Microchip Technology Incorporated (MCHP) Earnings Call Transcript & Summary

Great. Well, good morning, everyone. My name is Craig Hettenbach on the semiconductors team at Morgan Stanley. Really appreciate you joining today, the fireside chat with Microchip. Just before we kick off, I do need to highlight, disclosures can be found at www.morganstanley.com/researchdisclosures. So with that, I'm very pleased to have with us today, Steve Sanghi, who's now Executive Chairman; as well as Eric Bjornholt, CFO. So welcome. And I think Steve is going to make some introductory comments, and then we'll get into Q&A.

Thank you, Craig. Before we begin today, I wish to remind you that in today's fireside chat, we'll be making some projections and other forward-looking statements regarding the future financial performance of Microchip. These statements involve predictions, and the actual results may differ materially. I refer you to Microchip's filings with the SEC regarding some important risk factors about the company. So let me start with a brief update about the business environment. The business environment continues to be very strong, with bookings and backlog continuing to set new records. In fact, bookings and backlog have exploded, and I have never in my 42-year career seen such strong bookings and backlog. We're also getting 12 months of bookings from a lot of customers. So therefore, the total size of the bookings order just look extremely large because they're not for 1 quarter, they're not for 1 month, they're for 12 months. So as a reminder, the record level of bookings and backlog are not an indicator about the current quarter's business as a lot of the backlog continues to layer in over the next 12 months. This provides us with more visibility than we normally have, which is very helpful in a constrained environment to use our capacity more efficiently. Our demand over the next 3 to 6 months is running considerably higher than our capacity to service it despite of bringing on more capacity. As a result, our lead times have extended on a broad base of products ranging from 4 weeks to over 40 weeks. The exact lead times are product-specific and dependent not only on what the capacity constraints are, but also on what the demand intensity is. On January 1, we changed our noncancelable window from 45 days to 90 days in order to give us a better opportunity to utilize our constraint manufacturing assets as well as to discourage placing orders in excess of what is truly needed. Two months after implementing this change, it has worked out as we expected, and this will continue to be a new standard practice. About 4 weeks ago, based on broad customer feedback, we launched Microchip Preferred Supply Program, or PSP. This program offers our customers the ability to receive prioritized capacity starting in the 7th month when they place 12 months of backlog that is noncancelable and non-reschedulable. The response from our customers to join this program has been swift and very broad-based with over 1,200 customers who have committed to the program. We expect many more customers to take advantage of this program in the coming days as they work through their internal procurement logistics. In fact, every single major global account has either already signed onto this program or said that they will sign on and they're working through their logistics. Now a few words on the supply side. With the strong demand and low levels of inventory, both at Microchip and our distribution channel partners, we are experiencing constraints in all of our internal and external factories and the related manufacturing supply chains. We started ramping our internal factories in September as well as investing in capital additions to expand our internal capacity. We also work closely with our supply chain partners who provide wafer foundry, assembly tests and materials to secure capacity. Through the combination of internal and external capacity actions we have taken, we expect our overall capacity will continue to grow every quarter in calendar 2021. While our capacity will continue to grow every quarter, we also believe that wafer fab as well as assembly and test constraints are here to stay with us through calendar 2021 and possibility into calendar 2022. That concludes my update, and let me pass it back to you Craig for the Q&A session.

Great. Thanks for that update, Steve. Perhaps before we get back into the supply chain, just from an end market perspective, 2020, there were a lot of puts and takes in terms of areas of strength due to the pandemic, areas of weakness, how are you seeing things in 2021 from an end market? And any particular highlights?

So the end markets really haven't changed. Compared to what we have been seeing for the last 90 days or more, the automotive is seeing a very, very strong V-shape recovery and so is industrial. Data center benefited a lot from the work from home and PCs and all that, so they are seeing some digestion, and the business is returning back to normal. The Military, Aerospace and Defense business is very budget-driven. The new budget recently came out, and now people will start placing their orders. But mainly, I think the takeaway is that industrial, automotive and home appliances continue to be strong, and our data center is still digesting.

Got it. And then relating to the supply chain, you and many companies kind of went fab-light over the years. As we look at this current situation, what are the thoughts in terms of what you're doing internally versus externally? Are there things you can tweak in terms of bringing a little bit more in-house? Or how are you thinking about that?

