Great. Good morning, everybody. My name is Rajiv Gill. I head up Global Semiconductors Automotive Technology Research for Needham & Company. Welcome to our seventh Annual Global Automotive Technology Conference. We're very pleased to have Microchip with us. Microchip was the first participant in our conference all the way back in 2017, so we're pleased to have them again this year. In terms of the format, this will be a fireside chat. For folks that are online, and I see a number of folks, if there are questions that you have, there's a chat box. Just type the question, I will relay the question over to Eric and Matthias, you know, we'll take it from there.
With us from Microchip is Eric Bjornholt, our CFO, as well as Matthias Kastner, a VP of the Automotive Division. Gentlemen, welcome to our conference.
Thanks, Raji. Appreciate you hosting us.
Absolutely. Maybe, Eric , if we could just start a little with kind of the near term earnings, and want to get your thoughts as well. On your recent earnings and guide, they were kind of in line after guiding kind of essentially 2 quarters. You also mentioned on the earnings call that you expected, you know, growth in June, and the September quarter, you talked about that quarter potentially being flat, not down. Now, is that kind of the still the right way to think about it, that you're expecting, you know, perhaps flat growth in September? That is kind of a deviation from some of the other large, kind of, analog microcontroller companies. Any thoughts on what you're seeing in the near term?
Sure. Well, let me give my normal safe harbor here and just say that as Matt and I are going through this discussion, we will be making certain projections and forward-looking statements about the performance of Microchip, and we refer you to our filings with the SEC that identify important risk factors about the company. That being said, you know, as you said, we've guided the current quarter to up 2.5% at the midpoint of guidance, and the quarter is tracking towards that. Nothing's changed since our May earnings call. We continue to believe that it is highly unlikely that our September quarter is going to be sequentially down.
You know, that is really based on, you know, where we're seeing the business, the backlog that we have in place, you know, what our customers are requesting from us, and our ability to service those orders. You know, clearly, we operate in a cyclical industry, and we're not immune to business cycles, but, you know, we are performing quite well, and we are navigating our business to what we hope to be a soft landing. To that, you know, that really means for us that, you know, if we are faced with a situation where there is a semiconductor cycle or an inventory correction, that Microchip manages to maintain very high gross and operating margins, as well as free cash flow. You know, we've done that throughout various cycles in the past.
I think there's various reasons why we are set up to even do a better job of that through the current cycle. You know, we've highlighted for investors that, you know, we don't see a situation where our trough non-GAAP operating margins could fall below 40%. Honestly, you know, we believe that, you know, it's going to be significantly higher than that. We remain cautiously optimistic about our ability to navigate to a soft landing. We have experienced relatively soft bookings quarters for the last couple of quarters, and that is just a function of lead times coming down.
You know, capacity in the industry is improving, you know, our internal capacity, the capacity at our foundry partners and subcontractors, and with that, lead times are coming down, and customers are not really required at this point in time to give us as much backlog visibility. That is really within our expectations.
Your lead times have been coming down, as you mentioned, and your bookings and backlog trends are a function of that, those lead times. What are your current lead times, and do these vary much by nodes or end markets? Are you seeing any variation across the different markets or nodes?
We have a wide variety of lead times. We have probably 100,000 SKUs in the portfolio. We have inventory that's in stock that we can ship in very short order, and we have other pieces of inventory that, you know, have 52-week lead times. We do have several capacity corridors, both internally and externally, that are still constrained. You know, we still have a lot of backlog sitting on the books. You know, we have still an unusually high level of backlog compared to what is normal, whatever normal is anymore. With that, you know, we do still have extended lead times.
Average lead times today are above 26 weeks, but we've proactively told our customers that, you know, we believe in the second half of 2023, our average lead times will fall below 26 weeks, which will be more competitive.
Got it. As the kind of backlog continues to decrease and lead times come down, the quarter becomes more turns based rather than shipping from the backlog. What any sense of kind of what % of the quarter is now coming from turns, where you actually have to go in the market and get business versus say, you know, just shipping from the backlog? What would that say a couple of quarters ago?
