All right, good afternoon. Let's get started. Everybody's still awake. I see you have some coffee. I'm Tristan Gerra. I'm the Senior Semiconductor Analyst at Baird. I would like to introduce Diodes. Maybe they don't need an introduction, but I'll do it anyway. It is a leading supplier of analog and discrete solutions, serving a variety of end markets. They've gained consistent market share over the past many years. We're pleased to have with us today Emily Yang, Senior Vice President, Worldwide Sales and Marketing, and Gurmeet Dhaliwal, Investor Relations and Corporate Marketing. With that, let's get started. I think you have a few slides for us.
Thank you. Yeah, I'll just go through the high-level introduction, and then we can get into the Q&A. So Diodes is a semiconductor company. We serve analog and discrete power solutions, and then we serve five segments, our customers in five segments, which is automotive, industrial, compute, communications, and consumer. We have been in business for over 60 years. For the last 33 years, we have been profitable consecutively. Last year, we closed 2024 at $1.3 billion revenue. And in those market segments, if you look at it, you know, back in 2018, Dr. Lu made a goal that 40% of the revenue should come from automotive and industrial market segments. For the last 12 quarters consecutively, we have been above that goal. That is the focus.
Just to give you some highlights on the automotive market segment, since that's one of the focus segments, we are focused in three areas of the automotive segment, which is connected driving, comfort, style, safety, and electrification. The way we have been, you know, we've been growing at the CAGR of 22% since 2013 when we established this automotive BU, with the focus on these three subsegments within the automotive segment. The first two is connected driving and comfort, style, and safety are agnostic, whether it's an ICE car or an EV car. That's where, you know, the majority of our silicon is. The other thing is, back in 2013, when we established the automotive BU, we had $28 content per car. We have been growing this over the years. As of 2025, we are at $213 per car.
Just to give you a profile of the revenue, one-third of our business is direct, two-thirds is through distribution, and that's based on the ship-to revenue. Also, from the segments point of view, 78% as of last quarter was from Asia, followed by North America, and followed by Europe, and then North America, 9%. Our market segments that I mentioned earlier, five of them, is computing is the largest segment for us, which has been, which did grow from last quarter as well. That was 27%. And automotive and industrial, as I mentioned before, combined is 42%, and then compute and communications. And these revenues that we show here are based on the ship-to either location or segment or direct versus the distribution channel. This is just high-level profile. I will stop here, and Tristan will ask follow-on questions.
Great. Before we talk about the more high-level questions about the company direction strategically compared to some of your peers, there's a lot of questions, obviously, on tariff, lots of cross-current. Intel has talked about some advanced purchases in PCs in Q4 and Q1. What do you see in terms of orders? Do you think that we're still at a point where things are tracking pretty much in line with seasonality, or do you see anything abnormal one way or the other? Maybe we can tie this to inventories as well.
Mm-hmm. Yeah. I think overall, what we're seeing, we didn't really see a significant pull-in, like Intel mentioned, maybe a little bit in Q4 or maybe beginning of Q1 because Chinese New Year. Current quarter, we don't really see that. We have a lot of customers because they don't know what to expect. They don't know what will be the tariff policy tomorrow. Most of the customers are actually more on the conservative side. That's what we see. They just kind of wait and see, almost like business as usual. We're definitely not seeing that. We are not that worried about it's going to impact the Q3 or Q4 revenue. Of course, tariff is a little bit unknown. That's really more a concern on the end customer consumption, not specific to the pull-ins and push-outs and stuff like that.
When you mentioned that people are still conservative, should we read that they're still reducing inventories, or how's that process? I mean, there's been already a lot of inventory burn over the past several quarters. Maybe if you can remind us of your position in terms of inventories and distribution. What do you see, you know, other customers doing with their own inventories?
Yeah. So overall, on the channel side, right, distribution side, we actually define a normal range of inventory, 11-14 weeks. At this moment, still slightly higher than that. But if you look at the Q1 results, we actually have a decrease in internal inventory as well as channel inventory. Our POS, POS revenue, point of sales revenue increased, Q1, which is actually usually not normal. Q1 is usually a down quarter for us. As an end result, if you look at our actual revenue, we actually decreased 2.2% for Q1, which is actually better than usual seasonality. Usually, Q1, we down around 5%. Second quarter, usually it is an up quarter, about 5%. We actually guided more than 7%. We actually view it as actually better than usual seasonality.
