Power Integrations, Inc. (POWI) Earnings Call Transcript & Summary

Good afternoon, everyone. We're very pleased to have the management team from Power Integrations here with us today: the CEO, Balu Balakrishnan; the CFO, Sandeep Nayyar; and the Head of Investor Relations, Joe Shiffler, also known as the best dressed management team in semis today. They even made me put my jacket on.

So guys, thank you very much for attending and joining virtually. First question I have is somewhat specific to you guys. But if I went back to the beginning of this year, you were amongst the most cautious management teams. Some of it was company-specific and end market-specific. You were worried about excess orders, double orders, inventory, et cetera. And you upsided those cautious numbers in the first quarter and second quarter. And then the most interesting thing to me was you flipped around and became maybe not the most bullish, but a lot more bullish than you were before. So what changed?

There were 2 things that changed. One is that we were expecting the Huawei normalization to happen in the second half. And it actually happened to quite a large extent in Q2. There's still more to go, but a lot of it happened in Q2. At the same time, we saw that consumer and industrial markets came very strongly up in demand, driven by our share gains in those markets. And the same as the computer. So we were, I guess, pleasantly surprised on both the share gains and the demand increase in appliances, which was significantly more than we originally anticipated. So in essence, we still believe the second half will be lower but not as low as we originally thought. And of course, the consumer part of it still has work-from-home demand in it, which will eventually normalize. But the amount of share gains we're getting will really soften that normalization going forward. That's the reason we've become more bullish, if you will.

So that all makes sense. It seems like demand is stronger, and some of the supply was digested. How do I interpret the channel inventory went down to like 2 weeks, and now it's back to 5. It seems like some of that would be more inventories coming through, not less. So am I -- I must just be confusing something. So can you help me out there?

Well, it's -- a lot of it are straightforward. The inventory build, the distribution -- increase in distribution was almost entirely driven by cellphone -- China cellphone market. That's all it is. We shipped what we thought the customers needed, but they didn't pull it through. And that's part of the correction or normalization that happened in Q2. All other markets are very strong.

Got it. So why don't we -- so it's great that you guys have shifted over to the more bullish side. We'll talk about some of those end market drivers, technology drivers as we look forward here in a bit. But before I get on to that, I want to just talk about the supply side of the equation. Talk a little bit about any limitations that you are seeing, whether they're direct or indirect. Is that playing any part in your second half being below your first half? And then the eventual question, is there a difference between the GaN and the silicon side of things in your supply chain?

Well, at a high level, we have enough supply to meet the real demand. It's as simple as that. We have definitely enough supply for that. Now when you go into the details, there are some mix issues in some cases, but they're relatively small because customers can never seem to be able to forecast that mix accurately. And so other than that -- and also there are some supply chain issues in Malaysia and Thailand that are talked about quite a bit. And that, again, it doesn't affect our overall capacity. But to the extent there are some products in a specific location that gets shut down because of COVID, we do have -- one of our locations has been shut down for a few days. It came back up. May, shut down again. But the good news is most of those products are also second and third sourced in other areas of the world. So overall, capacity is okay. But when things happen suddenly, there are short discontinuities. But I would say that we are in a much better position than almost any other semiconductor company because we have captive capacity. And as I said, the overall capacity is not a problem. It's just when there's -- when things start shifting, there is a delay in time. And I think the...

And Ross, the only other thing I would love to do is build the inventory that we have. We are at 92 days, and we want to get back to our thing. So that's the one area. Even though we can meet demand, we would love to build as the opportunity provides us.

So on that note, I guess, 2 follow-up questions. Balu, when you said you're able to meet real demand, sometimes I hear that from CEOs and what they really mean is, yes, there's a bunch of double-ordering, but we're going to shift to what the customer truly needs or your belief into what the customer needs. Is that what you're implying by saying true demand?

Absolutely. So we talk to customers, at least our largest customers, and we tell them not to build inventory because it's much better to keep all the inventory in one place so that we can serve all of our customers better. And we have good visibility into the chip inventory they have. We don't have visibility into the further down the chain, whether they have power supplies built up, watt chargers built up, that we don't know. But certainly, in terms of making sure that we are supplying to keep their production running, I think we are doing a pretty good job. Is it perfect? It's never perfect because sometimes customers don't tell you the whole truth. But I believe, overall, I mean, if you talk to our customers, they will say we are one of the best suppliers that they have today.

