Power Integrations Inc

NASDAQ:POWI

Good afternoon, ladies and gentlemen, and welcome to the Power Integrations Q4 Earnings Call. [Operator Instructions] This call is being recorded on Thursday, February 6, 2025. I would now like to turn the conference over to Joe Shiffler, Director of Investor Relations. Please go ahead.

Thank you. Good afternoon, everyone. Thanks for joining us. With me on the call today are Balu Balakrishnan, Chairman and CEO of Power Integrations; and Sandeep Nayyar, our Chief Financial Officer. During this call, we will refer to financial measures not calculated according to GAAP. Non-GAAP measures exclude stock-based compensation expenses, amortization of acquisition-related intangible assets and the tax effects of these items. A reconciliation of non-GAAP measures to our GAAP results is included in today's press release.

Our discussion today, including the Q&A session, will include forward-looking statements denoted by words like will, would, believe, should, expect, outlook, forecast, estimate, anticipate and similar expressions that look toward future events or performance. Such statements are subject to risks and uncertainties that may cause actual results to differ materially from those projected or implied. Such risks are discussed in today's press release and in our most recent Form 10-K filed with the SEC on February 12, 2024. This call is the property of Power Integrations, and any recording or rebroadcast is expressly prohibited without the written consent of Power Integrations. Now I'll turn it over to Balu.

Thank you, Joe, and good afternoon. We will discuss our fourth quarter results in a moment, but I'll begin with 2 other pieces of news that we are announcing today. The first is that I have informed our Board of Directors that I intend to retire from my role as CEO. The Board has engaged an executive search firm to help identify our next CEO and will consider both internal and external candidates. I will remain as CEO until the search is concluded and a successor is in place. I have been CEO for 23 years, and I will turn 71 later this year.

And while I have not lost an ounce of my passion for this company or my excitement for the opportunities ahead of us, I believe now is the time for me to step back from the day-to-day responsibilities of the CEO and take on a reduced role supporting a new leader. That role will include serving as Executive Chairman of the Board for as long as needed to help my successor settle into the job. Once the transition period is passed, I expect to remain on the Board with the consent of the Board and our stockholders, of course.

Second piece of news we are announcing today is that Gregg Lowe will join our Board on February 15. Until last year, Gregg was CEO of Wolfspeed and previously served as CEO of Freescale Semiconductor through the time of its merger with NXP in 2015. He also spent 27 years at TI, culminating in the role of Senior VP running that company's analog business. Gregg's experience in analog and power semiconductors makes him an ideal fit, especially his extensive knowledge of the sales and distribution landscape and deep customer relationships in key end markets, including automotive and industrial. His long history in the industry also means he can make a significant contribution to our CEO search. We are delighted to welcome him to the Board.

Now turning to the results. Revenues were in line with our guidance, up 18% year-over-year to $105 million. Revenues for the full year were $419 million, while that was down 6% from the prior year, the underlying details demonstrate why we are excited about the year ahead. The decline was driven primarily by the communications category, I should say, entirely by the communications category, which fell more than 60% following our exit of the China OEM cell phone business at the start of the last year.

The rest of the business grew 17% with consumer up more than 35%, computer more than 10% and industrial up about 3%.

Looking ahead to 2025, the cell phone headwind is behind us. And in fact, we expect our communications category to grow driven by the 5G fixed wireless rollout in India and increasing dollar content in our remaining cell phone business. We also began the year with channel inventories down more than 2 weeks from the prior year-end. Most importantly, we expect incremental revenue this year from an array of markets and products, and I will touch on several of those in a moment.

Our Q1 revenue guidance is for flat sequential revenues at the midpoint of the range, which equates to a year-over-year increase of 15%. While forecasting beyond the current quarter is difficult in light of uncertainty around the trade policy and end market demand, we expect to sustain a healthy rate of revenue growth over the course of the year. Growth should accelerate this year in the industrial category, driven partly by lower channel inventories compared to a year ago, but also by design ramps in high-voltage DC transmission, renewables and traction in our high-power business as well as metering, home and building automation and automotive.

In the consumer category, the rate of growth will moderate after last year's strong recovery, especially with soft housing markets still holding back demand for major appliances. However, we expect growth in air conditioning this year based on share gains and a solid demand outlook from our customers. We also expect new revenues from the TV market after recent GaN design wins, which I will discuss in a moment.

Potential growth drivers in the computer category this year include notebooks and tablets, auxiliary power supplies for AI servers and also monitors where our InnoMux-2 ICs are in production with a major PC OEM. Underpinning our growth across 4 end market categories are 2 common themes. One is our success in India, where we have expanded our presence in recent years. A priority of India's government is to design and build domestically more of the products purchased by its growing middle class. The country is also modernizing its infrastructure with electric transportation, residential broadband, renewable energy and a more robust power grid, including the planned installation of 250 million smart utility meters.

