Microchip Technology Incorporated (MCHP) Earnings Call Transcript & Summary

Got it. Okay, got it. I was going to ask about inventory next. I think as of today, inventory on your balance sheet is nearly 2x of what you had before the pandemic. And Rich, you just talked about some of the strategic moves that you guys are making. But how should we think about inventory management going forward? Is this sort of the new normal? Or do we start to come down over time?

It's fascinating. Pre -- during COVID, there was one population of customers that we didn't have delivery issues with. And majority of those customers were our Japanese clients. And so our Japanese customers based inventory prediction on cycle time, not on lead time. Lead time is the worst indicator for you to do your capacity planning on, right? That's not a true indication of how long it takes to build the semiconductor product. So a classic semiconductor product takes anywhere from 6 to 18 months, right? So as you start to drain, the only reason that we sitting at 8 weeks in [indiscernible] today is because we have filled up our inventory buffers, iBank, probe, finished goods, and so you're drawing from inventory. But as orders start to come in and we're trying to replenish those banks, that's when lead times start to go out. And so Japanese customers because of volcanos and tsunamis have all looked at what it takes to build critical products and they based their inventory buffers, their inventory buffers on the cycle time of those technologies. So they will have anywhere from 2 months to 6 months of inventory sitting on the shelf. And that particular customer base did not have the issues that the rest of the world suffered with. We thought that we would have more customers that would have learned from that because that would help smooth out the overall cycles in the semiconductor industry. We're finding a lot of customers are trying to revert back to the 6 weeks or 3 weeks or 8 weeks of inventory. Some have learned from that customer group, and we have some customers that are now assuming, "Hey, you know what, we're always going to keep 3 months. We're always going to keep 2 months." But unfortunately, I feel like we can almost schedule the meeting with some customers because they were reverting back to their old ways.

Interesting. Shifting gears a little bit, long-term demand drivers. I know Microchip in a very positive sense, you guys are diversified, right? Not one end market and application accounts for 20%, 30% of revenue. And I know you guys have talked about secular growth drivers over the past couple of years, particularly at your Analyst Day. But as you sit here today, what are you most excited about? You talked a little bit about AI, not only leveraging it internally, but also as a potential growth driver or growth driver.

A growth driver. I mean that is -- when it comes to data center right now is about 17% of Microchip's revenue and is continuing to grow. And AI or GPU-based servers are continuing to grow as well. Another area that is also pretty exciting on Microchip is FPGAs. We typically report once a year on FPGAs. And what's fascinating when we first acquired Microsemi, they were solely focused on A&D, Aerospace and Defense. And the reason for that is they had a unique combination of technologies that were required for Aerospace and Defense. So when you're -- you need an FPGA for, let's say, a Boeing or an Airbus and you have 100 FPGAs on that plane, it's running off a battery in the air, how you need to have extremely low power devices, FPGAs. When you put -- you're taking pictures on the James Webb and you're needing to transmit those video images back to earth, you need to have an extremely low-power FPGA that's processing that information and then sending that back right in order for it deliver for battery. Well, you also need to make sure that, that's secure, highly secure from a communication standpoint and nonvolatile, so you can reprogram it and update it if needed over time. So those 3 elements were the basis of the Microsemi FPGAs. We took that FPGA product line and then showed that to our industrial customers. And those were product attributes that they didn't think were possible in FPGAs right, the highly secure nonvolatile memory that's reprogrammable and extremely low power. In fact, 60% lower power compared to all other FPGAs, right? And so you take that combination now and people looking at new ways to use that technology, whether it's an ultrasound application, where you may have a technician with fatigue by using the heat that's generated in the ultrasound wand that goes over your body, now you can make that smaller with less heat so that you don't have a technician that's fatigued during the day if they're using that for 8 or 10 or 12 hours. IoT nodes that run off of batteries are now possible using an FPGA. Vision systems on the edge are now possible, secure -- highly secured because we have military-grade encryption on our devices, highly secure communications on the edge where needed. That were not possible for -- and that's why we're seeing tremendous amount of growth in that area.

Great. I just wanted to double click on GPU-based accelerated servers. What exactly is Microchip supplying into that ecosystem, if you will?