So traditionally, looking in our history, we have always had an inside bias. If we can do something inside, we often do inside, especially on the assembly and test side. 4, 5 years ago, prior to Atmel purchase, Microchip did 95% of its test inside. And that may include maybe one more acquisition before, but classic Microchip business was 95% test inside, over 70% of assembly inside. And then we bought a number of these companies, Atmel, Micrel, SMSC, Microsemi, and a lot of those companies really had a very large portion of their manufacturing outsourced. So when we combine those 2 companies together, our percentage of the work we did inside just dropped dramatically. And then we have been on a never-ending quest to bring more assembly inside, more test inside. It is accretive. We have more control. We have a higher control of quality. We've been doing that. But the large leverage over the last couple of years, it constrained us for a while on really how much we wanted to spend it, bring it -- to bring it inside. But even with that, those numbers have been continuously growing. And Eric can give you an update on where we are today. On the fab side, we have often less ability to bring in because many of the foundries that these acquired companies used, those were proprietary technologies of the foundries that we cannot bring them inside. So we have been able to make less impact on bringing fab inside. With one exception, Atmel had a number of technologies that they owned, which they had outsourced to the foundries, but the technologies were owned by Atmel. So some of those technologies, we're able to bring it inside. We qualified them in our Fab 4, Oregon facility, and we are ramping output on those factories. So there, we have some flexibility. But most of the foundry technologies are proprietary and those we cannot bring them inside.

Yes. So maybe just to add on to what Steve was saying there. On the assembly side, in the December quarter, we did about 55% of our production in-house, and that was actually a pretty significant jump quarter-on-quarter. Tests were at about 57% inside. And assembly, over the course of time, we expect to go over 60% and test over 70%. So still investments to make there, and that's included in the capital plan that we're executing on today.

Got it. Steve, just going back to the Preferred Supplier Program, and nice to see the increased number of -- the breadth of that program from your customers. Can you touch on just perhaps -- one of the things with the pandemic last year as well in terms of all this volatility in orders, the difficulty to kind of plan a business, to run your business, how does that help in terms of visibility operationally as we go forward, that program specifically?

So it helps a great deal operationally because, at any point in time, historically, we enter a quarter with something around maybe half the quarter booked, sometimes less than that. And then the rest of the quarter, you have to take turns. You have to take bookings that will turn into the current quarter. The assembly -- the fab assembly test time is longer than that, often 12 to 16 weeks, to complete a product. So therefore, large amount of product that runs into our factories is not really being built to backlog often, it is being built through a forecast. And then there is always -- forecasts are only that much accurate. So at any point in time, then there's an amount of product you produce in the current quarter where the bookings or turns don't come in, they come in for the next quarter. So there is an efficiency factor where you're not able to ship every unit you make in the current quarter. It might -- some might ship next quarter or the quarter after because you're building it through a forecast. In the current environment, with the strong visibility and backlog we are getting, to give you a feel for it, on many of our product lines, the next quarter backlog is already higher than the current quarter. And there are some product lines -- again, the product lines, which are very much data-center specific or communication specific, few areas which are weak, those, we still have a very, very strong backlog, better than ever before, but they're not higher than the current quarter. But most of the product lines, most of the microcontroller 8, 16, 32, analog, wireless and others, their next quarter backlog is already higher than the current quarter. So with that level of visibility, every wafer you're making in the fab, every assembly unit that you're mounting and every test unit that you're performing, is going out the door for revenue. So that efficiency goes up dramatically.

Got it. And then in distribution, I think your inventory last quarter was a record low. And it doesn't sound like, given these constraints, that it's going to change materially near term. But how do we think about the distribution channel for your products? And are there any differences you're seeing by geography and distribution currently?

Let me have Eric take that.

So you're right, Craig. Last quarter, we ended with distribution inventory at an all-time low of 26 days. And in the current environment, there just is not the possibility of distribution, building any sort of significant inventory. The capacity just isn't there to take it. They'd like to have more, just like our customers would like to have more, but we can't do it. So we've seen, over the course of time, that distribution has had a large range of inventory, and that varies based on what point in the cycle that you're at. But, over time, we would expect distribution to build some inventory, but I just don't see that happening in any material way in the next few quarters.

Understood. Got it. Maybe we can shift gears and just talk about some of the growth drivers, and specifically the total system solution, which is something you've been talking about the last number of years and something that's an effort inside the company. Any update in terms of what you're seeing for that program? And the traction that is having on growth?

