All right, let's get started. Good morning, everyone. Welcome to the Power Integrations session here at the Stifel 2024 Cross Sector Insight Conference. My name is Tore Svanberg. I'm a senior semiconductor analyst. I cover analog processor and connectivity companies, and it's my pleasure to introduce Power Integrations today. With us to my immediate right is Balu Balakrishnan, who's the company's Chief Executive Officer. To his right, we have Sandeep Nayyar, who's the company's Chief Financial Officer and VP of Finance, and we also have Joe Shiffler, Director of Investor Relations and Corporate Communications, here up front. The session here is gonna be a fireside chat.
Before I get started, just to sort of keep everyone on the track that's been here throughout the morning, we started the day with Texas Instruments, the largest analog company in the world. We then had Cirrus Logic here, which was the first fabless company. We then had MACOM doing high-end RF, and they're definitely working quite a bit on GaN. And now we're gonna talk about power management, especially from a high voltage perspective. And certainly, GaN is gonna be a very important part of this session. So with that, we'll get started. Balu, Sandeep, Joe, thank you so much for coming. Really appreciate it.
Thanks for inviting us.
As we do with these sessions, we sort of start off very light, and we ask you to do a sort of two to three -minute introduction to Power Integrations, especially for those that may not be that familiar with the company.
Yeah, I see a lot of familiar faces, but just for the rest of you, we are in high voltage semiconductors, which is a very unique space. The way to think of us is, if you look at carbon net-zero initiatives, there's gonna be about $100 trillion spent over the next 20 years to get that to, you know, the Paris Accord levels. But if you look at where that reduction is gonna come from, the number one is from energy efficiency. It's about 37% of that is expected to come from energy efficiency, which is our major play. Number two is renewables. I know there's a lot of talk about EVs and so on. If you look at the contribution of the EVs, it's really tiny. So we are not very greedy.
We just need a small portion of that, hundred trillion dollars, I think we'll do great. So those are the, you know, 67%-70% of that comes from those two areas, and that's where we play. And to achieve that, you do need high voltage semiconductors, and, our focus is to be able to get as much out of that investment as possible.
Great. Thank you for that, Balu. Let's start with cyclicality. This is a question that I asked the other management teams prior to you as well. I mean, this cycle has been the craziest cycle we've ever seen, and we've been doing this for a long time, right? Each cycle is different. This one was crazy. Maybe a question for both of you: What are some of the things Power did right during the pandemic years, and what are some of the things that, you know, if you had an opportunity to do it again, you would have chosen? You know, 'cause I'm sure there's always a few things that you learn when you go through a cycle like that. So, yeah, maybe, Balu, you can start.
You know, I've been in this business for a very long time, seen many, many cycles, and they're all different, as you say. But one thing is very common to us. That is, every downturn is an opportunity for us to grow our share and be different than other companies. During downturns, most companies just retract, whereas we say, "Okay, my God, you know, this is the time to hire people. It's the hard time to invest in new products. This is the time to go get more share." And that, that concept has worked independent of the type of, you know, downturn we have. We have always, you know, benefited from the downturn because of that. This happened in 2008.
It happened when, you know, the bubble burst in 2001, and it's happening now as we speak. But if you look at what happened during the COVID times, when the whole market was going downhill because of COVID, we built a lot of inventory, and that helps us, you know, it helped us to supply products to our customers better than almost any other company. In fact, I don't know of any other company who were able to supply the entire demand. What we learned from that was we shipped too much. Because we were able to ship, our customers decided to build a lot of inventory, and we are kind of paying the price for it in the last two years. That's kind of the bad news. The good news is I think it's over.
We always come out of this recession before other companies do, so we look forward to a, you know, growth from here on. The big question is how fast it's gonna grow, and that's something I'll leave to you since-
Sure. Well, I guess, I guess even the better question is, you know, when are your customers gonna learn? 'Cause I remember you telling me a story about a customer where you said, "You don't need the product. I'm not gonna ship it to you.