Okay.
Mm-hmm.
Just similar question in terms of end market. We've seen mobile starting to pick up a little bit in, at least in China. What's the commentary that you would have in terms of the key other end market, and notably industrial and automotive at this point? Do you think that your, you know, those end markets, you're shipping in line with end demand, or is it, you know, people are rebuilding inventories, or it's too early? What's your outlook medium term for those markets?
Yeah, I'll just give you an overview on all five segments, and then we'll specifically talk about automotive industrial. You know, three Cs, what we see is the inventory is relatively clean in all three Cs, especially computing and consumer. With the communications, it's a little bit mixed because we put smartphone in the communication segment, which behaves just like consumer. Some of the communications part, which is networking, telecom, enterprise, it's starting to pick up in cleaning inventory. Versus automotive and industrial, we believe that it's still going through inventory rebalancing. It varies from customer to customer, region to region, which could take another quarter or two.
But overall, based on our focus in compute segment as well as in the automotive industrial, as we are bringing new products to the market, last year alone, we introduced more than 300 products for the automotive segment, and last quarter in Q1, 49 new products. With our focus, even though the number of cars being produced is flat to down, I think with our dollar content increase that I mentioned earlier and new products, we are continuing to see momentum.
Yeah. Maybe just, let me add a little bit, right? We are seeing stronger book-to-bill ratio. We actually have stronger beginning backlog, you know, getting into the quarter. POS, we actually continue to expect to improve. We also openly mentioned second half of 2025 will be stronger than the first half of 2025. I think with all the indication, we actually expect some recovery coming. You know, I did mention it's a positive sign of recovery. Even with auto and industrial, we start to see a lot more urgent orders, a lot more expedites, a lot more short lead time orders. That's also the reason I think a lot of customers is more on the conservative side on the buffers inventory. You know, I don't really think people are building more buffer.
It just, you know, they probably haven't really recovered what the buffer level from before. I think all this adds up is actually a pos-positive indication of recovery of the market.
On the rush orders across end market or in specific end markets?
I would say it's across different market segments. The majority, I would say, distributions, you know, probably a little bit more on the auto industrial side.
Okay.
Mm-hmm.
I wanted to go back to, your strategy in 2021, 2022 when things obviously were very tight, 'cause we're seeing a divergence of some other companies, with revenues drying up, you know, we model you up high single digit for this year at the top line. Some of the companies are going to be, some of your peers are actually going to post decline this year still.
Mm-hmm.
I know that you've, in 2021, 2022, you've said you traded some pricing against market share.
Mm-hmm.
I think, you know, a few other companies have been pretty conservative with pricing. I think TI is one of them. I know others that have been much more aggressive on pricing.
Mm-hmm.
Does that position you differently from some of the peers that raise pricing more aggressively in the upturn because you have less downside in pricing or any other metrics that maybe you can mention? You know, how, what is the customer feedback on your approach over the past several years? What does it mean in terms of market share and, you know, going forward?
Yeah, definitely. I believe a few years back, probably on similar stage, I talk about long-term benefit is more important than the short-term gain, right? At that time, instead of raising prices, we actually work with the customer, gaining some of the new products, designing additional products onto the board, and build a stronger relationship. I think, you know, throughout the years, I think at the time, probably a lot of people questioning about me, why we think that way. You know, my philosophy at the time is like, if it's a deep commodity part, if you don't have product, you know what you're gonna do? You're gonna pay whatever the price you can get. When the supply coming back, you're not gonna pay the same price.
If I use opportunity or leverage the opportunity to design in products onto the board, even if you do not have a demand right now, when the market comes back, I will actually have opportunity to gain the market share. That is really the strategy we put in place during that time. I think now we are actually seeing good results and also payoff. You know, not only do we have better print position, but we also build a much stronger relationship with the customers. I think customers definitely are looking for solid suppliers who can supply them multiple devices in different product categories for the long term. I think, you know, that actually shows that we are a solid supplier, that we can actually partner with the customer for the ups as well as for the downs.