So we're providing PCIe Gen 4 and Gen 5 communication devices. We're providing timing on devices that all surround those chips. We're providing power semiconductors that go around those ships, we're providing the secure boot for those devices. So the security that also surrounds that chip where the leading a provider of secure boot devices for that technology. So timing, secure boot, power, memory products that may go around it, microcontrollers that do some support functions that surround the device as well. Ethernet is also provided by Microchip for those applications. So there's probably about 5 to 7 key technologies of ours that surrounds those GPUs that are provided in the marketplace today.

Not to mention support that we provide in the infrastructure, the AC to DC conversion historically or our 16 [indiscernible] dsPIC devices, and that's been there for a long time.

Great. Great. The TSS strategy, as you spoke to earlier, has been in place for a while. I guess my main question on this one is there are other companies that also have analog technology, microcontroller technology, maybe less so FPGAs. But how does your TSS approach differentiate vis-a-vis perhaps others that have a similar integrated approach when it comes to delivering solutions?

So it didn't -- Microchip was put in as a way to combine all of the technologies from all the acquisitions, right? So it's not just putting parts together, right? It's how do we leverage technology from one group to the next. So TSS starts all the way from product definition, right? And so we're coming out, for instance, we're coming out with the new 64-bit microcontroller, PIC64. We have teams of people from all the groups sitting down together, okay, what timing devices, what interface devices, what power devices, what PMCS do we need. And so all of the different BUs of the company now are developing the products that go with that product. So when a customer picks out that PIC64, there's a unified support for that technology. And so the TSS goes pretty deep in the company in terms of from definition of the products to how we go after markets and how we share in that overall growth. It's not just putting a bunch of parts together.

And your peers is more of that.

Right. Because a lot of our peers, it's still very much siloed in terms of the way the companies run. The way Microchip runs it, not siloed like that. It's very much where we're sitting down together in the group and we're saying, okay, how do we mix or match or how do we work together how do we fund this effort so that you have this product that goes along with that product. And so there's multiple team we call them anchor teams. And so we have an anchor team for dsPIC and anchor team for PCIe5 type devices and anchor team for our 64-bit devices or Ethernet products. And those anchor teams then bring in all of the other BUs to make sure that their product development strategy is aligned with their product development strategy.

Yes. And if I could add a little bit to what Rich is saying. So today, we're probably the only company that's innovating all the way from 8-bit micros to 64-bit and FPGAs. And the whole concept of TSS is largely to reduce the cost of capital that the customers are looking to go to market, speed to market and total cost of ownership, right? And this fits them perfectly with that strategy as well as we're almost kind of product agnostic, and we're looking for the solution that the customer is trying to drive and really focused in on that and then bring in all the best of the TSS part. And under the same software environment as well, right? So an engineer can start developing on the 8-bit side and graduate as you get more input for marketing and everybody else into 64-bit to 32-bit and they can move their codes around as well. So it really makes it easy for our customers, and that's the end goal.

I appreciate that. I am going to pause here and see if we have any questions from the audience. I'll keep going. So I wanted to pivot back to markets and demand and pricing. From a pricing perspective, clearly, over the past couple of years, it's been inflationary given all the dynamics that we talked about. I think on the most recent call, Ganesh reminded us that you're not in the commodity business and pricing doesn't go up and down in a week or a quarter or a month. But he did say from a design-in perspective, I forget the exact language you used, but perhaps being a little bit more aggressive, the market being a little bit more aggressive. How do you guys think about pricing, not in the near term but medium to long term? Do we go back to pricing being flat to down low singles? Or do we stay flat? How do you plan around that?

We're planning -- we're looking -- we look back at what our model was and that's what we're planning now. So you've got -- you're going to be highly competitive on the front end, right? It's why we're competing against whatever company to win that socket, like most mixed signal or analog companies, even microcontroller is considered more mixed signal than anything else. Those are designs that's been there for 10, 15, 20, 25 years, right? So you've got many years of layered designs, right, that sit there. And then you've got new design. So you've got these layers of designs that maybe 15 or 20 years old, where there's inflationary pressures causing pricing to go up. And then you've got new designs where the prices are going down, but also the cost of those components are going down. So the margins are relatively same. So you see some degradation in the ASP, but it's buffered by the old designs that aren't going anywhere but then you've also got new designs coming out where we're integrating more on that device, driving higher ASPs. And so it's a combination of mix, a combination of standard products and a combination of older technologies that are still going up in price. That's how you get that relatively flat or slightly down ASP range. At least for traditional analog type-based companies.

