why they matured. Everything from the geographic spread, that much broader spread over a much wider diversity event market sectors. And all those things are very true but they're always ongoing. 

We're always expanding the market horizontally and vertically, so they were nothing really new. But what was different over the last decade was that the general world economy had been very very staged, it hadn't really been dynamic. And the recovery from that financial crash was really not a proper economic recovery, it was more driven by all of the economic stimuli, all of the money that was pumped into the economy, the quantity of easing, and all those kinds of issues there. Which was really trying to keep it alive. So, it all led people to this false sense of security.   

But then we had the Covid crisis hit, which kind of stopped the world as we all went into lockdown and literally everything shut down. And then people learned to work from home, which suddenly switched the lights back on again. And that upset the fine balance of what we’d achieved in the industry there and put it on the stimulus of immediate extra demand, where we all had to go out and upgrade our phones, upgrade our computers, a better TV so we can entertain ourselves because we couldn't go to the cinema, we couldn't go to the theater. So that gave a huge spike in demand.  

And there were shortages everywhere, I mean shortages not just in chips, shortages hit everywhere. Simply because of this spiking demand. And the industry, and no industry, can accommodate these spikes. The supply chain can't react that quickly to any significant distortion disruption in the demand on the upside, and that's why the shortages started to occur.  

So, we really flip back into what really was a classic industry uptown driven by under capacity and that led to an increase in unit demand which then led to price increases which then added together or - multiplied together actually - to give you this huge 26%-dollar growth that we saw in 2021.  

KEITH: How is this affecting the consumer-driven markets versus enterprise markets? What I mean by that is, I’d like to go back to what you said: There was this huge spike in demand, everyone's ordering new equipment, but the enterprise markets, at least so far, are remaining relatively resilient. And do you expect that to change this year, or do you expect it to stay the same? What are your thoughts on that? 

MALCOLM: Well, the consumer market and sectors like that are always much more dynamic and much more - I guess - driven by sentiment. They're always ‘gonna be more a little bit more sort of fluctuating than the enterprise markets which are a lot more stable and driven by investment, whether by industry investment or infrastructure investment. So those are a little bit more kind of mature in the way that they grow. And they are more steady because those investments are longer-term investments, they are not driven by emotions so much as by longer-term commitments to growth in the infrastructure. 

So, you always see different market segments behaving under a different time scale there, and the consumer ones are very fast to react. And automotive is kind of somewhere in between the two there because that's quite a high purchase, a discretionary spend. So, it does take a little bit of consideration to go out and buy a new car, but it's not quite the same as building out your infrastructure, building out a new data center. Those are always expanding, demand for data is insatiable. 

KEITH: You mentioned the automotive segment. Historically, the wireless communication has long been rated as the industry’s most important revenue drivers. This year for the first time it slipped into second place and was supplanted by the automotive sector as the most important sector for driving semiconductor revenue. Can you touch on a little bit of why that is? And could you give us a bit more color to that? 

MALCOLM: I think with the automotive: It is on longer lead times in terms of the production cycle of that there. The design cycle is quite long, the approval cycle is quite long. It’s still built on a production-line basis. Most of the products are built in batch related basis but you still have car production line. There's a lot more sort of lethargy in these lines here, than there is compared with some of the other major markets for electronic component semiconductor devices. So, the lead times to change the dynamics there, are a lot longer than they are in other sectors. But I think you have to put that into perspective. 

Because when you compare for example the number of mobile phones that we built, maybe 1,5 billion mobile phones a year, to the number of cars that get built which is a 15th of that value there, 100 million cars. The unit volume is significantly lower in cars than it is in mobile phones. So, you're gonna have to have an incredibly larger semiconductor content in the car before that market will actually be the same size as the market for mobile phones, which is the number one driver for semiconductor devices at the moment.  

