2024-2025 Semiconductor Industry Outlook with TechInsights

Prepare to be illuminated by the insights of Risto Puhakka from TechInsights as he joins us to map out the semiconductor industry's explosive growth, driven by the thunderous influence of AI. 

During our adventure through CES 2024, we bear witness to AI's pervasive touch, from robotics and healthcare to the sizzling innovations in micro LED technology. Risto's expert analysis doesn't just skim the surface; it plunges into the economic, political, and competitive undercurrents that are redefining the semiconductor landscape.

Venture further into the conversation as we unravel AI's sprawling web, extending its reach from agriculture to retail, and sculpting the semiconductor market's future. We dissect the burgeoning world of spatial computing, and the surging demand for data streams, processor aptitude, and memory. 

Gaze into the crystal ball with us as we reflect on 2023's rollercoaster ride – spotlighting industry bottlenecks and breakthroughs – and cast our predictions for 2024's market climate. 

It's not just a tech talk; it's a strategic analysis of the global shifts, the CHIPS Act's ripples, and how different nations are vying for supremacy in the AI arena. With Risto's guidance, we're decoding the blueprint of tomorrow's semiconductor industry, where AI reigns supreme.

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: 0:30

Welcome, youkoso, to another great episode of Advantest Talks Semi. This episode titled Navigating the Semiconductor Horizon 2024 and Beyond, Risto Puhakka, general manager from Tech Insights, the authoritative information platform on the semiconductor industry, is going to be guiding us through the dynamic world of semiconductors. Back again, Risto joins us to share his invaluable insights on the current state and future of the semiconductor industry. From the impact of AI and robotics to geopolitical influences and the latest from CES 2024, we're covering it all. So let's get started and explore what lies ahead in this ever-evolving industry. Risto, happy new year 2024 and welcome back to the show.

Keith: 1:25

So, Risto, I want to kick this off. It's hard to believe, but this time last year, ChatGPT from OpenAI had just debuted on the scene and it's only been a year, but it feels like the entire world is now obsessed with AI. We've now got chip manufacturers and hyperscalers battling for market dominance and, to paraphrase Lisa Su, CEO of AMD, she said last year has shown us that AI isn't just a cool new thing, it's the future of computing. I thought that was interesting. I would say it's the future - full stop, and nowhere was this more apparent than CES 2024, which ended, as you know, just a couple of weeks ago. So, our Advantest team was there and what we saw, literally every single company's messaging and product centered around AI in some way. So, the question is: What happened in 2023? And did we literally just see the entire world change in the blink of an eye?

Risto: 2:29

It certainly changed in the blink of an eye, literally in a matter moments. ChatGPT was launched late in 2022 and became available. I think it experienced the fastest ramp of usage that the world has ever seen. The adoption rates, the usage rates just went through the roof and people were fascinated because the ChatGPT literally could take a lot of human tasks, which were traditionally human tasks, and suddenly they became very, very easy to do.

Risto: 3:06

But you know, they joke around, like my people are, why don't you take this data check, call the ChatGPT and write the report on this? And then, just, you know, you mentioned Lisa Su, and AMD, a great company, and Lisa put out this $400 billion dollar market size in four or five years and it became a huge head scratcher for us. For example, are we going to go against Lisa and say, no, it's not 400, or are we going to go figure out what Lisa's thinking is or AMD's thinking is? And those are kind of the ramifications. And then we can't forgo the comment that NVIDIA became a trillion dollar market capitalization company. And there were not a lot of those companies and now we have a semiconductor company there with that. So that obviously got the fascination of the whole financial world in a single say then. So we had a, let's say, a perfect storm that was lit up in the late 2022. So yeah, that's where we are.

Keith: 4:16

Yeah, great, it's exciting times, for sure. So, aside from this AI phenomenon that has seemingly taken over the world and was definitely one of the overarching themes that we saw at CES 2024, I noted six key technology themes for 2024. AI, obviously number one the ubiquitous integration of it. The advancements in robotics - the robots in your homes to mow your lawn and to help the elderly. Healthcare- expanding applications and computer vision, which go hand in hand with the AI. And then the momentum in spatial computing, which is the rebranded term for AR and VR. Of course, EVs and automobiles and the innovations in micro LEDs and transparent displays. So, the question is, I mean, these are the six themes that we sort of took away and that people were talking about. Do these six themes resonate with TechI nsights and your observations, and why these six? I mean AI notwithstanding why these six, and are there others that we should be thinking about?