So the total system solution is going very well. First of all, as we engage with customers in various different markets, we're able to see that our product portfolio is very strong, where we are able to populate a customer's design with Microchip products across the needs, whether it may be the central processing unit, which could be a microcontroller, it could be an FPGA, it could be an MPU or SoC. And then all the chips that surround it from analog, converters, power management, op amps, memory, flash memory or static RAM, all the connectivity, USB, Ethernet, Wi-Fi, Bluetooth, it's just -- it's never been that our product line was so full that we're able to really put every chip on that board made by Microchip, and only thing remaining are resistors and capacitors and switches that are mechanical and comes from other suppliers. So we're getting a confirmation that the product line is full and is very, very good, and we're winning designs with a large number of chips around the central processing unit. What Street would like is some sort of measurement, which is kind of hard to get, because we work with 120,000-plus customers. Majority of them really work with distribution. We work with a few thousand customers directly. So every quarter, we give you samples of block dive ins that show in various markets, whether it's data center, electrical vehicles, industrial market, consumer market, how well that is working. But overall indicator is very, very hard to put together on really how many chips because we have more visibility at the design stage than we have at the manufacturing stage. It goes to manufacturing somewhere, and they're acquiring parts for multiple projects. And it's difficult to get hands around. We're trying to put some indicators together, but don't have it right now.

Okay. And understanding you don't have specifics on it, I mean, at a high level, I guess, how would you characterize just perhaps the tailwind that it can mean to your growth? And I don't know whether that's looking at it versus the semi industry overall? Or just how this can kind of enhance the growth?

Well, TSS's ability to attach more chips around a single design, together with focus on the 6 megatrends that we have talked about, is the fundamental reason why we believe that we could outgrow the industry by a few percentage points.

Got it. Okay. And maybe since you mentioned just the megatrends, any update there in terms of -- and I know because for some of these trends, some of them are near-intermediate term, some have perhaps a longer tail, if you can just kind of talk about how these trends are flowing into place for Microchip?

So we aren't finding any issues. The applications we have defined in the 5G market, in the IoT market, in the electrical vehicles, in self-driving or driver-assist functions and AI and ML, and all those areas, we are getting a large amount of design wins. There, we have more numerical measurements internally. We'll not share them, and I don't know whether we want to share them. But from a funnel database, we can query how many of those designs are tagged for them to be in electric vehicles, or 5G, or IoT, or whatever. And those funnels and all those 6 mega markets are growing quite strongly.

Got it. I just wanted to take a step back. As the company has evolved for the years, organically and through M&A and all the pieces that you have, I mean, sometimes there's a debate around analog versus microcontroller. Just how you see these pieces fitting together, and just kind of various growth rates by segment? Like what are the pieces where you see kind of the strongest growth within your business?

So my feeling is that the analog and microcontrollers largely go hand-in-hand. Any application that we go serve out there, whether it's in 5G, or electrical vehicles, or IoT, or whatever, there are microcontrollers on the designs and there are analogs on the design. So my feeling is they go hand-in-hand. So broadly saying that the analog and microcontroller growth rates would be similar, with one exception, which is how you categorize the part to be analog and how do you categorize the part to be microcontroller. So if you take an example, let's say, it's a point of load detection and managing chip that is largely an analog chip, but you add a little microcontroller to add intelligence to it. At Microchip, the classification of that analog chip will transfer to being a microcontroller because it has a microcontroller CPU. If it's the exact same chip being done in Maxim, and they add a little 8051 engine to add intelligence to it, they would still call it an analog chip. That's the only difference. So over time, we're winning designs in both. And if you add it up, the growth would be similar. But because of that classification, constantly, some of the analog revenue gets reclassified into microcontroller revenue when we add a microcontroller engine to it.

Got it. And specifically, on microcontrollers, I think your company has stayed committed to 8-bit, which I think has served you well in terms of many other suppliers kind of moved on and didn't focus on it. So can you maybe just talk about kind of the breadth of the business across 8, 16 and 32? And why that helps in terms of that the focus that you've remained?