I guess the most extreme scenario, the question that we get is if pricing increased double digits per year over a 2-year through time frame, why can't it come down? Why wouldn't come down all the way, right, which is, again, very extreme, but based on what you just said, you don't really see that happening.

Yes, it doesn't happen because what happened was there was a great deal of investment by Microchip and the entire chats, supply chain. And the depreciation of those assets won't start rolling off for another 5 years or so, right? And so for a long time, it was this trickle-down economy of equipment and services, appreciated equipment. That trickle down has ended, right? There's no more trickle down. So as they're building these 2-nanometer and 3-nanometer and 6-nanometer fabs, those are all surrounded by analog microcontrollers, timing devices, and we're having to grow capacity in order to meet those demands, right? And the technology that's used to create those products do not have the ability to trickle down and service the demand anymore. That ship has sailed in the semiconductor industry. So if you have older technologies or lagging nodes, for you to expand them, you have to invest in capacity. Trickle down is no more, doesn't exist anymore.

Got it, makes sense. The other question I'm sure Sajid gets all the time is competition in China. It's come up on a couple of conference calls. Like what do you see from a competitive standpoint in China? Is it the usual suspects, the incumbents primarily in the West and some of the East? Or are you starting to see local competitors show up and do you start running into them?

We've been dealing with local competitors in China for 25 years, right? And so the idea that it's new, it's always been there, right? I think what's new is the demands by the Chinese government to have x percentage locally made. That's new, right?

I guess is there an acceleration or more intense competition, I guess.

The difference on -- at least on the analog front, the -- you don't have many designs, a number of our SKUs are maybe less than $1 million a year, right? And so our business is built on these SKUs or individual devices that are $0.5 million a year, $1 million. You see extremely rare that you have a singular product that's $10 million a year in analog, right? And when they do, everybody sets their sights on that product and it's trying to copy it, right? And so what happens is as these companies are coming up and trying to -- they're targeting those higher-volume SKUs. But then to layer on and go after these $0.5 million a year of devices will take years for them to get there. And that's sort of the way the analog business runs is always a new entrant. There's always someone coming in to try to go after the product line, but it takes years to build up large enough portfolio to do that, to make a substantial change.

Got it. That makes sense. From a gross margin standpoint, you're just above 60% at the moment. I think long term and always that's on depressed -- revenue depressed factory loadings. Longer term, 67.5% to 68.5% is your long-term target. You were kind of there, I think a couple of quarters ago. But the bridge from where you are today to those levels, is that purely just revenue utilization rates driven? Or is there more going on there?

I think it's a combination of [indiscernible]. I think it's a combination of utilization and it's also a combination of -- we have products that are driving substantially higher than corporate average margins, right? Our FPGAs are driving higher than corporate average margins. Our connectivity devices, such as PCIe 5 and others are driving higher than corporate average margins. So we're releasing a number of products that are driving or being adopted that are driving higher than corporate average margin. So you can...

So to your point, I think the 67.5% to 68.5% range remains are North Star and the actions that we're taking and our expectation is to get back to that 68% range at some point. The timing of that is somewhat unsure of it today because it will depend on the recovery curve as well. Obviously, as we kind of get back into growth mode, the inventory reserve charges will reduce, the utilization charges will improve. And then as Rich was saying, the richer product mix will help as well get us back to that level. So we're pretty confident that we will. It's just a timing part on when that happens is somewhat subjective right now.

Okay, got it. Just wanted to come back to FPGAs real quick since you made that comment. You talked earlier about taking sort of the A&D heavy exposure at Microsemi to something broader. Are you still in that process as in could that process alone drive continued outperformance in that market because I think based on what you reported vis-a-vis your competition, whether it be the classic Xilinx business or the classic Altera business, you're outperforming. So are you in the early innings of that process or are in the middle innings?