Risto: 5:25

We had people at CES. They're still writing some of their observations and comments and analysis, but those are the key things. But I would like to add a comment here: You know, first is AI, which is effectively a little bit of, let's call it, an unknown application. We are still trying to figure out what it is.

Risto: 5:43

Then you got robotics, computer vision, spatial computing, which is a technology. EV, a kind of application, but driven by technology pieces. And micro LED, which is definitely a device. Those five are all centered and driven by AI. So, it's a kind of catch all AI show. I was not able to make the show myself, but you know, we had a 10- person team there and when you look at the comments, they are either driven by AI or they drive AI, one way or the other. So, you know, take micro LED driving data to AI. Take an EV that uses AI or, let's say, autonomous vehicles: they require that AI component to become fully functional. Spatial computing, a world of AI and computer vision driving data to AI and robotics is going to be a great user of AI. So, you have actually a huge amount of technologies that are basically co-habiting the marketplace, and the AI is in the center.

Keith: 6:57

And just to sit on that for a second, I heard recently also that we talked about AI going into different segments. But one of the big takeaways is, that AI isn't going into different segments, it's going into all segments. Like everything now doesn't matter what it is: agriculture, industry, gas, oil- mining, your grocery stores, everything has AI as part of the growth behind it now.

Risto: 7:30

That's exactly it. It is an all-encompassing application across various industries and uses.

Keith: 7:39

Okay, so we could probably talk about this for quite some time, but there's so much to cover here. But I want to focus on two of those trends: The first one, spatial computing, and the second, micro LED. The reason I want to talk about spatial computing is because Apple's Pro Vision is actually coming out next week, and there's been a lot of sneak peeks at it and a lot of videos, and it's getting a lot of press in the media. So, from that aspect, how do you see this technology impacting the semiconductor industry?

Risto: 8:16

Well, again, you're going to need a higher data stream for those devices. So, and that corresponds, you need more, better processors, you need better logic components, you need more memory. All of the things that basically drives that experience are going to be driven. Basically, I'm going to say, from the perspective of silicon, it will take more silicon per whatever your measurement is, compared to what the previous applications were, whether it was your smartphone or your TV having that experience. But yeah, that's kind of the experience we should expect from that.

Keith: 8:59

So let's pivot then over to the transparent micro LED displays. I think a lot of consumers, this was really their first time seeing this technology, at least in a big way. It's a very futuristic technology with transparent displays and Samsung had one of the largest displays there as one of the largest semiconductor manufacturers. As it unveiled it there, I think it was 77- inch micro LED display to sort of open CES. But I kind of want to draw everyone's attention to the technology has been in research and development for the past six- plus years. And Advantest has been working with those companies on developing next generation test solutions for that space. The technology is still expensive, I think, compared to traditional displays, but I do see a lot of potential industry use cases. So, from TechI nsights and from your perspective, how do you see this market and its impact on the semiconductor industry?

Risto: 10:00

I mean, obviously, we still have the cost issues to address, but it is a technology that we should expect to follow the normal cost reduction curve that we see in display technologies across the generation. So, I don't have any issues with that. And then, once we get there and we start to see those cost reductions, it will come everywhere. You know, again, it's a progression of display technology providing the users, the consumers, richer experience. I wish I had seen that 77- inch display from Samsung. And, you know, you need to start building your house around the display technology rather than fitting your display into your house. So it's kind of an interesting way to start to change the world around us. But, yeah, you have a completely different experience as a consumer or user of that display technology. So that's what we will see.

Keith: 11:08

Well, I think also they're hoping and thinking that we're going to build those displays on the outside of our houses as well.

Risto: 11:15

Exactly, yeah. Instead of buying your Christmas decorations for your house, you'll just have a display. Well, it makes life easier.