So from the very beginning, we organized our businesses to be business units. There's an 8-bit business unit. There's a 16-bit business. There are actually multiple 32-bit business units, because there are so many different factors in it. And each business unit within this segment -- because what are the customer needs and how to continue to expand applications and add intelligence, 8-bit business unit constantly figures out how to create new applications going from electromechanical control to microcontroller-based control and they kind of continue to grow. What we have seen is when you combine all those together, then the business unit always sees more opportunities on the high end. And 8-bit becomes uncompetitive. Every new product line pick up the next 32-bit because there's infinite performance you can add. By organizing it by business units, we have continued to grow our 8-bit business unit. We currently expect our 8-bit business unit next quarter will probably hit all new -- all-time new record. And that's not what the analysts write about and opt for an investment -- what the investors think, they think 8-bit is something that's declining, it's a headwind, and it's not growing. All those are wrong, and it's been a wrong bet for nearly 25 years now. Our 8-bit continues to do very well. Like I said, we expect to really make an all-time new record on 8-bit next quarter. So similarly, 16-bit would be a new record, probably this quarter. And 32-bit also, there'll be new records this quarter. So each of them are growing. Each of them keeps finding new applications. If you take an 8-bit application, and you don't do anything to it, and the customer needs something, they need to add a feature, they need to connect it to Internet, they need to add a sensor, then they, obviously, will go to 32-bit or 16-bit to meet that need. But if 8-bit keeps serving that need, the customer doesn't have a desire to go to 32-bit. It's a bigger chip, costs more, takes more area on the board. It's more difficult to write code on it. It requires a larger memory. It requires more power. Has all those things, customer -- 8-bit customer wants an 8-bit as long as it keeps meeting its needs. And the way we are organized, it allows the 8-bit business to seek out what those customers need, and they keep expanding their customer base and they keep growing their business. It did an all-time record in 2018 and the '19, '20 were slow, driven by COVID and U.S.-China trade war and all that. And this year, it would hit a new record, starting next quarter.

That's great context. Maybe we could shift gears towards capital allocation. The company has talked about kind of this pivot. You've been very acquisitive over the years, but now the focus seems to be more on the cash flow generation, returning that. And so if you can just kind of talk about why that makes sense at this point in Microchips currently?

So we often first define what the need of the business is, and then figure out a strategy how to meet that need. About 12 years ago, we started around figuring out what the need of the business was and what strategy would satisfy it. At that time, Microchip was a subscale company compared to many of our competitors in analog and microcontroller space. We were still winning based on strength of our products and design and everything else, but we needed to build a nearly 7x to 10x scale from -- in the 2006, '07 time frame to, going forward, 12, 13 years. And the semiconductor growth during that entire period, about 2002 to 2016, was only in the mid-single-digit, and that kind of growth wasn't going to get us anywhere in terms of building the scale and outperforming our competitors. So that required a strategy of M&A where we drum roll. Every year, 2 years, we bought a large semiconductor company, and we built the scale to what would be over a $6-billion company this year. And the companies we acquired were all the companies which helped us complete the customer solution. And 12, 13 years later, we find that we can implement TSS because every chip can be made at Microchip. And the companies we bought, each of them were accretive, successful, merged into Microchip's business very, very successfully. We brought its margins, no matter what the margins were when we bought it, to Microchip level. So that was a very successful M&A strategy, and we paid a very reasonable price. For SMSC, we only paid 1.7x sales. For Atmel, we paid 3.5x sales. For Microsemi, we paid about 5x sale. Today, those deals are going for an 8x, 10x, and sometimes they're less profitable than the companies we bought. So as you look at the current time, what is the need of the business? M&A is no longer the need of a business because we don't find ourselves with scale disadvantage to our largest competitors, to ADI, to Maxim, to Linear, to other companies, $6 billion versus $7 billion or $8 billion is not that different, $500 million versus $6 billion is. So we don't have the scale disadvantage. We complete customer solution and TSS is successful and the pricing that companies have are very expensive, [ with a property valuation ] and our balance sheet is stretched. So given by -- given all those factors, M&A is no longer needed. So therefore, we are rapidly paying down debt, having paid well over $3 billion of our debt in the last 2.5 years. And as the leverage comes down further and we get investment-grade rating, which I think we're talking about 1 year, although it doesn't depend on us, depends on the rating agencies. When that happens, and we're already building a glide path by starting to increase dividends so that when we achieve that objective investment-grade rating, we can start giving significant amount of cash back to the shareholders. Today, we're doing it with a dividend, and we're doing it with convert buyback, which is actually the most expensive instrument we have. When we buy it back, it buys the underlying shares of dilution for the future. But there are only about $600 million of converts left, the old converts, which are very expensive. The new convert we issued last year is really not very expensive at all, and it's covered by a cap call. So the old converts, only $600 million are left, so they won't cause us much dilution. But at some point in time, over the next few quarters, we started a pure buyback programs, stock buyback from the market. So that's really -- that's the need of the business today, and that's what drives our strategy.

Got it. And you just increased the dividend by 6%. So can you maybe just talk about that in terms of how you're thinking about growing the dividend over time versus buying back stock? And how you make those choices?