Yeah.
They insisted, and then I think a couple of months later, they cancelled the product. So-
Yeah
... I guess that's the better question.
Yes. That is. Yeah, it's a big customer of ours in, let's say Korea. And I will tell you, we knew we were overshipping, and they, the senior management, were there. They said, "We need the parts. You shall ship." How do you say no? So we ship, and then later on, literally two months later, they decided they have too much inventory, and they literally shut us off for a long time. Of course, they're back on now, so we are happy.
Right. And, Sandeep, would you add anything, especially on the finance side?
Yeah, I really think it's just the way during this period. It's a perfect... as Balu indicated, how we use the capital allocation. During this period of downturn, we are investing in working capital, we are investing in our people. I mean, that's something, as Balu said, you know, we actually, you know, when COVID happened, people were cutting back. We actually gave normal raises, increased the 401(k), and that enables us not only bringing the talent, but retaining the talent and having a lower turnover rate. So I think not only with this, but also with our suppliers, you know, where we treat them because we are a fabless company, but we treat them that they also have an economic way, you know, from their business standpoint.
Showing that partnership has actually enabled us, and this long-term thinking and not trying to run the business by the quarter and investing in R&D at that time has always paid off. As you see, what we did, even we did, in 2012, walked away from silicon carbide, but invested and doubled down on GaN, and that's paying us rewards. And you saw what we have done recently with the acquisition, accelerating what we are working on, what I call as the next decade of GaN.
Right. Great. I'll get off cyclicality. I promise I'll get off it, but I do have a few more questions. And especially because sometimes Power Integrations is kind of viewed as a canary in the coal mine, because you sell into power supplies, power supplies then go into systems, right? So you tend to get a pretty early read on the overall demand environment, some of your end markets.
Mm-hmm.
You know, you know, I'll have to say, you called the bottom last year. It didn't quite work out. But, you know, it does feel like now you've finally found that real bottom, right? And you've guided for very strong growth, you know, here in the June quarter. So either one of you, could you talk a little bit on what's driving the recovery? And maybe emphasize a little bit on that sort of revenue run rate, right? Because clearly, you are shipping still way below consumption and, you know, how fast do you think you know, it will take for you to get back to that sort of true consumption level?
Well, the first indication is the inventory level. It's overall, it's come back close to normal, but if you look at communications, which is one of our largest segments now, it's way below normal, and that means they've just gone too far. Now, you know, this happens all the time when you're coming off a downturn, especially for us, because we have so much inventory. They know if they need parts, they can order any time and get them, even though we say, you know, six to eight weeks lead time. But they can see that if they need the parts, we are able to ship.
Mm-hmm.
So they get trained very quickly. They say, "Okay, well..." You know, they've had, they have burnt themselves by building too much inventory, so they're very cautious about buying products. That's kind of the bad news. The good news is the bookings really indicate what their demand is. So you don't have all this buffering that's going on with the inventory before. So that gives us good feeling that, you know, communications is out of this thing. And then when you look at... Sorry, consumer, I should say. Communications is a different story. It's been a declining market for us. It will continue to decline but for various reasons, including that the demand for phones is not increasing. It's a flat, you know, numbers market.
But on top of that, people are going out-of-the-box, which reduces the number of charges. So as we said in the analyst day, this is one market that we expect to decline over time, but all other markets should grow very nicely. And you can see that happening, you know, in computers. We have grown very nicely, thanks to shared gains in notebooks and in monitors, and we are also growing very well in industrial. Our high power has been growing for the last two years, even though our overall industrial was down, because the broad-based industrial is still digesting some inventory. But we think that by the end of this quarter, that inventory should also come back to normal, and so we expect growth in the second, I mean, second half of this year.
So that's another area we can see that. So we see growth in computers, consumer, and industrial. So that should bode well for us. We expect this year to be a growth year. Don't know exactly how much, and next year should be a very good year as well.