So last question on this general topic, then we'll move on to some of the sexier GaN technologies and things like that. But Sandeep, you mentioned a bit ago about you want to regrow the inventory. And that makes total sense. I think last quarter, you were down closer to 90 days, and usually, you have 110 days or so as your target. Without being too much of an accounting nerd, I was shocked to see that the raw material side hasn't grown in about a year. Was that just a reflection of your initial caution...

No, the reason is you look at...

That we talked about earlier?

No. The reason is that if you look at it, we had grown our inventory last year, and most of it was in raw material. But with the demand really coming strong, that build that we did, and we kept it in wafer form, really has enabled us to meet the demand that Balu talks about. And that's why it is a tight situation. I would love to build it. But we actually -- the reason we have been so successful is we think long term, we didn't want to jerk the foundries. And we let our inventory go to 180 days last year. It was all in wafer form.

Okay. So -- but going forward, now we'll see the raw material side ramp up?

Well, gradually. But again, I'm saying because the demand is tight, it's going to take a while to get to our model, which is what I would like to do.

Yes.

Got you.

Every time we build inventory, we try to keep it as -- keep as much of it is in wafer form, simply because it gives us more flexibility to build different types of products.

So you guys don't even put it in die bank? Is that what you -- you don't even package it at all?

Well, die bank.

Same thing?

It is die bank.

Okay.

We won't package it because packaging cycle time is relatively small. It's about 4 to 6 weeks, whereas the processing time is much longer.

Right, right. So let's switch over to the GaN side of things. POWI's been a great innovator over the years through all kinds of different integration techniques versus some of the linear transformers, all kinds of different things historically. The latest integration path that you're doing is on the GaN side. Talk a little bit about what GaN brings to the market, how that solution compares to the pure silicon version, and then I'll get into some kind of competitive base questions after that.

At a very high level, GaN is far superior to silicon in terms of the power switch. The controller is still silicon, just so that you understand, a lot of the control functions are silicon. At -- as a high-voltage power switch, it is far superior not only to silicon but also to silicon carbide at the voltage levels we operated, which is below 1,200 volts. And the advantage to the customer is that it allows you to build a much higher efficiency power supply, which means you don't need a heatsink on the power supply. You can make the power supply a lot smaller because you can go to higher frequencies. And the combination of all of that is that it gives you a much more compact, lightweight power supply. It also has other advantages like, for example, it can handle transients much better than silicon can. So if you are in countries like India, in lightning-infested areas like Florida, it truly is a much more reliable technology. But lastly, I truly believe, in the long term, GaN will be very competitive with silicon. Right now, it's more expensive. But in the long run, we can see that happening. It's very different on silicon carbide, which is fundamentally more expensive to process.

So the GaN side is not going -- it's going to be just scale to get GaN to the price points of silicon. It's not -- there's no fundamental barrier beyond the scale side?

Well, I can't talk about other people's GaN technology because we have a very unique GaN technology, which I believe is the most cost-effective GaN technology in the world today. And based on our technology, we are confident it will become -- it will replace silicon above certain power level, meaning about 25, 30 watts.

So POWI's business model has been based on a lot of unique technologies, manufacturing, design, integration. But when I think about those benefits for the GaN side of things, you always had priced the bill of material equivalency to what you replaced, but you were -- it was a simpler design process, a better performance in aggregate for that bill of materials, that system. Is that the same strategy that you're implying in the GaN side?

Exactly the same. Exactly the same. In fact, at the system level, we are already cost-effective because there is no way to make the power supply as small as they are able to do it without GaN. So we price it on a value basis. So we look at what the system cost is using our solution versus using an alternate solution, and we base on that. But when it comes to small size, there is really no other solution that can make the power supply as small as we can with a combination of not only our InnoSwitch products, but also our MinE-CAP product and the ClampZero product. If you combine all 3 of them, you get the smallest form factor, lightest power supply and the highest efficiency power supply. There's really no other technology today that can beat that.

So when you put that all together, is the bill of materials equivalent for your customer? Or is there actually a cost increase that, that customer has to bear to get the better performance, whether it be size or switching speeds that GaN will enable?

If you need the smallest size, it is very cost-effective, there's no question about it. If size is not a constraint, silicon can be cheaper in some applications. But what is really fascinating to us, in applications where we thought people would not pay the higher price are now paying the price for reasons we didn't imagine, which is like, for example, appliances, we have a number of design wins in appliances. And you would think, why would anybody care about size or weight in a washing machine or a refrigerator? It turns out they don't care about those things, but they do care about efficiency, which is significantly higher than silicon. But the other thing they really care about is the lack of heatsink. What they're telling us is when they ship a refrigerator or a washing machine or a dishwasher, the biggest failure mechanism is somebody drops it and jerks it during transition, and that causes the heatsink because that's the heaviest thing on the board to fall off or break the board. So they think that alone is worth paying the premium to use the GaN solution, but they also love the efficiency because that's always a driver in all of the consumer products.