Keith: 11:25

Yeah, it does. Okay, so, like I said, we could talk about CES for the entire show, but let's stop there and move on to talk about market situations. What happened in 2023 and what the outlook for 2024 is. So, 2023 was a tumultuous year. The semiconductor industry went into a slowdown. There were pullbacks, a lot of companies announced layoffs, orders were shifted out and the entire ecosystem sort of slowed down for a bit. So, it's over, 2023 is completed. How did the semiconductor industry end up versus the expectations?

Risto: 12:03

Well, if you look at what was forecasted in late 2022 for 2023. From our perspective, we didn't expect semiconductor down- turn in the end and also equipment down turn along the way. Obviously memory was leading the way. There definitely was a substantial cyclical downturn. There was a debate about whether non-memory logic and power analog would follow. We did keep our view that it is unavoidable, probably milder, and that certainly was the case.

Risto: 12:45

The interesting observations from last year is, that when you think that the business is darkest, that's when we are at the bottom and sure enough, the bottom was second quarter around May and June . A and now in the third quarter things started to turn up.

Risto: 13:06

It did not feel like that, but you know, if you look at the data and how things evolved, you noticed the space in the memory obviously started to see that improvement. But also the logic, especially in the fourth quarter, started to get back to it then. The interesting observation is there that the power in analog segments didn't follow the downturn but almost like a year behind. Now that you see memory and logic recovery nicely, we actually see power and analog having a slowdown now. So that's kind of interesting how the different segments experience the cycle. In our expectation, for power and analog, that the cycle is a lot milder and so there were such a shortage of those components in 2021, 2022 or even still.

Keith: 14:04

You mentioned 2024, but before looking out at 2024, I do want to get to that. But we had a very notable acquisition that occurred this month, which made some big news in the semi-industry, with the announcement of Synopsys acquiring Ansys for $35 billion. So, can you walk us through that? Why the acquisition and what could it mean for the industry and what should we be thinking about there?

Risto: 14:29

The way I look at this, Synopsys is what I would call, and this is with the huge respect the classical EDA company providing. And they have recognized the value of simulation. They have definitely invested money on their simulation capabilities. And then Ansys comes from different origins, not the classical EDA, but they built out their capabilities through simulation. So, there is now a strong EDA and a strong simulation capabilities brought together. I haven't seen and haven't had a chance to really talk to customers and users, but my understanding is that it's really customer-driven that they actually see that. If these two guys can combine their tools, there's going to a very very nice outcome for the customers So, So that's also driving the acquisition from the background and I think it's listed as a merger. But basically that's the setup.

Keith: 15:35

Okay, great, I think we're just seeing a lot of things shifting around with this new globalization and one of the ways I've heard it described. So there are two transitions. There's the transition from the age of discovery to the age of implementation. Second, from the age of expertise to the age of data. Maybe it's a bit convoluted, but the way I interpret it, is that it's saying, right now, we're in the throes of AI implementation, with big data, and the companies that embrace and implement AI throughout their ecosystem. And the ones, of course, that have access to the big data sources, those are the companies that are best positioned to win in this winner-take-all scenario, or winner-take-all game. I guess maybe you could think that acquisition is kind of part of that thinking. Is that the right way to think about this, or should we be thinking about it in a different way?

Risto: 16:37

I'm thinking of it in a more different way because, well, I mean, in certain ways it's a way, but it still stays within the semiconductor design environment. So, you can have that two nanometer, one and a half nanometer design, loads of data that you have to go through, and now you have a substantial simulation capabilities to build into that design. So, you actually start to see how that design is going to turn out in the real world. So big data plays a huge role. It's just that the big data is within the semiconductor world. When we talk about the big data, we talk about Googles and Facebooks, and this is a different application of that big data.

Keith: 17:30

So, PricewaterhouseCoopers (PWC) has highlighted that the transformative potential of AI, projecting that it could contribute over 15.7 trillion to the global economy by 2030. And a large portion of that growth is driven, or is expected to be driven, by generative AI. So, in your perspective, how could this impact the semiconductor industry in 2024 and 2025? What should we expect for the semiconductor industry in the next year or so?