So yes, it is incredibly important for several factors: the first factor is of course the move toward electric vehicles. And that is something which gained pace in recent years when we all actually finally became a bit more conscious of keeping our world intact, and not polluting it, and not generating a lot of the emissions which are harmful. But the batteries have to be very very strictly controlled and the only way you can do that is with semiconductor devices. So, there is a huge spur of demand for semiconductor devices adjusting the battery alone. 

And then in parallel with that, we've got all of these other features in the car: the buzzword is, of course, self-driving cars, driver-assisted cars, collision avoidance, accident avoidance, and all those kind of things in there are going to happen and are happening. And whereas you probably will never have a driverless car for a long time in a normal environment you certainly will have cars which are literally stuffed full of computers and microcontrollers and stuff in there making them safer for us to drive - and they've done a good job on that there.  

So, you’ve had that aspect going on and that's been certainly meaning that semiconductor content in cars has been growing probably one and a half times as fast as overall in the industry itself. I think then you had to put that into the context of when Covid hit you couldn’t go and buy a car: The car showrooms were shut, you couldn’t leave your house, so demand for cars overnight, when that lockdown happened, demand for cars just evaporated completely and went to zero.  

The car manufacturers reacted to that by canceling all their production orders. They just simply said we can't run our lines. But then of course eventually demand started to pick up again maybe 3, 4 months later when the lockdown started to be loosened up a little bit there and suddenly the demand was still there - it was just pushed out. And when they went back to buy the parts, they suddenly found the factories were full and they couldn't place any orders, so that kind of caused the car shortage, even though the demand for cars was still fundamentally there. Driven by the move to elective vehicles, but of course, the supply side was under constraints because they couldn’t get the parts.  

KEITH: Yes, I think it went from Just In Time to Just In Case at least for a lot of companies because of the supply chain issues that were incurred from Covid. I want to talk a little bit about memory versus non-memory: You mentioned also the average selling price for automobiles, for example. You were talking about a $50,000 dollar automobile, versus a $1,000 phone. There seems to be a lot of room there to include a lot more semiconductor technology.  And memory has been growing quite substantially over the last few years for precisely for some of the reasons that you mentioned. Where all these new horizontal markets are coming into play. But for this year at least, how is the memory market being affected in 2023, and how much of that is affecting the overall market? 

MALCOLM: Memories always are the first product into an industry correction. Whether an increase in demand or a decrease in demand. Memories are the most commoditized product that we are make in the semiconductor industry.  

When you think about it, a memory has to be identical, no matter whether it comes from Samsung or SK Hynix; they have to be identical, they have to plug into the same slots. So, there is no way to differentiate really your part from your competitor. So, it is the definition of a commodity part. And that means that there is that intense competition, and the way they exercise that competition is really nuanced more by pricing than it is by secondary factors. Although secondary factors come into play, reliability dependability, and all those issues, but all those suppliers in memory are all pretty good, the top-tier supplies. It has whittled down to a handful, and they're all very very good.  

It is a highly commoditized item. And they are the first to go into a downturn and they are the first to come out of the down (turn). And their changes are fairly dramatic, because they are an amplification of the dynamics from the rest of the industry.  

They did have a large increase in the average selling price. Because when you are in a shortage the first thing you do, when you can't literally make any more parts, you just simply ration them by putting up the price. And you can do that in memory. And likewise, the opposite happens when the demand starts to slow down. You drop the price to try to stimulate people to keep buying the memory there. The underlying demand for memory hasn't gone away – it's insatiable.  

Well as other sectors, analog for example, are always the last to feel the pain, and they are the last to recover as well. So, the memory market collapsed pretty much in June of last year and that was followed by the micro market pretty quickly because that's commoditized – to an almost, not quite the same degree as memory - but it is much more commoditized.  

Logic is then much longer to come in because of the long lead times and that's just starting to slow down now and the analog likewise, similar kind of timeframe there, that starting to slow down as well.  

KEITH: Maybe all we need to do: look at when memories recover: that’ll give us our first indicator. 

MALCOLM: Memory and micro, yes. Once they start to recover. Memory gives you the first clue, micro then cements that.  