We are winning in each of these areas, supplying gate drivers to one of India's largest suppliers of traction systems for electric locomotives and winning a substantial share of the metering and fixed wireless rollouts. The second key theme for this year is GaN. Last year, we talked a lot about progress in our technology road map, including the launch of 1,700-volt technology. We believe 2025 will bring an inflection point in terms of adoption and growth. We expect revenues from GaN-based products to grow at a high rate this year and to comfortably exceed 10% of our sales.

In Q4, we won a follow-on design at our Indian 5G fixed wireless customer, which is upgrading to GaN after ramping last year with the silicon-based InnoSwitch. Metering customers in India are also now moving up to 900-volt and 1,250-volt GaN products to gain extra safety margin against India's fluctuating grid voltages.

We also recently received our first purchase orders for GaN-based InnoMux-2 ICs at one of the world's largest TV manufacturers. We have won power supply sockets in 3 models, largest being 65-inch screen, which will not only use InnoMux-2, but also our GaN-based HiperPFS power factor correction chip.

Along with accelerating customer adoption, our leadership in GaN technology and products is also being recognized by industry experts. Our 1,700-volt InnoMux-2 ICs received a 2024 Product of the Year award from a leading U.K. technical journal and a PowerBest Award from Electronic Design magazine. InnoSwitch3 with 1,250-volt GaN won an engineering achievement award from Design World, Best Power Management Product from AspenCore in China and 2 industry excellence awards from [ 21 Dayayun ] also in China. While 2025 is shaping up as an exciting year for GaN, we are still in the early -- we are still very early in the GaN revolution with huge opportunities still ahead in the short, medium and long term.

Short term, GaN has just begun to penetrate the power supply market and adoption is accelerating across a wide range of low-power AC to DC applications, including the ones we talked about today and many more.

In the medium term, the opportunity for GaN at high power levels is massive, nowhere more so than in AI data centers. While data center operators are eager for innovative power solutions for AI, adoption of GaN has been inhibited by the challenges of using discrete GaN in high reliability systems. We are tackling that problem with our system-level approach to product design and expect to have our first product for AI server power supplies next year. We estimate the SAM for this product alone to be more than $0.5 billion in 2027, with additional products to follow that will take our data center SAM to well over $1 billion.

Longer term, we believe GaN can achieve power levels sufficient for EV drivetrains at much lower cost than silicon carbide. We are pleased with the progress we are making on high-power GaN, aided by our acquisition of Odyssey Semiconductor last summer. And we continue to believe that a market-ready high-power GaN technology is attainable within the next 3 to 5 years. I'll conclude with an update on our automotive efforts, which are progressing nicely. EV power architectures are not only evolving in ways that benefit power integrations, but in some cases, are being shaped by our expertise in high-voltage systems and the unique capabilities of our products.

Automotive revenues will grow rapidly in 2025 from a modest base of few million dollars in 2024. More importantly, we are building an impressive roster of customers in the EV industry, including pure battery EVs and plug-in hybrids, which should result in a more substantial revenue contribution starting in 2026.

Building on our early success in China, we are now expanding quickly into other markets. We have several customers scheduled to begin production this year in Europe and the U.S. In Japan, we were initially expected -- we initially expected resistance as a non-Japanese supplier, we were instead being invited into the market because of the capabilities of our products. Following our recent qualification at one of Japan's largest Tier 1 suppliers, we have now been invited to begin qualification at Japan's largest Tier 1, and we hope to complete that process by the end of 2025. With that, I'll turn it over to Sandeep for a review of the financials.

Thanks, Balu, and good afternoon. Our Q4 results are straightforward, so I will just briefly recap the numbers and the outlook, and then we will take questions. As usual, I will focus on non-GAAP results, which are reconciled to GAAP in our press release. Fourth quarter revenues were $105 million in the middle of our guidance range. Revenues were up 18% year-over-year and down 9% sequentially. I will briefly speak to the sequential changes in each category.

Consumer revenues were down mid-teens sequentially, reflecting continued softness in major appliances in the U.S., Europe and China. Finished goods inventory at Chinese OEMs remain elevated as the impact of consumer stimulus program thus far appears to have been fairly modest. However, we did see a meaningful drawdown in channel inventory for consumer in Q4, giving us added confidence in our growth expectations for 2025. Industrial revenues were down 10% sequentially, largely driven by the timing of shipments for metering applications. Revenues from the computer category were down mid-single digits sequentially on lower sales in notebooks, while the communication category was up mid-single digits on stronger cell phone revenue.