Risto: 18:08

Yeah, well, this leads to a great point. On the long term, let's go to the long term because, you know, PricewaterhouseCoopers mentioned 2030. There is always this trend and discussion in our industry. So, when will we reach a trillion dollars in semiconductor sales? We haven't changed our forecast yet, but we have internal discussions. How do you count that? Let's assume Lisa Su's forecast of 400 billion becomes true, it will definitely push that one trillion number higher, for sure. Now, Price waterhouse Coopers, whichever ways they came up with that, the analysis of 15 trillion, well, we don't have to worry about the funding of that trillion dollars anymore because the money is there. So, you got to get both sides to kind of align. That actually describes that. The environment around AI. Lisa Su basically saying, we have to build up the supply here to to reap those benefits, and the numbers get big very quickly. That's the long term. Let's look at the the shorter term, the 2024 and 2025.

Risto: 19:33

I mean, I presented early January our 2024 and 2025 forecast publicly. We are very positive. You know 16% up on semiconductors this year and after 10% to 25%. When I was presenting, I made comments that I am a little bit concerned on the demand side because at the time we added that demand side. It was still a little bit cloudy worried, about the inventories. You know once TSMC announced their fourth quarter numbers gave the projections for 2024, those demand concerns are way gone. We're going to see a classical upturn, for sure, in the in the next couple of years. And then the only question is, what kind of numbers we can truly see. But but the very healthy two, three years ahead of us right now.

Keith: 20:35

So right now we're looking at least forecast 16%, and next year it's targeted to be 10, something like that.

Risto: 20:42

Yeah, 16% this year, 10%, 25% or so, and the further we go out from the cycle, we actually see the demand improving. And we already now see that the demand is actually fairly robust, you know, and it's not only AI but also other parts of the of the semiconductor businesses start to see the demand improving.

Keith: 21:10

I want to feed of a little bit of what you said. Obviously, with these large numbers like 15 trillion and 400 billion we're seeing now countries positioning and investing at- you know, that they're all vying for their position to compete for what looks to be a very lucrative economic opportunity, and so we have the CHIPS Act here in the US and then other countries have their own versions of the CHIPS Act. So, considering the challenges around AI governance, reliability, AI security. Are there any geopolitical shifts or anything that you're seeing, and what should we be watching for?

Risto: 21:53

I mean we're dealing with, I would say, three large powers. We can always debate what are the relative strengths of each. We have the United States, China, European Union, as you know, and of course there are opinions of who is stronger and what are the weaknesses and strengths. But each of them will have their own approach for AI. In case of China, they are limited on access to the most advanced technologies, which creates its own challenges. Then United States is clearly given

Risto: 22:33

Nvidia, AMD, US government is taking a very aggressive approach to it. And then Europe is still kind of a little bit, let's say, evaluating and looking what they potentially can do. They see the opportunity to some other parts of the economies, but for sure AI. And you can look at it from a regional, geopolitical perspective. But equally, one step down, corporations, the competition - they got hyper scalers. The competition is fierce because you may decide really winners and losers on this technology cycle. You know, look at Amazon, Microsoft, Google, Facebook, Apple how those have to invest and find the best applications and what fits their business. On that geopolitical level you go one step lower. You have that corporate level and the competition is as fierce, if not more fierce than geopolitically.

Keith: 23:41

Risto, we're at the end of our program, and I want to thank you for sharing your invaluable insights and a shout out to Tech Insights. Before signing off, though, any final thoughts?

Risto: 23:52

I mean just the general thing is we are actually at the beginning of a very nice cycle here. What we didn't really mention here: One of the beauty of AI from just the pieceless and forecasting perspective: We are not limited by the number of humans, we are limited by the number of machines and it's growing. So, that's an interesting thing to think about at the end of the interview here relative to AI. But very nice to be back. I really enjoyed it.

Keith: 24:19

Good point. Risto, thank you for joining us today. And to all our listeners, we hope you found this conversation as enlightening and thought provoking as we did. To learn more from Tech Insights, you can visit them at techinsights. com. And don't forget to tune in to our next Advantest Talks Semi episode, where we'll continue to delve into the world of technology and innovation. Thank you for listening.