So Board has indicated that it intends to grow the dividend every quarter. And this was the first -- every quarter when the dividend is announced, it's a new decision. They haven't laid out, the next 5 quarters, of what the dividend would be. But I think a reasonable expectation would be that the dividend will grow every quarter, and will grow maybe in that range. As far as the stock buyback is concerned, as long as we can continue to buy the converts, there's not too many left. At some point in time, the people who hold them want an exposure to the equity, and they do not want to sell it. So when that happens, then we convert to a stock buyback, which could happen. I can't give the exact time, but it's not too far.

Got it. Eric, maybe we can touch on the target model. End of last year, you increased your profitability targets. Just kind of where you stand today? And what are the most important drivers to get to the new targets?

Sure. So back in early December, we increased the target model on gross margin from 63% to 65%, and took the operating margin target up to 42%. And we've made really good progress. I think that the strength of our gross margin was really a key highlight for Microchip over the last 2 years, which have been tough from a top line perspective. And we had some underutilization charges, and that was kind of the biggest short-term driver of gross margin improvement that we saw in the December quarter and what we're also seeing in the March quarter as our midpoint of guidance is 63.5% on a non-GAAP basis. From there, on gross margin, we've got continued expansion into existing clean room space. We don't need to add another large wafer fab clean room space, we've got room to grow into. We're increasing the amount of assembly and test activities that we're doing internally. Those take capital investments and we built into our forecast, and we've been making progress there. But we talked about where we're at from a percentage standpoint, and those percentages that we do internally have high returns as we increase those percentages over the course of time. We have some smaller, less efficient factories that we've acquired in various acquisitions that we can transition to our more efficient factories over time. And that we're going to continue to have a very disciplined pricing strategy. Pricing environment right now is obviously different than what we've seen in the past, with input costs going up and passing those costs on to our customers. But outside of that, we've got a very disciplined approach to keeping prices flat and getting the benefits that we're getting in the cost structure to flow through to gross margin. So investors should view on the gross margin side that it's a gradual increase as we go from this point at 63.5% roughly as we increase to that 65% target. And on the operating expense side, we're always quite efficient there. We had pay cuts last year, and those have all come back into the model and what investors are seeing for our guidance for the current quarter, which, at the midpoint, is about 23.4%. We're not that far away from our long-term model, which is 23%. And we know we need to continue to invest in technical resources and R&D and sales support to drive the long-term health of the business.

That's great. As we come up on time, just last question, and it relates to the business and the margins. Particularly through cycle, I mean, I think the company is known as a strong operator through cycles. You've managed the business well. But I think this last cycle, in 2019, '20, really stood out. And so if you can just touch on the durability of the business and margins? And how you think about it kind of cycle to cycle?

So I think if you look at the performance cycle to cycle, it has been always -- in every low point in the cycle, it's a higher low than the prior cycle. And every top of the cycle, it's a higher high. So higher highs and higher lows. But when the business is soft, the margins do go down a little bit. But our intent is to make sure that they don't go down too much. And as you mentioned in the last cycle, the margin didn't go down that much. And part of that is a strategy where if we can keep our factories full, especially in the assembly and test area, there is no package that we build more than 80% of it inside. So there's always some outside. So in the slow time, if we can keep our factories full by having that flexibility to bring it more in and similarly on the fab side -- historically, on the fab side, in the slow times, we have built a little bit of inventory rather than completely taking the fab down, we built a little bit of inventory. And when the cycle comes up, the inventory helps to keep the lead time short. This time, we were not able to do that because of the leverage on the balance sheet. People's memories are short, but in June, when I have these kind of meetings, investors and analysts will ask me, well, what happens if the business goes down 30%? What happens to leverage? What happens to covenants? What happens to this and that and that? A lot of the Street models were very, very draconian, and none of that happened. Certain businesses went down like automotive, but it got covered by data center, work from home, PCs and others. So in that environment, we put our factories on attrition. We didn't build as much inventory. And now as the business is rising, we find going into a strong cycle, our inventory is too low. That was the cost to pay because we didn't have a very, very strong balance sheet and keep building inventory in the last cycle. But in the next cycle, 2 years from now, or whenever that is, we'll have a much stronger balance sheet. And we'll be able to cushion the down cycle by continuing to build inventory on long-time products.

Got it. Excellent. Well, Steve and Eric, thanks so much for spending time with us today. Great discussion around kind of the business, the drivers and near-term conditions as well. So appreciate that, and hope you both and everyone has a good day.