Total channel inventory fell slightly to 8.4 weeks compared to 8.6 weeks last quarter as distribution sell-through exceeded sell-in by about $2 million. For the full year, sell-through exceeded sell-in by about $12 million. Revenue mix for the quarter was 37% consumer, 35% industrial, 15% computer and 13% communications. Non-GAAP gross margin for the fourth quarter was 55.1%, unchanged from the prior quarter. Full year non-GAAP gross margin was 54.4%, up more than 2 percentage points from the prior year, driven by the weaker yen and favorable mix with lower c cellphone revenues and higher percentage of sales from industrial and consumer.

Non-GAAP operating expenses were $44.6 million for the quarter, consistent with our guidance. For the full year, non-GAAP OpEx was $174 million, up 4% from the prior year, including about 0.5 percentage point stemming from the Odyssey acquisition. The non-GAAP tax rate for the quarter was negative 3%, reflecting a reversal of FIN 48 reserves. The reversal resulted in an EPS benefit of $0.02 per share.

Including the tax benefit, non-GAAP earnings for Q4 were $0.30 per diluted share. Share count for the quarter was 57.1 million, up slightly from 57 million in the prior quarter. Inventory days rose to 315 at quarter end, up 24 days from the prior quarter, reflecting the sequentially lower revenues. Inventory on the balance sheet fell by $2 million during the quarter. Inventory will remain well above our target level throughout the year, but should begin to taper down in the second half.

Cash flow from operations was $15 million for the quarter, while CapEx was $3 million. We used $12 million for dividends after the 5% increase that took effect in Q4, and we used $2 million to buy back shares. For the year, we generated $81 million in cash from operations with just over $17 million in CapEx, resulting in free cash flow of $64 million. We returned $74 million to stockholders through dividends and buyback.

Turning to the Q1 outlook. We expect revenues to be flat sequentially, plus or minus 5%. Non-GAAP gross margin should be between 55.5% and 56% compared to 55.1% in Q4. The sequential increase reflects favorable end market mix and incremental benefit from the dollar-yen exchange rate. For the year, I expect gross margin to be around 55.5%. That would be up about 1 point versus 2024 with higher back-end production volumes and favorable mix offsetting higher input costs.

Non-GAAP operating expenses for Q1 should be around $45 million, a slight increase from Q4, driven by the resumption of FICA taxes and modestly higher headcount. For the year, I expect non-GAAP OpEx to increase by about 6%. About 1 percentage point of the increase is a result of the full year of Odyssey expenses compared to half a year in 2024. Finally, I expect our effective tax rate for the first quarter and the year to be in the range of 5% to 6%. And now operator, let's begin the Q&A.

[Operator Instructions] Your first question is from David Williams from The Benchmark Company.

And Balu, let me just first say congratulations on the retirement. It's well deserved and we'll miss you, but wish you the best.

Thanks, David.

So first, Balu, it really seems -- I mean, you've obviously been very upbeat on GaN opportunity over the last several years, but it sounds like you've got greater conviction today in meeting that higher power and even talking about the EV drivetrain in those areas. Has anything changed maybe over the last couple of quarters that's given you more confidence there, maybe something that's come along with Odyssey or just maybe internal efforts. But just anything around that -- your confidence in that -- in the GaN transition to the powertrain potential?

Thanks, David. Absolutely. We talked about it in the last quarter as well. I would say in the last year, we have made significant strides in our technology that will allow us now to address tens of kilowatts. And that's why we are now talking about AI data center opportunity. We think with the technology we have today, we can go up to perhaps 50 kilowatts worth of applications.

Now the new technology we are working on, which is the reason we acquired Odyssey Semiconductor will allow us to go to much higher power levels, hundreds of kilowatts. That's when we'll be able to get into EV drivetrain. And we believe we can be very competitive or actually much lower cost than silicon carbide and provide even higher performance than silicon carbide.

So that's kind of the holy grail that we are working towards. And there, we are making good progress. We still have some ways to go. We think we can be there in the 3- to 5-year time frame. And acquiring Odyssey is really going to help us get there faster than we originally thought. And that's the reason I'm very optimistic about GaN in general. But I would also say from a business side, I'm really impressed how broadly GaN has been adopted. It's no longer just cell phones. We are talking about notebooks, tablets, monitors, TVs, appliances, industrial applications. So it's really across all of the markets. And that's why we believe this year is going to be a significant growth for us. And going forward, it will grow even faster. So I am very, very confident GaN will be the high-voltage switch of choice for the future. And of course, all of our products have them. So that's going to make a huge difference to us.