So if we wrap up the GaN conversation, it's -- I think it's, what, $25 million to $30 million in revenues is what you guys said it would be this year?

Right. Yes, we did about roughly...

And I'm not trying to put words in your mouth, but...

No, it's roughly about $10 million last year. We said about 3x this year, yes.

Right. How big a percentage of your business do you believe that will be? And what sort of revenue growth rate do you consider likely going forward? If it's tripling this year, what, $30 million out of your total isn't a huge percent, but it's growing really fast. How are you thinking about the revenue growth rate going forward?

I never thought we could grow as fast as it did because of the higher price, but it has always surprised us. I wouldn't be surprised if it doubles again next year. Can we do better than that? It very well can. We certainly are putting enough capacity in place to be able to handle that. It's actually been quite surprising how well received our GaN solution is. I think the biggest reason it has been well accepted is because it's the ease of use. They can't even tell there is GaN. Whether it's silicon or GaN, our chip looks exactly the same. But you get the higher performance, but you don't have to deal with the idiosyncrasies of a discrete GaN, which is really, really hard for people to design them.

And the last question, I guess, on this, the margins for these products still more dictated by the end market they serve and all the traditional dynamics, Sandeep? Or is it -- are there some GaN-centric items that people need to appreciate as that rises in your mix?

Well, the end market will be...

Fundamentally...

Go ahead, Balu.

Well, fundamentally, it just depends upon volume concentration. There are markets like cellphones where you have very, very large customers. And therefore, it's more competitive because even the alternate solutions are priced lower, if you will. So we always value price our product. And so when you go to more fragmented customer base, we get higher margins. So that in itself doesn't change. But what is happening, however, is that even in the cellphone market, a portion of that market is going to aftermarket customers who are all very fragmented customer base. So our margin in cellphones is actually increasing now as the aftermarket is becoming a reasonable portion of that business. It's about 10%, 12% right now of the cellphone business, but that's much higher margin than the higher volume customers. And really, that's all differentiated. The differentiation is based on customer concentration, which happens to be more in cellphones.

Sandeep, is there anything else you wanted to say on that? Or is that...

Yes. Ross, I think the key part to look at it is it's accretive to our operating margin because of the higher ASPs. So that is really helpful to our operating margin, in addition to what Balu has already summarized for you.

Got you. So if I pivot over to your end markets and talk about communications, that's probably about 1/3 of the company or so. And by my math, it's roughly $250 million. So I think that's that 10%, maybe 12% of revenue you were talking about, Balu, is the GaN side of things. Is that going to remain the primary beneficiary of this GaN dynamic with the ASPs rising for the foreseeable future? Or do you think in 2022 and 2023, we're really going to see benefits in other segments where that 2x ASP or even higher will start to be a tailwind for you?

Well, as I mentioned, we truly believe we will replace the silicon switches with GaN going forward. I mean, we are certainly targeting that in all of our new products, I would say -- most of our new products, I should say, because very low power levels, again, doesn't make sense. You can't even make a GaN small enough to put it into the package. It's so tiny. So about 25, 30 watts, it's got to be all GaN. And so it's not market-specific. What will happen is that some markets will adapt faster like cellphones and computers. The consumer market will take a little bit longer to adapt, and the industry will take a little bit longer. Eventually, all of them will use GaN. So just a question of design cycle, right.

And generally speaking, is there a rule of thumb about the ASP boost that you get with GaN if you want to make it kind of application-specific or end market-specific because I know there's a wide array of prices depending upon the power levels you're addressing and the end markets?

Yes. It's -- it -- well, it really depends upon a lot of factors. But let's say, for example, the easiest one to talk about is cellphones, right? There if -- there are several things that increase ASP. The power level increases the ASP. The complexity of the part increases the ASP because now you have USB PD charges which have quite a bit of complexity on the secondary side, which we actually integrate. The efficiency dramatically increases the cost because then you need to synchronize electrification, the new ClampZero and so on. The size increases ASP because then you can use our MinE-CAP and ClampZero products. And if you have multiple outputs, that can increase the ASP. So if I take the really the best case, which is like, I said, 2 output, say 100-, 120-watt power supply, now our ASP is in the low single-digit dollars, okay? Whereas if you go to the other extreme, where you go to like, say, 30, 20 -- 30 watts or below, then it's, of course, well below $1 -- below $1, let's say. It just depends upon the wide range of power supplies -- the wide range of applications, I should say.