And, you know, Sandeep, what's your view on the sort of recovery? I mean, you and I talk about this a lot, and it's funny, sometimes I go with my gut, sometimes you go with your gut, sometimes that is actually a better signal than bookings. So what, what's your gut telling us about the second half, Sandeep?
So I think the trajectory that we talk about is that it's upward. The part I think that I'm looking for is the sell-through and the pickup in demand, especially in an area like China with housing and other issue. It is pretty hot there, hot out there in Asia, so I think the air conditioning business is going to do, continue to do very well, and the rest of the region is doing well. I believe with the inventory levels coming back to normal in industrial and the strength we are seeing in metering and high power-
Mm
... will start boarding very well in the second half for the industrial segment.
Mm-hmm.
So as Balu indicated, I think the only segment that is not going to grow, and in fact is going to decline this year, is communication. And part of the reason is, which you when you talked about the COVID, the good, good and the bad, well, we were had so much inventory that we were able to supply, and what other competitors have done had moved their demand to higher margin, and we were able. So we got a disproportionate share-
Mm
... in the communication, which now with this Made in China , plus the going to the lower power levels, is moving away from that, along with the out-of-the-box thing that Balu indicated. So I think this year there'll be a meaningful decline in the communication segment, but the other segment will grow. But I think with this turn, the next year, I truly believe, should be a good driver-
... driver of growth, but the question, as you and I have talked about, is the levels.
Right. Right. No, that's fair. And I'm a little bit surprised that you still view communications as a declining market. I mean, I get it. I mean, you know, it's a mature market. You know, there's what? 2 billion phones sold every year, right? 7 billion people. So yeah, I get it. But there also seems to be some possible product cycles coming, right? You know, we talk a lot about AI, you know, starting to hit, the phone, the notebook. And I would assume that that would require much better power management, much better fast charging, and so on and so forth. So do you not see that as an opportunity, or are there other reasons why you see communications still as a declining market?
Well, the AI really doesn't have a direct impact on what we do because the charger power supply is related to the battery size.
Right.
If the battery size gets bigger, yeah, you need more power, and to reduce the size of the charger, you know, it certainly would help us in that regard. As far as I know, they're not planning to increase the size of the battery very significantly, so it won't have any impact. The out of the box is a pretty significant impact because, you know, if people have chargers at home, they will not buy a charger with the phone.
Mm.
The attachment rate drops from 100% to something like 30% or so.
Mm.
30%-35%. So that has a huge impact. Even though our ASP goes up significantly when it becomes an accessory versus being a freebie-
Right
... it doesn't quite compensate for the loss in volume.
Got it.
So that's why we think it's gonna be a declining market. But the rest of the other three areas will more than compensate for it, and we should have very nice growth.
That, that's fair. And maybe related to that topic, but maybe you could talk about, you know, the GaN technology, the R&D, 'cause all your new products now coming out are based on GaN. I think investors still have a perception that, you know, Power has high share in the smartphone market, but to your point, it's declining, right? So, how would the revenue mix look like three to five years from now, especially with GaN products ramping, especially with automotive becoming part of the story? My understanding is that you're now working with data centers for-
Yes
... you know, high-voltage technology. So how will the revenue mix look like? And I assume that's also driven by R&D. I mean, it's gonna look quite different than what it looks like today.
Absolutely. So there are several things that are going to be happening. One is our computer business is gonna grow very nicely, not only because of notebooks, where GaN plays a huge role, but also in the monitors, where InnoMux-2 , we introduced recently, is getting a lot of interest. In fact, we are going into a high-volume production with one of the major, you know, computer companies, in Q3, for monitors. Notebooks, you know, we still have relatively low share, and GaN is becoming pretty much a requirement for notebooks because of the power levels.
Mm.