KEITH: So Malcolm, you talked about some of the companies and their geographies. So, a lot of analysts have US and Europe and Japan actually growing this year. But the rest of Asia Pacific, and China in particular, declining. So, I wanted to get your thoughts on the reasoning behind that, and can we expect that perhaps for the remainder of 2023 and 2024? What are your thoughts on that? 

MALCOLM: The bigger picture aspects of that in the market overall and then the nuances for the individual reasons. For the overall; When I think back to this time last year, we forecasted the market growth would come to an end in the second part of this last year and that the market would actually crash in 2023. 

And that was a pretty brave statement to come out with, two brave statements really, because the market was really really going like gangbusters at that time there. But we said it would slow down in the second half and it would crash in 2023. And as it turned out we were absolutely right.  

And there are several reasons why we believed it was gonna be that way. And it did slow down, and that slowdown is working it’s way through the system so the growth that was originally projected for 2022 - the general consensus was in the 10 to 15% growth range. That now looks like being much closer to 3% growth this year. Simply because the second half of the year did indeed slow down and has gone negative compared with the first half of 2022. The forecast overall for 2023 in our view is still a significant market contraction overall. 

We're in a double-digit type of decline somewhere around 20% decline. Most of the analysts now accept that it will be negative this year but they're still very much in the low single-digit number there. We don’t see that happening. When you look at the momentum, the current momentum, when you look at what's happening in the first half of the year it is very very hard to get a decline less than 10%. You really have to struggle to get it down to 10%. We will be refining that number the week after next, we have our update event. 

But it looks like it will be a significant decline. Now, when we have a decline like that no market is going to grow. OK, so the markets that will grow - some will decline less than others and the one that declines less is Europe actually. Because Europe’s product mix is more kind of boring and stayed. So, it's still growing right now - slowly but still showing a bit of positive growth. Whereas the other markets are actually showing a decline in growth in there.  

China is in a much deeper decline simply because its supply chain has been so disrupted, they've only just released their Covid lockdown strategy and allowing factories to reopen. Whereas for most of this last year a lot of the factories there went routinely into lockdowns and so the supply that they were making or should've been making was simply not available to be made. So that means I don't need to chips, so that means the market for them was declining the fastest. So yes, they were the worst performer. Whether they will be the worst performer next year really depends on how much they and how fast they can recover from that supply chain disruption, but it also depends on what the market demand would be. Whether if there's a need for that product now and next year. Compared to what it would be, because of the general overall market decline. 

KEITH: So, you mentioned the global big picture versus the various geographies and segments which leads me over to the nationalization of the semiconductor technology. And as you know we have gone from, a few years ago,  where semiconductor importance wasn't even on the radar. And in fact, if you said the word “chips” often much of the public thought you were talking about potato chips. To now every country has semiconductors as a top priority, and it is vital to their national security and countries are investing billions of dollars, and subsidies, and some hundreds of billions of dollars in the semiconductors, things like the Chips Act here in the US. Help us with understanding why the shift, why has this occurred and is it going to remain this way moving forward? 

MALCOLM: Yes. I mean the good part about what you just said is that finally people know semiconductors exist. Even the man on the street understands. So that's kind of gratifying that now people are aware of this. Because we've always known that they’ve been driving the world, but they are the unsung hero of the of the world really. But now people understand that. So that's been very good and the shortages, which was probably the automotive industry was the most vocal in shouting and screaming about the need for chips, and more investment and things like that: they got the attention of politicians, particularly in the US I think, and in other parts of the world. 

And so, we did have this tremendous impetus to investing in semiconductor manufacturing because at the same time, it exposed the fundamental kind of limitation of the factory that everything has been outsourced now to a high degree. And the concentration of production was really in one or two areas around the world particularly in Taiwan for logic and non-memory parts, and in Korea for memory devices. And both of those countries are pretty close to China but suddenly we were aware, politicians became aware of the fact, that the world and their companies and their governments and their GDP’s were very highly dependent on production from Taiwan, which China regards as a province.  