Right. So it's nicely accretive to ASPs any way you cut it. But...

Yes.

No single rule of thumb that investors can use for some sort of multiplier effect.

Right. And by the way, it also depends upon how many chips they use. So for example, if they use a MinE-CAP, that adds more ASP, which depends on the size of the product. So the range is roughly what I told is about 3 to 5 weeks kind of range compared to silicon.

So the last communications and GaN question I'll ask is one about the competitive landscape. This market seems to be growing very, very rapidly. There don't seem to be too many players that can touch the GaN side of the market. But generally speaking, how do you view your competitive positioning in GaN versus where it was in silicon?

I think we are way ahead of competitors in GaN. I can't think of anybody who has the type of technology we have, a very cost-effective technology. They don't have the level of integration we provide, the ease of use we provide. It's a lot easier to throw a discrete GaN at the customer than to build a complete system that's really, really easy to use for the customer. That's why I think our GaN has been adopted so quickly. In fact, we were shipping GaN for quite some time before it was discovered it was actually shipping GaN. We never even announced GaN because we don't sell technology. If you're a small company, you have to sell the technology. What we sell is a system-level solution that's easy to use, and there is no reason for us to advertise the technology. But of course, now everybody knows it, so we talk about.

So let's take a step beyond the end market specifically and even beyond the GaN technology. We've had discussions at this conference in the fall of many years about your 4 main segments, which ones you thought had tailwinds or headwinds as you looked into the out year. As we look into 2022, given how we started this conversation, your optimism on true demand and product cycles, et cetera, it seems like you're pretty confident that all 4 of your segments will grow heading into 2022. First, is that true? And then second, walk me through the hierarchy of the segments and then -- and the drivers of that growth.

Okay. So let's start from the smallest one, which is computers, where we have gained significant share in tablets and notebooks. And in notebooks, we have a long way to go. We have a relatively small share there, but we really see a trend towards smaller chargers. People have gotten tired of these bricks that come with sleek notebooks, and there is definitely now focus on the size and weight of the charger that goes with the notebook. And it's, in fact, driven by cellphone chargers we have done to USB PD, and USB PD has been much more widely accepted in the notebook market than in the cellphone market. But the same charger that you use for the USB PD cord for a cellphone can be used for notebook now. So we see a significant growth ahead in the market. We are also going into monitors, which will also help us grow that market. And we have grown very nicely. I think it will continue to grow. In cellphones, we have a pretty large -- we have -- I believe we have the largest share of the cellphone charger market of any company in terms of the power supply chains. And our ASPs are much higher than our competitors. If you just look at it from a dollar share, we are already above 50%. But if you look at unit share, we are below 50%, which means we have a lot more room to grow. And we have such compelling products. We are finding that many of our competitors, they are deemphasizing this portion of the business because they just find it very hard to compete with us. So we see a lot of growth there. Then if you go to consumer, there are a lot of things happening there. The power levels are going up, thanks to all the features that go into appliances these days, like WiFi, like web connection, monitors, LED displays and so on. As a result, our content, our SAM, dollar SAM is increasing significantly. But on top of that, we also have the BridgeSwitch, which allows us to address the motor control market, which is a total new market for us, we think is about $0.5 billion of addressable market with the power range we offer. We plan to expand that power range so that, that will get to more like $800 million. By the way, when we say SAM, we truly mean SAM. We don't throw around numbers that we -- some of our competitors do, billions of dollars. We truly mean this is what we can address if people use our products everywhere. So that's going to do very well for us. And we are already seeing a significant share growth because our biggest competitor there is actually moving away from the market. Our second biggest competitor has moved away from the market and has actually requested us to take over their customers. So that's how compelling our products are. Then if you go to the industrial space, the fastest growth we are seeing is in home and building automation. These are things like voice-controlled home appliances. These are things like USB wall outlets, smoke detectors, door locks and so on and so forth. And that's grown extremely well for us. I think year-over-year, year-to-date, it's almost 200% growth, which is phenomenal. Then the next biggest one is the tools, power tools. These are things that used to be electrically operated. Now they are battery operated like your floor appliances, your -- what you call lawnmowers and leaf blowers and so on and so forth. That's done extremely well for us. And on the high-power, because of COVID, we haven't -- we have been relatively stagnant for the last year or so, but we see that slowly coming back. But in the long run, I believe high-power is going to be a huge market for us because there is so much investment going into infrastructure go to carbon net zero. The estimation is that between the major countries, there will be about $56 trillion invested between now and 2050 to add more capacity for renewable power, to add infrastructure for electric cars like chargers and so on. Everything is coming towards us in that respect. We really specialize in high voltage, and we see that this infrastructure spending is going to benefit us. So we see a very, very good growth going forward. We are also going into automotive, which is in the industrial space. And even though the revenue is very small, and it's going to take a gradual growth for the next 3, 4 years. But after that, we see a much more significant growth. And we already have significant pull on the products we've already introduced, which is like InnoSwitch, LinkSwitch, our Qspeed diodes and our iDriver. In fact, everybody we have shown the InnoSwitch immediately wants to use our product. So the reception is very, very positive. It just takes time to get the revenue growth. So you can see that there's a lot of stuff going on. The whole world is going green, and that's good for us.