You know, we used to have 65 W chargers that looked like these bricks. Now they wanna go to 140- and 200 W chargers that are much smaller, and the only way to get there is GaN, and we have the best solution for USB PD type of notebooks. So we are working with all the major, you know, laptop companies, in, in that regard. So that, that should do very well. It should be, you know, mostly, I mean, InnoMux-2 is mostly GaN.
Mm-hmm.
Except at very low power levels, it's all GaN. That's another benefit of InnoMux-2 over the previous generation. So GaN is gonna dominate the computers. Now, if you go to consumer, we are seeing a lot of new designs using GaN. In fact, you know, we expect to win a large refrigerator design with GaN. But there are hundreds of designs. If you look at overall, the company, cell phone has the least number of designs-
Hmm
... compared to other areas. But-
It's the opposite of revenue, right? I mean-
Opposite of revenue.
Right.
But that's because, you know, GaN is really not tied to any specific market. It's really the... It will replace, I believe, silicon, silicon high-voltage switches, over time. In fact, right now, anything above 30 W, we use GaN, and below 30 W, it's kind of strange that we can't make the GaN small enough-
Mm
... to use. You can't even handle it at that point. You can't put a pad on it to bond. So that's kind of the limitation on the low end. But everything else, we are working on the new products, you know, use with GaN. So GaN is gonna just replace it. So GaN is more of a technology than a market, and therefore, it should go across all markets. So we expect to see a huge number of, you know, designs going into production in industrial. You know, metering is an example, IoT. You know, IoT is probably the largest use of GaN in industrial right now. This is for USB outlets on the wall.
There, the space is very constrained, so you need a really, a very efficient, very small, 100 W kind of a, you know, USB, power supply there in that respect. So, I think, you know, GaN is gonna be a significant part of our revenue going forward.
Mm-hmm.
The next level in GaN is to push the power level. With our current, the latest GaN technology, we can go to I would say several kilowatts, up to about 10-15 kW. That will cover data centers, that'll cover onboard chargers, that'll cover DC to DC converters in the car. And so that will dramatically expand our SAM. And we have the technology already, but we don't have the products. We're working on the products right now. Once they come out, we'll be able to address those markets. The next level for GaN is much higher power levels, 50-500 kW. There it is dominated by IGBTs right now, and silicon carbide in EV-type applications.
Mm.
We think GaN will be a much more cost-effective technology, much higher performance technology. But the current type of GaN technologies can't get there. You've got to have a different kind of GaN device that can handle much higher current levels. So there are two things you need to meet: one is voltage, and the other one is current. Voltage, we are the only ones in the world who can do very high voltage GaNs. We just introduced 12-volt, 15-volt GaN, which is now gonna directly compete with silicon carbide up to maybe 10 or 20 kW. But we need to get the power level high enough to directly compete with silicon carbide at the, in the traction-type applications.
Mm.
That's our goal, and that's one of the reasons why we did the acquisition recently, to be able to develop the technology faster.
Mm-hmm.
So we'll stop there because,
No, no, that's-
I'm out of time.
Yeah, because I'm just wondering, you know, how this changes the mix. And, you know, I guess maybe, you know, Joe eventually is gonna have to do some work on the segment analysis, right? Because at one point, will you start reporting on, let's say, automotive revenue? Would it have to be 10% of revenue, or is there another metric?
Yeah, you know, it gets meaningful. So to, to answer your question on mix, I think at the Analyst Day, we talked about... And if you see the cell phone business last year, did about 21% of revenue. This year, with the decline, it'll be somewhere around, give and take, around 15% of our total revenue. But if you look out a few years, I think it'll be like 13%-14%.
Mm.
While as the computer segment, which four to five years ago was 5%, has grown nicely to ten percent, but I think will become, like, 15%. So if you look ahead, the industrial and the consumer segment will be about 70-odd. And now the industrial segment obviously includes the automotive. So it'll be like a 40, 30, you know, 40% industrial, 30. But automotive will become a sizable business. So if you say if automotive became a $100 million business in 2028, 2029, that's meaningful-
Mm
... at that point of time in terms of revenue. And how we decide to break it out, we can make a decision. But I'm just trying to give you an idea of directionally what will happen-
Mm
... in terms of the mix if you're looking four or five years out.