In terms of pricing, do you feel that trends are fairly normal, or are we still in the midst of kind of retracing back to pricing and, you know, maybe back to 2020 or 2021? How do you see those trends and how you're comparing with peers?
Right. Before COVID, we actually built in a model, 1.5%-2% quarterly price erosion. When people ask me the question about pricing, I always say it is more stabilized. It does not mean we are not dropping the price, but it is within the 1.5%-2% range that we define. That is really more what stabilized means. I would say, you know, overall, we are seeing a lot of stabilization overall in the pricing. We are seeing a lot more new designs and the pipeline continues to improve. I would say that is what we are seeing. People probably focus a lot more on the functions and features. I think at the same time, people talk about risk mitigation, agility, flexibility. I think that is really where some of the areas that Diodes can provide a really good value.
Great. I think a couple of years ago here, you had mentioned, you know, there was some level of competition from local Chinese supplier, and it's been more of a theme, I think, over the past few quarters. Is that level of competition stable, or is it getting tougher in China against local players over the years? How do you feel you're positioned against that?
Yeah. I think the competition is actually good, right? I mean, we live in a world, there's definitely competition. Whether it's Chinese competition or American competition or European competition, I think that's part of our life, right? I think, you know, in China specifically, we've seen a lot more stabilization, right?
Okay.
There's once in a while still some of the names, even I'm Chinese, but I can't remember, I can't pronounce it. But overall, right, I personally believe there's just so many. There's probably gonna be more of the consolidation going on, which also, you know, verified and agreed by a lot of my customers. So, you know, I think over years, they are improving. The key for Diodes is we need to improve faster too, right? That's what Gurmeet mentioned earlier with new product introduction. If we don't grow, if we don't push our technology to the next level, you're always gonna have someone at the bottom chasing you, right? So, you know, I view this as part of our life. I don't really think we can say no more Chinese competition.
I also view that with all this localization, tariff restriction, it's actually a benefit overall for Diodes getting into some of the new customer base, expanding some of the new opportunities in different areas as well as different countries. All in all, when I look at China market, I think, Tristan, I told you this, China market's so important. We definitely will continue focus in this area. What we've been actually changing our focus is actually focus more on the differentiated, more higher-end product in China market. If my focus continue to be commodity and deep commodity.
Yeah.
That just writing on the wall, there's no future. So, we are adjusting. One good example, for example, Pericom product family, in China, we're still seeing a lot of momentum because it's on higher protocols, higher frequency devices that we don't really have a lot of Chinese competitors in this area.
Mm-hmm.
How does your local production in China, in Shanghai, match the local demand? I don't know if your local demand in China is about 15% of your revenue. How does that compare with the percentage of production that's actually coming from China?
Yeah, so, that is right. It's about that percentage, and it's in the teens. I think the production coming from China is hard to align with the local consumption because it's like Emily mentioned, you know, there are different products that are being sold in different applications and different market segments. I am not sure if we can, you know, say, okay, 10% of local production is being consumed in China. That's not the case. As you know, our assembly and test footprint is heavier in China compared to the other locations. That is the backend, which is in China.
The front end, as we are fully very well diversified, we have a fab in the U.S., we have two fabs in Europe, one assembly and test in Germany, and then assembly and test in Taiwan, and the fab in Taiwan, as well as in China, fab and assembly and test. I think we are very well diversified from the manufacturing footprint to serve the customers and be closer to the customer where they are. I do not have the number that I can say, okay, X percentage from China production is consumed in China.
About, maybe 13 months ago, you announced that PCIe packet switch platform went with NVIDIA, which I think was unexpected and speaks for some of the differentiation in terms of products that you have. I know GB200 has been very slow to ramp, so probably contributed less in the past few quarters than the initial expectation. How should we look at this in the second half and how material does it get, notably as it is probably higher margin and notably as potentially it cascades down to other data center platform and maybe even traditional data center?
Yeah.