We are still bringing it into customers. We are still doing seminars or technology days at numbers of customers that are looking to want to enhance security are looking to significantly lower the power consumption of a product. So you're finding from an ESG standpoint or sustainability standpoint, many customers have to show significant power reduction in their products. And those products that are using classic FPGA technology are using substantial amounts of power. And that's where these FPGAs come in and solve a considerable issue with customers. And so I would say power reduction is probably a leading reason why people are adopting this technology, where they have to meet certain savings. And I'll probably say security is probably the second thing. And then we've done a good job with our FPGA group of trying to simplify the usage of the product making it more appealing and easier to -- for our clients to use. And it's sort of been -- that secret behind Microchip is creating a development environment that makes it easy for people to migrate to move around and adopt new technology. And we're trying to use those same learnings and work with our FPGA team and turned it into more of a mass market type product.

Got it. I guess in the last couple of minutes, I wanted to hit on 2 topics. One is OpEx and the other is capital allocation. From an OpEx perspective, you guys have this culture where you share the upside, and when times are tough like today, you should kind of share the pain, if you will. The OpEx reductions that we're seeing today, is any of that permanent or structural? Or as you come out of this downturn and you start to recover as the analysts should we model OpEx to snap back.

I'll let Sajid lead and I'll jump in.

No, no. Definitely. I think, yes. So majority of the OpEx reduction that you've seen is temporarily. Obviously, a large portion of it is the salary reductions that people have taken. The one component of it, which if you remember the last couple of years, the bonus, we pay quarterly bonuses and that was 2x, 3x the target range. So that aspect is probably going to be muted here. Obviously, get in a situation where that is background on. That will be a good problem. But -- so that -- and the salary reductions, again, expect to take a more graduated approach in kind of bringing that back, and that's what we've done historically. So a portion at a time, a slower recovery on the back -- on the OpEx side. And then on a longer-term basis, our target model is 23% of revenue. So we'd look to kind of settle in somewhere around that. Pre -- in the COVID time frame, we were running well below that at 20% because we just couldn't keep up with growth and now obviously, we're above that. So somewhere in that range is where we should expect us to settle.

Okay. And then again, lastly, in terms of capital allocation, historically, you guys had been quite acquisitive, and it's worked really well. Rich, you talked a little bit about tuck-ins. Steve has been super transparent about capital return. So I feel like we have all the information. But how do you think about priorities internally?

So we have a list of technologies that we're going after, right, or things that can bolt-on or enhance our TSS capabilities. And so when we have our megatrend teams and our anchor teams, we actually put together the technologies that would add to our solution that we're offering. And so we're constantly on the hunt for those assets. And can we get those at an affordable price that makes sense for the company? And so we're still looking, we're still acquiring, we just did 2, we'll do some more each quarter. Large acquisitions, we have to believe that they have to be closed and integrated within a 90-day period. Otherwise, there's too much leakage of technology or people or information. The reason why you buy companies or -- you're not just buying revenue, you're buying talent, right? And you want that talent to still exist there. And so we'd like to close larger transactions within that 90-day period, and that's almost impossible in this period of time. So we're going after smaller companies wherever possible.

Okay, great. In the last 90 seconds, anything that we didn't ask or we didn't discuss that you'd love to highlight or -- before we close?

We have a number of interesting technologies at Microchip. One that probably is not discussed much is just the amount that we do in clocks and timing and everyone's phone in here, the time damp on it is probably our cesium or atomic clocks and GPS signals are done by Microchip. Those are also growing businesses within the company itself, and it also is needed with all of the hyperscalers and driving that our time same products are all high-growth products within the company. And so it's more than just the mixed-signal microcontrollers. Now, it's really expanded. And I think with the PIC64, that gives a lot of customers some new choices that they didn't have before and building that into the same ecosystem that we have allows people to use code and the work that they've done and migrate it up and down that compute stream. And we're going to continue to widen that as well, whether it's adding CPLD or adding switching regulators or other nontraditional analog that you would typically see outside of microcontroller, we're going to continue integrating those onto microcontrollers.

Okay. Great. Awesome. Thank you so much, really appreciate the time.

Thank you.

Thank you, everyone, for joining.