Yeah, we will separate automotive as soon as it becomes reasonable revenue.
Got it. And maybe data center, too. Maybe. I mean, that comes probably later, but-
Might, yeah.
Yeah. Well, again, it has to become significant within our computer segment. Yeah.
Right. And just back to GaN and sort of the bigger picture, Balu. So you've been very outspoken about, you know, GaN versus silicon carbide. I think investors are very excited about silicon carbide. You know, maybe starting to get a bit more excited about about GaN. But what are you seeing some of your bigger competitors doing on the GaN side? So I'm now thinking about your your STs, your ON Semis, your Infineons. I mean, there's been some acquisitions in the space, right? So I assume that means that they're gearing up for a much bigger GaN cycle. But yeah, could you share with us some views there?
Well, when you look at, you know, whether it's GaN Systems or Navitas, they don't have their own GaN technology. They use TSMC's GaN technology. The Transphorm is different. They have their own GaN technology, which is now acquired by Renesas. A lot of people have tried GaN for a long period of time, and Infineon especially has tried GaN for many, many years. They had their own GaN, and then they acquired IR for the GaN, and neither of those worked out. Then they took a license from Panasonic, which is a different kind of GaN that didn't work. Now they have acquired GaN Systems.
Mm.
It is a very tricky technology if you don't address it properly. And we have a very unique GaN technology, which is not only very reliable, but it's very cost-effective, and that's really important. Cost-effectiveness is very important because if you can't get to silicon cost levels, it'll be hard to replace it in the long run. Yes, it has higher performance, but the higher performance itself is not necessarily necessary for all applications. I'll give you one application, like motor drives. Going fast in a motor drive is actually a negative. You have to slow down the GaN, in which case you have no speed advantage, if you will. But if you have the fundamental cost advantage, you can go across all markets. And that's exactly our focus.
Now, I know that there are a couple of start-up companies who have been working on vertical GaN to go to higher power levels, and one of them we just acquired. But the actual GaN technology is that we are pursuing is the one we have been pursuing for some time, which is very cost-effective once we solve all the problems. The ones they were working on are actually quite expensive to do. That's the difference. But there are portions of it that are very important to us in terms of knowledge and so on. So we're not only acquiring people with knowledge, we are also getting a small fab, which we call a flab, 'cause it's more like a lab than a fab.
But the advantage of that is that we can do experiments much faster doing it inside than having to go to a foundry outside and, you know, for every little experiment. And that's the reason we believe this acquisition is gonna be very beneficial in speeding up our development. It speeds up not only the cycle time of experiments, but we also have acquired all of the key employees who have a lot of knowledge about vertical GaN, and that will be very helpful. There are different aspects of the GaN. It's epitaxy, it's a device structure, and so on. In the device structure side, they have a lot of expertise that we could use. So we believe, and we don't just believe it, we know that this will be, you know, GaN will be more cost-effective.
15 years ago, we invested $65 million in developing silicon carbide technology, and when I saw the electricity bills, I said, "Well, this will never be cost-effective." Now, forget the material part. Material is expensive, but let's say material will eventually come down in price. You can't get around the, energy costs. It is so much... It's, you know, silicon carbide is the second most hardest material after, diamond. To process it, at those temperatures, 200, 2,500 degrees centigrade, takes so much energy, you can very clearly prove that it'll never be cost-effective compared to silicon. And that's the reason we stopped our investment, wrote off $65 million, and doubled down on GaN in, I think, 2011 or 2012.
So, to me, I just don't understand how it could be cost-effective long term. But GaN has some challenges to get to those power levels. But we have gotten this far, and we, you know, we think that we are very confident we'll get there in the next few years.
So the silicon carbide reactors, sort of, you know, they consume so much power that it takes away from the power savings you're getting from an EV in the first place.