As much feedback on this, I appreciate it. Yeah.
I'll start, and Emily, you can add to that. As you know, as you alluded to, we did mention that we are in the GB200. We are working with all of the hyperscalers. We're in all of the AI servers and also in traditional servers and data center server storage applications along with the compute, other applications such as notebook and desktops. What we are saying is it's not just the packet switch that was one of the incremental products, which is still there and is being designed in and used. We also have, as we, in the traditional server, we had about $60 content in the box. With the AI server, it's $90 content per box.
A lot of it is in addition to the packet switches around the signal integrity, Mux switches, IO components, power management, and even discrete. I think we will continue to work with the likes of NVIDIA and other partners on the reference design, and especially as the AI moves to the edge computing, I think that's even a bigger opportunity for us.
Yeah. So let me just add, right? Not only GB200, GB300, the next generation.
Yes.
You know, the demand, right, switched from GB300 back to 200. I think the good news is it does not matter, right? Because we are on both. So, you know, like Gurmeet mentioned, NVIDIA is just one out of a few hyperscalers. What we think, thinking even more exciting, is actually when the AI is getting into the edge level, getting into the compute, getting into the smartphones, and that is actually going to help a lot more, drive a lot more volume, and also with the refresh cycle. I think that is really where we are seeing exciting. So I think from our point of view, not only the packet switch, and also this is just the beginning of the deployment.
Is there a way to frame this in terms of the TAM, the CAGR, maybe three- or four-year CAGR, and then the market share that you see in those circuits that you've won so far?
Yeah. I think I don't really have the TAM or the SAM classified or the CAGR at this moment. The challenge is actually it really depends on the customer's design. Not every customer is gonna need a packet switch. Not every customer has a different power block that they are designed. It varies a little. It's not easy to say, "Hey, if this for sure, this is a volume, this is a percentage," because, you know, NVIDIA is gonna be different than the AMD platform, is gonna be different than the Qualcomm platform, or, you know, a different one, right? That's really where the challenge is. I think the exciting thing is the potential, right? Even it's not narrow to a specific application, but it can be very broadly used.
For example, on the networking side, we're actually seeing a lot of new designs, a new requirement. Basically, the packet switch you refer to, the PCIe packet switch, is anytime you need additional PCIe ports to support a Gen 3 or Gen 4 speed, that will be a perfect use case. It can be everywhere.
Okay.
Mm-hmm.
It's beyond compute. It's we're seeing in automotive networking as well.
Right.
Any other product that you think have a lot of potential? Pericom, obviously, has been really a success story in data center. It's now moving into automotive. Maybe you can talk about this a little bit. Anything that's incremental that leads you to believe it, it's, you know, we should look at this in terms of additional share gain or mix gain going forward?
From.
In terms of new products.
Pericom product line point of view?
Yeah. You know, besides that PCIe, you know, packet switch discussion.
Yeah.
What are some of the products that, yeah?
As you know, Pericom product line is heavy in the compute segment. In the last 10 years or so, we had the opportunity to expand that, those products into other applications, which is, industrial is one of them with a lot of embedded computing and that power in the industrial segment. That's an extended market segment for Pericom product line, as well as automotive. We talked about not just the packet switch, but also signal integrity, right? Ten years ago, we never thought there was gonna be Ethernet in the car or you'd need a signal integrity or redrivers and HDMI ports in the car. All those are Pericom products that are going in the automotive segment. Also, if you look at our timing portfolio, which is part of the Pericom product line, that is expanding, right? We have both sides.
There's the clock IC, which is clock generators and clock buffers. We are the, you know, we have a very broad portfolio. I would say we are the leaders in the PCIe clock generators. We have, last year we announced PCIe 6 clock generators and clock buffers. In addition to that, we also have crystals and crystal oscillators, which is for the, that does the complete timing tree. We have from crystal all the way to clock ICs, generators, and buffers. I think if you look at it as timing is everywhere, regardless of the market segment. I think that also is expansion for Pericom product line.
Great.