Any headwinds for next year? Like not macro stuff or subjective, but within a given segment of your business, you're a little bit concerned on something? And if there isn't one, that's fine.

I can't think of any. Sandeep, can you think of [indiscernible] -- go.

Yes, the only thing I -- and I think is the normalization, if any, from work from home on the appliance area. But I believe, with the market share gain, that will kind of offset that headwind, but I have to be upfront that we have seen the normalization in the cellphone, and we have seen phenomenal growth in the appliance area because of the upgrade cycle that people have done because they are at home. And there's a possibility of a little normalization there.

And where would that reside by your end markets? Because I -- some of it seems to be in consumer and some of it seems to be in industrial, given your description a bit ago, Balu.

No, it is in consumer. So we put all of the appliances and the air conditioning, we call it comfort appliances for air conditioning, that's in consumer.

But like power tools and things like that, that's all in...

Those go in industrial.

Okay.

Yes. You could argue some of them could be in either place, but that's how we have done in the past.

Yes, yes. There's no right or wrong, I -- but I thought that they could be in either place. So I just wanted to understand that. So...

We've had a lot of our peers put the appliances in industrial because it really behaves more like industrial, long-term life cycles and so on.

For sure. So why don't we, in the last couple of minutes we have, switch over a little bit to the profitability side of things, the spending side. How do you, Sandeep, view the gross margin puts and takes as we go forward into the next couple of years into your, I think, 50% to 55% target range. Historically, that get pulled to the bottom when the communications side was doing really well. But it seems like you have a much more diversified revenue stream, not only within communications, but in overall revenues for Power Integrations. So I would assume that's a tailwind to your gross margin going forward. Am I missing something there?

I think the model of 50% to 55%, I think, holds well. And if you look at the long term and not look at short term, we've been in the mid area of that. And yes, because cellphone in the past pulled it down. But as you and Balu have attributed earlier that we've got aftermarket, which is getting the margin up. It's no longer a commodity. It is more a value proposition. I think we have continued to say and I think continue will -- is to optimize our operating margin. And if you remember, historically, we used to talk about getting to 25%. With the revenue growth we have, we've already outpaced that. It's -- but right now, the revenue growth is so high, but we have to catch up with our expense growth, which, as you see in the models, we are going to be spending more than the revenue growth rate next year. So -- but in the near term, we still think our operating margin will be north of 25%, between 25% and 30%.

So you kind of front ran my final question in that was just the OpEx side of things, it's -- I -- I'm always hesitant to make it relative to revenue growth for next year because no company is going to guide to revenue growth for next year, but you have to start planning the spending.

Yes.

So how -- is there a way for us to think about OpEx growth next year without linking it to revenues?

Yes. I think we would grow in the low teens in spending next year. And the reason is we have so many ideas. And we -- as we have talked about, we really want to have our reach increased. And as a result, we have a lot of headcount that we are adding between Q3, Q4, Q1, Q2 next year, which will have a full year impact. And also, we are hopeful that by Q2 of next year, we are back to normal with things like travel and events, which also are important from a customer reach standpoint, we would start investing in. So I think next year, the expense growth, I think, should be in the low teens.

Got it. Well, guys, we are exactly on time right now, nearly the 25 after. So thank you so much for attending the conference. I don't mean it jokingly. You also are the best dressed team at the conference. Making me put my coat on just to fit in. But I really appreciate it. Congratulations on a really strong year, and I look forward to monitoring the growth continuing into 2022.

Thank you, Ross.

Thank you.

And thank you for inviting us.

Thanks a lot.

Have a great day.

Thank you.