And also, we are so far ahead of our competitors in terms of voltage level, in terms of reliability, in terms of cost. We feel very, very strong that GaN would be a huge benefit for us, differential benefit for us as a company.

Fantastic progress there and making a lot of headway. But I guess maybe secondly, on the automotive, you also talked about that now and where you're playing. Can you update us on maybe how many design wins you have in the automotive space? And maybe just kind of speak to the magnitude that you're expecting from GaN and automotive over this year and maybe in the next couple of years, how that should trend?

Yes. Just to be clear, a lot of the designs that we are already in, we have about 20 designs that are in production right now. They are either silicon or silicon carbide. We recently introduced the 900-volt GaN that will be suitable for the 400-volt battery systems, and we are actually getting design wins on that. And of course, we have the 1,700-volt silicon carbide, but we will also have the 1,700-volt GaN-based device for automotive as well. And that is required for the 800-volt battery.

So the 400 will require 900-volt GaN and the 800-volt battery will require 1,700-volt GaN. So we are in very good shape to address those markets with GaN. But on top of that, there are other sockets in the car, like, for example, DC-to-DC converters to replace 12-volt batteries. So we are working on products using GaN. It is a higher power version of InnoSwitch and that will provide us additional sockets and significant dollars for that matter.

Right now, our primary socket is the emergency power supply in the drivetrain and also auxiliary power supplies for various subsystems. But the DC-to-DC converter is a much bigger dollar content. And then once we have the 10, 20-kilowatt type of solutions that we are working on, we can also address the onboard charger application. And then we have to wait for our high-power GaN to go into the EV drivetrain.

Your next question is from Ross Seymore from Deutsche Bank.

A couple of questions. And first and foremost, Balu, congratulations. As David said, very, very well deserved and we'll be missing you on these calls. So I guess for my first question, the reporting guide were pretty much in line. It didn't even look like the end markets were terribly surprising. So what would you describe as the biggest update of what you're seeing overall from an end-market environment today versus 30 days ago? What's changed?

Well, that's a good question. Not a significant change. Only thing I would say is in 2025, that is this year, we expect to see a significant growth in industrial, which is kind of a little bit out of sync with the rest of the semiconductor market. And that's because we have some really unique opportunities that will go into production, things like infrastructure-related projects, renewables, high-voltage DC transmission systems, electric locomotives, and that's in the high-power part of the business.

But beyond that, we are also doing very well in meters, which is also infrastructure related. We talked about the India meter opportunity. And then we are doing very well in home and building automation and power tools and so on. So this year, industrial is going to be the strongest growth market, which again is a little bit unique, but part of it is because some of the infrastructure projects, which is supposed to start last year have been pushed to this year, but that bodes really well for the industrial this year.

And I guess as my second question and not to ask too much, but I think in your last call, you talked about you thought all 4 segments would grow this year. Is that still the case?

From the best modeling we have done, all 4 should grow, but industrial would be the largest growth.

Yes. In dollar terms, industrial will grow followed by consumer, but we expect all 4. And even in communications, we are expecting growth with the 2 areas of cell phone as well as in networking product. So we have got growth in all areas.

Yes. Just to amplify on that, we talked about the networking product. This is the residential networking, the fixed 5G network. But we are also going to see growth in the remaining cell phone market because of content growth. We have growth because we are able to upgrade to InnoSwitch-PD, which has the PD protocol built into it. So that increases our dollar content. Also, many of the products are good -- I mean, there is also a move towards higher power to charge AI-based phones. And that will also increase our dollar content because they'll then move on to GaN for higher power chargers.

Your next question is from Tore Svanberg from Stifel.

Balu, congratulations on your retirement. It's been a true honor working with you and covering Power Integrations during your tenure. I wish you all the best. So the first question that I had is on the near term by segment. I assume the Communications segment will probably be down seasonally. But how should we think about the 4 segments into the March quarter?

Yes. The way I would look at it that you would see that basically the consumer and industrial will grow and come and communication and computer would be down.

Great. And as my follow-up question, channel inventory, Sandeep, I know sometimes in the March quarter, there's a lot of moving parts, especially with Chinese New Year and so on and so forth. So I know channel inventory came down a couple of days, I believe you said in Q4, but how should we think about channel inventory target in Q1?

I think as the best I know, I think sell-in and sell-through to should be around similar. So I think the channel inventory should hold. And in fact, this is one of the good things for us for 2025. We are coming starting off with what I call the normal weeks. And with all the growth drivers that Balu talked about, I really -- we feel very good about our growth and really think even for the year, sell-in and sell-through should be pretty close, give and take.