Yeah. I think let me add, right? I think overall, we're looking for smaller size, lower cost, smaller package, sometimes higher voltage, better performance, right? We're talking about the speeds increasing from Gen 1 PCI Express to Gen 6 now. We're talking about USB Type-C power delivery. All this is actually creating opportunities for us, probably more on the Pericom side. On the other side, we're looking at power management, right? You know, that's quite broad across all applications, but everybody is looking for some power management, right? You know, automotive, we're talking about from 12 volt to 48 volt. All this changes is actually creating new opportunity for us.
Are you positioned as well in, in data center for power management? 'Cause that's a big bottleneck. And, I, I know that you're working on some silicon carbide technology. I don't know about gallium nitride. Maybe you can give us a quick update on that, for power in data center and, and also the rest of the market.
Yeah. In general, as you know, we've been working on silicon carbide for the last few years. We were not probably as aggressive. We were more conservative, which is good that we had introduced 20+ silicon carbide diodes and then MOSFET products too. We are building traction both in the industrial as well as in automotive market segments. That focus will continue, as we are bringing new products and continue to develop the technology. In a way, we feel like, you know, our approach was correct. It was not as aggressive, but we are aligned with the market, and the market demand along with our technology. In GaN area, we are, we're not doing much, but mostly in silicon carbide, but we're focused.
Yeah. I would say for the GaN, we are monitoring very closely.
Yes.
You know, definitely, I think at this stage, more understanding the technology. We're not at the time releasing product at this moment.
Okay. We have a few minutes left. Any questions from the audience? Otherwise, I'll keep asking questions.
If you mentioned the edge, once, you know, if the idea goes to the edge, are they there or is that a year from now? Any idea of when it will be, you know, pervasive on the edge that benefit your cloud?
I think everyone talked about they have some designs going on, right?
Yeah.
I think what we see maybe earliest would be end of this year, maybe some of the product. I would say majority is probably still gonna take a few years.
Yeah.
You know, this is still at the very early stage. A lot of people are still looking at the, you know, the case study, right? How to really implement that, right? All in all, if you really talk about, you know, big data, everybody need more compute powers, right? Everybody probably need a lower power to minimize the energy consumption. All this is in the right direction, but the actual product probably still gonna take a little bit of time.
All right. Thank you.
Mm-hmm.
Quick question on gross margin. Less than two minutes. So clearly, utilization rates is a nice driver. You've talked about raising the percentage of production that you wanna do internally.
Yeah.
And notably the South Portland, Maine Fab. You have also talked on the last earnings call about, you know, some, you know, timing on, the, the qualification, of those products. What does that mean in terms of gross margin medium term? I'm assuming as you move production in-house, you get some company-specific levers on gross margin that you would not have otherwise. Any other commentary on that?
Yeah. As we talked about in the Q1 earnings call, you know, that is one of the things that is, underutilization is impacting our margins. I think there are certain things we can do. There are some things we can control from the utilization and margin point of view. The others depend on the market environment. I think a couple of things. One is the utilization, underutilization. Currently, we have hybrid model, which is internal and external. I think we are around 55% internal, 45% external.
Oh.
As we are, you know, newly acquired fabs that you mentioned, especially the SPFAB and the GFAB , we are qualifying our processes in those fabs. As we are continuing to bring, qualifying our processes, bring our products into these fabs, working with the customers. Emily's team is working with the customers on PCNs and product qualification, and then eventually we'll ramp up. I think it'll take some time. We are in the middle of that process as we go through that. That's one of the levers that could help us with the gross margins, the improving utilization. Hybrid mix can shift, right? It could be higher than what our current mix is. The other one is the product mix.
As you have heard Emily talk about it earlier too, and it was evident in our Q1 earnings, is that our product mix is improving, and we will continue to focus on that, bring new products, higher margin products. The third one I would add is the market segment mix, right? As we continue to focus on automotive and industrial market segments that give us higher margins, that mix is also changing as well as compute. The other one is our analog and discrete products. Those products, regardless of the market segment, we also get higher margins. Three things are the utilization, market segment mix, and product mix that we can control and we are working on to improve margins. The fourth one will be the market itself.
Great. Thanks for being with us today.
Thank you.
Thank you.