Well, you know, I'll just make a comment on the June quarter. You know, the June quarter, we really didn't need any turns during the quarter. You know, our backlog has been significantly high for a long time. We've had a lot of what you've heard us refer to as unsupported backlog, essentially backlog that customers have requested sooner than we've been able to deliver. You know, we're balancing that now with, you know, various customer requests for pushout activity. You know, we are accommodating that where we can.
It's a mix right now, and, you know, with 125,000 customers that we serve, there's going to be customers that are still scrambling to get the inventory that they need, and other customers that are potentially in an over inventory position or expecting to be in one based on the backlog that they had in place before.
Got it. In the September quarter, and looking out, do you anticipate that shift, that mix of turn versus backlog to kind of change, you know, significantly? We've seen other IoT companies or other microcontroller companies where in the quarter it's, you know, 60% turns business versus the inverse, say, a couple of quarters ago. Do you anticipate that shifting more to turns and getting back to more of a normalized, you know, bookings quarter?
That will happen. You know, we're not gonna pinpoint the quarter that specifically is going to happen, but, you know, we've had very short lead times in our business historically. You know, even when we get, you know, below 26 weeks, which we're shooting for in the second half, that will still be significantly higher than where we've been historically. You know, historically, we've had 90% of our products that have had, you know, very short lead times, you know, 8 weeks or less. When that happens and we get back there, and eventually we likely will, we'll likely have a high level of turns in the quarter. We've had some of the best quarters in Microchip history, entering a quarter needing 50% turns, and that's just the nature of short lead times.
Can you just give us some sense of what you're seeing in terms of the end markets? You're fairly, you know, diversified. You don't have as much exposure to consumer IoT as some other folks. Is there any kind of softness that you're seeing in some of the core secular markets that have been strong the last couple of years, whether that's data center, in core industrial, automotive, or are those end markets continuing to be fairly resilient?
The three markets that you mentioned have absolutely been our strongest markets for, you know, some time, and we've really only highlighted consumer as being weak. You know, there's definitely a shift that is happening, and, you know, we have got, you know, more and more requests for pushout activity from customers across the board. That's not just in one geography, it's not just in one end market. You know, Matt can speak to what he's seeing in automotive, but, you know, even there, where that portion of the business has been quite strong, you're gonna have certain customers that are requesting some relief in terms of their backlog position.
You mentioned that your unsupported backlog was greater than what was shipped in the quarter, and that's allowing your customers flexibility, and you've been... You know, guys have been very, kind of, smart about this Preferred Supply Program to help you kind of manage this, you know, this potential soft landing that you're anticipating. Can you talk about, your backlog versus what's being shipped? Can you give us some sense in terms of what's happening with the customer order pushouts? Are you seeing, you know, order cancellations? Are you seeing order rescheduling? Obviously, the lead times come in, customers don't need to place long lead time orders anymore. Trying to get a sense of how you expect to manage a, kind of, a soft landing when you see kind of this order volatility.
I get the fact that you are providing targets on the gross margin and the operating margin. I think that's very helpful, you know, for folks to understand, look, you know, we're not gonna go below a certain margin or operating margin. The top line is also of interest as well.
Well, you know, the bottom line is, the top line is the hardest thing to predict, right? You know, we don't know what the next few quarters is gonna bring from an overall economic backdrop, right? There's definitely some uncertainty out there. We are managing our business appropriately to be able to adjust to whatever the situation that's thrown at us. Yeah. Yes, we are getting some requests for cancellation. You know, we have this Preferred Supply Program, and under the PSP program, that is non-cancellable and non-reschedulable backlog. We are not, you know, kind of, you know, coming back on the non-cancellable piece of that, but on the non-reschedulable piece, we are willing to work with customers.
You know, the last thing we want is for customers to be in a significant over inventory position, but we also want them to know that they have skin in the game with this program because, you know, we've gone out and made commitments on purchases from foundries and our suppliers, significant capital commitments, hired people to run activities through our factories. With that, you know, this can't be a one-sided situation where heads you win, and tails I lose. You know, we're working through it the best that we can, and there's definitely a higher level of requests for pushouts than we've seen over the last few quarters. We're not necessarily surprised at that.