Which would mean that the world will be dependent on production from mainland China which is kind of not want the politicians were trying to achieve. So, you've got all of these political influences and practical influences kind of all playing together. And the simple answer would be: build more chip factories at home and hence the Chips Act from the US. Now interestingly Europe always had this kind of government, European level government support and local government support for its industry. And for a long time, I mean 30, 40 years this has been going on.  

And so, I think they’ve been treading that part for a long time, it's not quite as vocally broadcasted as much as the US chip became a headline, but other countries are also realizing that they had lost control of that manufacturing local production of Semiconductors. 

So yes, we’ve seen these investment programs in there, and yeah at the moment they’re more promised than dollars but in the US they are real dollars now. We are seeing the first green shoots of that investment actually coming close to producing part.  

KEITH: OK, so Malcolm we are nearing the end of the show and wanted to open this up for any final thoughts. 

MALCOLM: Yes Keith, thanks. So overall the industry: it will go into a downturn next year, it's pretty clear that's gonna happen. There will be negative growth, there will be a contraction but that's not the end of the world. It is going to recover. And 2024 will probably be a modest growth year - that depends on the global economy, and what happens to that. Because the economy does kind of govern the whole ambiance of the industry overall. But certainly, we will see a return to strong growth in certainly 2025 because there will be some cutbacks and that will trigger the next shortage and that will trigger the next upturn.  

We had 17 downturns so we can look forward to the 17th upturn and unfortunately to the 18th downturn in three- or four years time. But it is still the greatest industry in the world. 

KEITH: And with that, it has been an absolute pleasure to speak with you. And the valuable insights you provided us are amazing and fantastic. And I look forward to having you back. Perhaps we can make this a recurrent event. 

MALCOLM: That is very kind. Thank you. Yes, I would love that.  

KEITH: And that does it for another episode of Advantest talks Semi. See you next time.  

Post Podcast:

KEITH: This is Keith Schaub, Vice President of Technology and Strategy, hosting “Advantest Talks Semi”. We have some important Advantest news from Hira Hassan, our Global Marketing Communications Specialist, followed by some recent event highlights with Junko’s Top 3.  

First, we go to Hira for the news. Hello Hira! 

HIRA: Hi Keith. I am thrilled to announce our annual developer conference VOICE is now open for registration. This year we will be holding VOICE in Santa Clara, California in the Santa Clara Marriott. Make sure you register as soon as possible to take advantage of our early bird prices.  

Register today at voice.advantest.com.  

And, as always, be sure to connect with Advantest on LinkedIn, Twitter and Facebook for all the news and much more.  

Now, on to Junk’s Top 3! 

JUNKO: Thank you! My Top 3 in this episode is about SEMICON Japan, which was held on December  
14 - 16 in Tokyo. 

First, I felt that the general public, people outside of the semiconductor industry, were a lot more interested in SEMICON Japan. The show's attendance was almost doubled from last year, and we met not only with our customers, but also with many analysts, bankers and investors in the booth. Also, SEMICON Japan was mentioned in one of the major local TV news shows, and the Japanese prime minister made a surprise visit on the first day. 

Second, Advantest yet again demonstrated our strong industry support and technology leadership. We were the gold sponsor of SEMICON Japan. And we also sponsored the Advanced Packaging & Chiplet Summit, which was one of SEMICON Japan’s new initiatives this year, and Smart Mobility Pavilion as well as semiconductor Test Symposium, where we had a paper, titled "Data Analytics in the Chiplet Era". 

And, last but not least, amongst many products we displayed in the booth, the most popular one, I thought, was our new inteXcell, it is a combination of our memory tester and handler, so it appeared massive in our booth, but actually it uses only one-third of the floor space comparing to conventional systems.  

This concludes my Top 3 from SEMICON Japan. Thanks for listening!  

KEITH: Thank you, Junko, and thank you Hira.