Very good. And just one last question for you, Balu. You talked about some early revenues in data center power supplies scheduled for early next year. Are those GaN and silicon products or either one or the other?

Tore, first of all, thank you for the congratulations. Yes, just to be clear, I said that our first product for data centers will be available next year, but we won't see revenue until '27 or '28. We may get a little bit in '27, but really, it will be '28 before we get significant revenue. And the products are entirely GaN because GaN is where we have really an advantage in the AI data center power supplies.

But I do want to remind you, we are already designed into AI server power supplies with our standby products, which are also GaN-based. And we -- that will grow very nicely this year and next year but that market is not as big as the main power supply. For example, the standby market is about roughly $100 million of SAM, whereas the main power supply is we are talking about more than $1 billion once we introduce the 2 or 3 products, we are planning fo it.

Your next question is from Christopher Rolland from Susquehanna.

Balu, I echo my congrats. You look great for 71, by the way.

That picture is a little lured.

I need your skin care routine. Okay. So my question is around GaN, and it seems like you're really focused on high power. Do you think like the low end of the market commoditizes between like in a science on one side and 300-millimeter from Infineon on the other at the low end? And then on high power, is this like a sustainable competitive advantage? Or do you think others are going to catch up here as well?

I believe -- first of all, thank you for the compliments. But I believe we can compete very well across the entire power spectrum from low power to high power. And let me talk a little bit of the 2 examples you gave.

First of all, we don't sell discrete GaN. People who sell discrete GaN, I think, have a very tough challenge with some of the Chinese suppliers, especially Innoscience. We sell a system and the benefit there is that we really can extract the most value out of the GaN at the system level in terms of efficiency, in terms of protection and so on and so forth. So that's one thing. So we think we can compete very well in the low-power area.

When you go to high power, the dollar content is much higher and the benefits we can bring is also more significant. And we think that, that's a huge growth path for us. If you look at AI data center, we think that will provide us a strong revenue growth because the products will bring such significant advantage in terms of size and efficiency.

Now talking about 300-millimeter, there are 2 things I think is really important to know. First of all, at this point, we don't know of any equipment that can do high-voltage GaN on 300 millimeters. I believe the announcement that were made are for low-voltage GaN. When I say low voltage, less than 600 volts, which is basically [indiscernible] in 100, 200 volts may be possible. But to really go to higher voltages, you need much thicker epi. And the challenge with that is, first of all, the cost of GaN is really in the epi. So the size of the wafer makes not that much difference. And to the extent the wafer size makes some difference, the fact that you grow such big epi on such a large wafer creates a humongous amount of stress. I mean they -- I think they call it as a potato chip. That's what it looks like. And what that means is you have to deal with that stress, when you do that, the yield goes down significantly.

So it is not clear to me that going to a larger sized wafer is a benefit for high-voltage GaN, just to be clear. What we are doing is reducing the cost in other ways like device design, process design and really dramatically reducing the die size. The difference in wafer size is not very significant in terms of cost.

Interesting. The next one is maybe for Sandeep. Sandeep, you always have macro thoughts. I appreciate those. It sounded like in the press release, there was some commentary around tariffs. I wanted your thoughts there. Also, you're usually plugged in on Chinese subsidies. It seems like you had some commentary there as well. And then anything else macro that you want to add, I would appreciate.

Yes. I mean, basically, we felt the subsidies -- the full effect of that hasn't been up. There was also -- there could be some put and pull in because of tariffs. I really want to see the sell-through in first quarter, which will give me a better idea of whether it was tariffs or whether it is the subsidies. So I'm waiting to watch that.

I think the trade policy is obviously something we are watching because if you have a lot of tariffs and has an impact on end market demand, that can also impact us. At the end of the day, we go into things like appliances and there are other and that it becomes very expensive, can impact demand.

But I think if things really don't go extreme and are normal, I think the areas that Balu has highlighted of growth in our appliances as well as in our industrial applications, we really believe that we will have a very good growth here this year.

So obviously, a thing which is hard to predict is the real impact of tariffs, which it's a wait and watch. But I think all these things that come even out over a period of time, but do have short-term impacts. So -- but I think we have enough growth drivers that if things are normal, that we'll have a very good year.

[Operator Instructions] There are no further questions at this time. Please proceed with closing remarks.

Okay. Thanks, Andrew. I know it's a busy day of earnings out there. Thanks, everyone, for listening. There will be a replay of this call available on our website, investors.power.com. Thanks again, and good afternoon.

Ladies and gentlemen, this concludes your conference call for today. We thank you for participating and ask that you please disconnect your lines.