You know, we've got a, an economic backdrop that is not nearly as strong. With the supply situation improving, you know, that's giving customers the ability to say, "Hey, I don't, I don't need to have 52 weeks of coverage." With that, or their backdrop for their business has changed from when they originally placed those orders, we're working individually when customers self-identify inventory positions, to help them with that.
Just a couple more topics on the overall market before we go into the automotive stuff and bring Matthias in. You mentioned that the PSP, the Preferred Supply Program, accounts for over 50% of the backlog. In the last earnings call, you mentioned that the current backlog is still in significant excess of sales, and you're kind of working to bring that down, move to a more turns-based business. I would be curious to see how you, how we should expect the PSP program to kind of migrate over the next several quarters as a percentage of the backlog as we kind of move through this transition?
That is difficult to predict. I mean, what I would say is, customers that have participated in the PSP program have been serviced very well. Most of them have really liked the program and realized the benefits that have come with it. That being said, you know, the percentage of our backlog that's PSP, has gotten to a much higher percentage than I think any of us internally at Microchip anticipated at the start of the program. That percentage of total backlog has remained pretty consistent over the last 18 months or so, you know, we still are seeing people place orders under PSP. We've talked a little bit about these long-term supply agreements that some of our customers are entering into.
Obviously, all those customers have been PSP participants also. They are seeing value there. I think from a customer's perspective, you know, this last upcycle in the industry has been so painful on the supply side, that particularly when a customer sells a end product that has a very high dollar content, you know, t hey don't want to be caught short in not being able to get $20 in semiconductors or a $2 part, you know, preventing them from shipping an $80,000 automobile or, you know, a piece of expensive medical equipment, or data center, or whatever it might be. I think the program will continue for some customers, but I also think, you know, when we get back to, what I would call normal lead times, that it'll be a smaller percentage of the overall backlog.
We'll see how it trends over time. If this was put in place to be for a tool for our customers to use, and it's an optional program, and again, I think it will come down on its participation level, but I think there'll be some customers that are committed to it longer term.
On the automotive side, the PSP ratio is still very high, as the carmakers really suffered badly from lack of supply of individual, even $0.20 components. They want to make sure that their suppliers, our customers, are well-stocked, and they're asking them to really place long-term orders ahead. The ratio on the automotive side is much higher than corporate line.
That is very interesting to know. We're gonna get into that. I just want one quick question on the capacity. You mentioned that you're, you know, running the fabs, you know, through the downturn, even kind of while you're lowering the inventory, and the utilization has been kind of very high over the last year, and it continues today. How do you anticipate maintaining that utilization level through a potential downturn, you know, despite the elevated inventory?
It really goes back to the fact of, and we're seeing this, I guess, across a lot of companies, is that a lot of chip companies are carrying a lot of inventory on their balance sheet for a potential upturn in demand, while at the same time reducing the inventory that they have into the distribution channel. Could you maybe talk about your capacity expansion plans and your CapEx plans through a downturn? How does that affect utilization? The inventory that you're carrying on the balance sheet relative to the distribution channel, any thoughts there, and is there risk there of a potential, you know, holding too much inventory?
Sure. We did end this last quarter with what I would call an elevated level of inventory. We were at 169 days. Now, you know, the products that we manufacture and sell have very long lives, so there's no real obsolescence risk with inventory.
Right
levels. We are targeting to reduce inventory by about 5-10 days in the current quarter, and I expect in the September quarter, we'll take similar actions to reduce inventory. What I'd like to highlight on the inventory side, though, is that our internal fab-generated die bank is still very low. There's.
Mm-hmm.
A number of reasons that inventory is high, and we're proactively working on those. I'll give a couple of examples of that. When foundry capacity started to free up in the December and March quarters, you know, our business units, and our operations team, and customers, were starved for that inventory. We took in a lot of product from foundry in both the December and March quarters. We are going to moderate that in the current quarter and bring some of that down. We've also had to buy about seven days worth of inventory related to last time buys from our foundry partners, where they are end-of-life-ing processes, where we see 10+ years of runway from a sales perspective on very high margin products. We've made that investment in inventory. It's the right thing to do.