That's, that's the other point, is that, you know, people forget that, you know, when solar panels came out, everybody was excited how good it is for the environment. But it initially, it took 15 years before it paid back the energy it used to manufacture the panel.
Okay.
Now, since then, they have dramatically improved. It takes about three to four years now to get back that energy. Similarly, in silicon carbide, if you're gonna use so much energy, it will take years before you get that energy back in savings-
Mm
... to really justify the technology. Now, nobody seems to have the exact numbers of that, and I've been trying to figure this out. All I know is that it takes 20 times as much energy to manufacture silicon carbide versus either silicon or GaN. Silicon and GaN take about the same amount of energy 'cause same temperatures they operate. So, so people have to think about the entire cycle, not just how much they save. This is just like saying EVs will save a lot of, you know, carbon emissions, but then it takes a lot of energy to build the EV.
Right.
So that's why energy efficiency is the number one as factor in reducing carbon dioxide, because every watt you save is a watt you don't have to produce in the first place.
That's great. Great perspective. We're running out of time. Any questions from the audience before we wrap up? Jeremy?
Yeah, I had a question. The White House recent initiative to upgrade the system. Do you guys see some place there? Maybe not requiring, but... Sorry. I thought how-
Absolutely. So most of the energy in the U.S., and in fact, all countries, is transmitted through what's called three-phase AC transmission system. You've seen these big towers with three conductors on it. If you can transmit DC instead of AC, you will save 40% in losses. Or for a given transmission system, you can transmit 40% more. So it's a huge benefit. The reason people are using AC today was because of Westinghouse, who came up with this idea for transforming voltages from low voltages to very high voltages, so that you can transmit a lot of power over long distances. In those days, obviously, we didn't have IGBTs, we didn't have, the ability to transform AC to DC and DC to AC. Now we do.
So you can actually convert AC to high voltage DC, and typically, it's a ± 800,000 volts to ±1 million volts, is typically what you transmit DC with. So you can also go to higher voltages with, with the DC. So, if you do that, you can get a lot more power out of the same amount of copper transmission system, and you can go longer distances. So the pioneer in this is, China. They are going forward with a, a complete DC grid system, which is much more complex than doing a point-to-point link. In the U.S., there are a few point-to-point links that are DC, but they use very old thyristor type of technology, which is not very efficient, but that was installed some time ago....
But all the new systems, I hope will go to, you know, IGBT-based power conversion systems. So there are two things. Point-to-point, the technology already exists, but for the grid system is more complicated. That's where we are working with the State Grid in China to come up with a complete DC grid. And the reason it gets complicated is that then you have to worry about stabilizing the grid, so you need what's called STATCOMs, which also take a lot of our chips, by the way. And then energy storage is required. So we play in transmission, we play in energy storage, and also we play in the stability of the grid.
And I think we are very close to having a full grid, and the goal is to convert the entire country into grid and be an example for the rest of the world. But initially, what we are seeing is the demand is coming from point-to-point links, specifically from wind farms that are offshore. Once you put the wind farms offshore, we've got to have some way of getting the power to the land, and the way it's done is through big cables, and these cables are incredibly capacitive. You just can't use AC. If you use AC, you'll have a lot of losses. So they have to use DC, and currently, we have a number of these installations worldwide that are you know, transmitting huge amounts of you know, power through these cables.
Like we have one in the North Sea. We are just working on one for Saudi Arabia. So the offshore grid, you know, wind farms are becoming very, very popular. It's the lowest cost of renewable energy you can get. Although both wind and solar is now cheaper than coal in most countries.
That's, that's great perspective, and that's why this is a cross-sector conference, right? Here's a chip company talking quite extensively about the grid. With that, we've run out of time. Balu, Sandeep, and Joe, thank you so much for coming to our conference.
Thank you.
To the rest of you, thank you for attending the POWI session, and enjoy the rest of your day.