With some of the supply constraints going away, we are actively managing to bring, like, raw material levels down, which we've strategically built up when things were really tight, and as things are normalizing, we're going to be able to do that. Bottom line is, that, you know, our intention would be to continue to run our internal factories hard. As I said, our internal die banks, internally fab-generated die banks, are quite low, and that's how we get to having very short lead times, is having that inventory position in die bank, and then being able to turn it through the assembly and test process quite quickly. If there is an extended downturn, again, that's not what we're forecasting, but if there is, then you might have to look at fab utilization at some point in the future.
Mm-hmm.
We are fortunate today that the percentage of our wafer fab that we do internally is less than 40%, and, you know, that's different than what it's been historically. You know, we've got graphs on our website that look at gross margins over the last 15 years at various points in the cycle, and you'll see a 200-300 basis point decline in margins at weaker points in the cycle. During that time period, we had a higher percentage of internal manufacturing, and additionally, we were typically integrating companies or purchasing companies with lower gross margins, which were impacting, you know, kind of, the troughs that you'd see there on gross margins. You know, gross margins are positioned extremely well today. I think 68.4% is the midpoint of our non-GAAP guidance, and I'd anticipate that margins stay high.
Great. Matthias, just pulling you in terms of getting your views on the automotive market. Just broadly, how would you characterize the supply chain shortages in automotive? The industry obviously has been plagued by capacity shortages, semiconductor shortages for 2 years. Is there any signs that that's easing up? If it is easing up, are there any particular areas of semis that, where you see more capacity coming online? From your vantage point, you know, how do you look at, you know, distribution inventory, dealer car inventory, at the OEM level? Are we still seeing relatively low levels of inventory at the dealership, or are we starting to see that open up a bit?
I think, starting with the last point, the dealer inventory, this is getting replenished. Good measure for that are the lead times, the waiting times for new cars. Now, in the U.S., people go to the dealer and just pick a car that's sitting there. In Europe, we're ordering the car, and depending on the model, still waiting for more than a year to get the car delivered.
Wow!
Those times are coming down and getting better. From a semiconductor supply perspective, I think it's getting a lot better. There are still some technology corridors, some very specific ones that are constrained.
Mm-hmm.
Overall, the supply situation is easing, quite a bit. A good measure for that is the time that I spend on escalation calls. This has come down quite significantly over the last quarter.
This might not be an apples-to-apples comparison, but obviously, a lot of investors are concerned of, kind of an inventory correction. We've seen a large inventory correction in, you know, smartphones and PCs and consumer and IoT. As those industries overbuilt, double ordered during COVID, during the supply chain shortage situation, and demand has slowed down dramatically, and then you've seen basically this large inventory correction that's been going on now for about 18 months. There's concerns that that situation will replicate itself in the automotive industry. The automotive industry is different. Obviously, the lead times are design cycles are longer. You have secular trends of EV and ADAS systems, but, you know, there is this still general concern that this is the next shoe to drop in the automotive industry.
Were there signs of potential double ordering or overbuilding, you know, the last year? You know, the second thing, in terms of demand, the OEMs have been prioritizing premium vehicles over, you know, mass mainstream vehicles. That carries higher semiconductor content. It's, y ou know, interest rates go up for leases and things of that nature. Do you see kind of a mix shift down to more mass market vehicles where there's less than content, or is that still the-
I think the high-end vehicles demand for them still remains strong. There's not a big change. What the car makers didn't build were the lower end cars, which they now started to produce as well again because buy is there. So it's a natural mix shift, but not because the higher end cars are getting smaller in quantity. It's because they're now starting to produce the lower end cars again as well. They don't have all the features and not fully equipped. From an inventory position, we do see it, as Eric mentioned before, we do see some push-out requests from customers, and we do see some very limited cancellation requests. Cancellation requests, mostly when there was a unexpected or unanticipated end of model, for example, that they didn't see 12 months before when they ordered the parts. Those cases are relatively rare.
What is more common, that we see, customers asking for some push-outs just to manage their cash position themselves. [inaudible] as tier one suppliers are not in a very strong cash position at the moment.
Is the capacity, you mentioned that there is more capacity coming out of the foundries. Is that in a particular semiconductor component, whether that's more MCU wafers coming out, analog, power? Is there certain areas that are still heavily constrained by the foundries?
I think there are specific selective process technology nodes at the foundries that are still constrained, i t is not the most advanced process technology. It's not the three, the six, the nanometer technologies that are usually not get used in cars. It's, let's say, 28 or 40 nanometers and upwards, including the power technologies, and technologies that are used for in-vehicle networking and for the smaller type of microcontrollers that we're selling into the automotive market. It's not that the industry brought up tremendous amount of new capacity in those older technology nodes. It is that other market segments, like consumers, are weakening, and this capacity is shared on a wafer foundry level. There's no automotive tab as such. It is shared across the different market segments. That's why there's more available for the automotive market segment right now.
Got it. As these other end markets get weak, there's more capacity freeing up for auto. That's interesting. Just maybe a quick question on silicon carbide, and this obviously is tied to electrification of vehicles. You announced a $888 million investment for silicon carbide and silicon production. You know, both of those areas, converting kind of the 6-inch at mill site into an 8-inch site, and that was gonna become silicon carbide capable. Matthias, could you maybe talk about, you know, the rationale for that investment? And what kind of applications are you targeting for silicon carbide and for the 8-inch wafers?
Let me start with the application. The main application automotive for silicon carbide is for the traction inverter, but there are many other applications that are in our focus, like for example, smart fuse boxes, so electronic fuses, that disconnect battery in case of an accident or another unforeseen event to maintain safety of the vehicle. The onboard charger itself that converts the AC current to DC current that's needed to charge the battery. Also on the infrastructure side, the high-speed chargers, that are built along the roadside, that charge the vehicles up to 300 kilowatts an hour, they all benefit from silicon carbide technology, and those are the focus areas from an automotive perspective.
Silicon carbide, we've been selling silicon carbide in smaller quantities for many, many years in industrial applications, including semiconductor manufacturing, applications like implanters, et cetera, that need high voltage and high current. We do see opportunities for silicon carbide, not only in automotive but also in the renewable energies and those type of applications.
The competitive landscape in silicon carbide. It's relegated to, you know, three or four players. You have onsemi that is making a very big bet in silicon carbide. As you know, they're trying to be vertically integrated through kind of acquisitions. You have Infineon, STMicro, and then, you know, Wolfspeed, and then some other players on the peripheral. Other companies like Renesas have decided not to get going to silicon carbide. These are, you know, entrenched players that are making, you know, fairly sizable investments in terms of CapEx. You know, in the case of onsemi, they're vertically integrated somewhat vertically integrated, somewhat, or not.
You made an $880 million investment in silicon carbide and silicon production. I'm curious, How do you intend to kind of, you know, compete with those silicon carbide players? Is it that you want to, as you mentioned, target certain niche applications outside of traction inverter, where you feel you can kind of carve into that market? Or are you gonna go compete head-to-head on the traction inverter?
Hey, Matt. Matt, before you answer that, I just want to clarify. In that Colorado expansion of $880 million, the larger piece of that for Microchip is the 8-inch expansion compared to silicon carbide. Silicon carbide is significant and a really nice growing opportunity for us, we are not investing $800 million in CapEx for silicon carbide. Go ahead, Matt.
Okay, got it.
The applications that we're targeting, you mentioned, are more the niche applications that have specific product requirements.
Mm-hmm.
We do have a very good track record, for example, on robustness, radiation robustness, as well as those parts already used in avionics, for example, to control the flaps of the planes. We capitalize on those robustness characteristics of the product to enter those applications, like, for example, the smart junction boxes that are safety relevant.
Got it. Okay. In terms of the, you know, automation, the ADAS, the evolution from kind of L1 to L5, you know, if you look at a company, you know, like Mobileye, they've tried to change the taxonomy, moving away from kind of these ambiguous L1 to L5 categories and basically try to break it down into kind of 3 categories. 1 being, eyes on, hands off applications. You know, the 2nd category being where both your eyes are off the road and your hands are off the road. Then 3rd, ultimately down the road, no driver/driver needed. I guess my 1st question is, you know, how do you view that new taxonomy?
How do you view, you know, where you're positioned in terms of more microcontroller power management, LIN transceivers in each of those kind of evolutions of ADAS systems?
Okay. Whether it's five levels or three levels, difference is not that big. It's just a finer granularity. Below the hands-off is an important section today, which is you need to have the hands on the steering wheel. If you don't, the car beeps. The car is capable already holding the lane, et cetera, but you still have to have the hands on, and there's a lot of technology that needs to supervise the driver, et cetera. There's probably an L2 application, as it's in the five scale taxonomy. There's still a lot of growth for our products. If you look at the car basically is becoming a data center on wheels.
There are tons of data information from CMOS sensors, from radar sensors, from LiDAR, et cetera, that all are getting captured, being transferred and processed. Processed in big SoCs that we don't provide. Those are the known SoC makers that also invested in the processing stack. There's not only one SoC in a car, there are several SoCs in a car, and they all need to share data. When they need to share data, it's like in the data center. That's why we call a autonomous car or an heavily assisted car, a data center on wheels. To interconnect those SoCs and to hook them up to a high-speed memory, for example, and to hook them up to an Ethernet switch that has a PCIe connection.
PCIe is the connectivity of choice because it's native to the SoCs, it's fast, it's ultra-low latency. Because of our presence in the data center market, we are one of the big players in the PCIe connectivity business, and we're bringing this into automotive and are developing dedicated PCIe parts for automotive, and are qualifying data center-centric parts for the automotive market as well, for exactly this type of application. There won't be an autonomous car without PCIe connectivity and PCIe switching inside the car. That's one of the big focus areas. LIN is probably not the most important, and cannot the most important for those autonomous cars.
If you look at zonal computing or centralized computing, there's a very strong push towards Ethernet, because Ethernet helps to simplify the software complexity that we currently have in cars, because there's so many different standards used, from CAN, LIN, FlexRay, Ethernet, A2B, MOST, et cetera. They all come with their own software protocol and their own language that needs to be translated to each other, adding latency, et cetera. The beauty of Ethernet is that the upper layers, the upper software layers, are the same, regardless of the physical layer and the speed grade below. That's why we're also heavily investigating, 10-megabit Ethernet that can be used in a, in a bus configuration like CAN today, but with the beauty that it just connects to an Ethernet switch.
To the same switch, you can connect 100 megabit, 1,000 megabit or gigabit, 2.5 or 10 gigabit, and the switch itself is connected through a PCIe connection to an SoC. It really helps us simplify the software architecture and the software complexity, which is becoming the biggest hurdle to launch a new car. Software complexity is the biggest issue, and Ethernet being used throughout the car is an important element to break this down. We hear from some customers that their Ethernet spend will be bigger than the microcontroller spend in the future, not including the SoCs.
The Ethernet spend, that's interesting data point there. We just have, like, 2 minutes left. I just want to get one quick question on the You mentioned about the architecture of the car potentially, you know, changing with domain controllers and zonal computers. The domain controllers that have kind of specific compute elements for each automotive system, whether that's a body electronic or ADAS or drivetrain. Do you envision the new architecture having any impact on your microcontroller business, the number of microcontrollers that are needed to sell? I get the PCIe, I get the Ethernet, the interconnectivity, what about your kind of core microcontroller business? Do you see any impact, positive or negative, with these changes in the architecture of the vehicle?
I think with the consolidation of individual control units into larger domain, there will be a reduction in microcontrollers. At the same time, the number of edge nodes, which are the sensors, is growing quite rapidly. Some of the sensors need some intelligence, need some. For safety purpose, for example, a watchdog. Does it still work? What is the temperature? Can I rely on the data, et cetera. This is usually done by small microcontrollers, which is our home turf. I'm quite confident. If we were a big supplier of, let's say, gateway controllers that translate between all the different, let's say, five CAN nodes for LIN and et cetera, and back and forth, I'd be more worried. I think for the type of microcontrollers that we do, I'm quite confident.
The same when it comes to microcontrollers that have dedicated peripherals with more analog functions or DSP functions to control, for example, motors. This will not be done by a large central microcontroller in domain architecture. This will be done by specialized controllers that have exactly those peripherals. We're quite confident that it doesn't impact our microcontroller business in automotive.
Fantastic. We'll leave it there. Thank you so much, Eric, Matthias. Thank you, everybody, for joining. I greatly appreciate it. All right. Thanks, everybody. Bye.