Hello, and welcome to the Navitas Semiconductor expert call webinar, and thank you for standing by. My name is Alex, and I'll be your coordinator for today. If you wish to ask a question after the presentation on Zoom, please use the Raise Hand button or you may use the Q&A box. If you have joined us via the phone, please press star one on your telephone keypad. I will now hand over to your host, Adithya Metuku, to begin. Please go ahead.
Thank you, Alex. Hello everyone, and welcome to the first call in our GaN CEO call series. I'm pleased to have with us today, Gene Sheridan, CEO of Navitas Semiconductor, which is the number one player in the Power GaN space today. In terms of the format of this call, I have a list of questions that I will be putting to Gene, but if you have any questions, please follow the instructions for asking questions, and you will be given the opportunity to put your questions to Gene as well. Without further ado, let's begin. Gene, firstly, thank you for making the time.
Maybe, just to start us off, can you talk a bit about your background in the power semiconductor space, your International Rectifier background and why Navitas was founded?
Certainly, thanks, Adi, a lot for this opportunity to talk to you and your broader group here. We'd love to share a little bit of a history. In fact, Navitas was founded 8 years ago, but our history goes back much further. We were actually founded by a management team, largely of colleagues that have worked together for 20 or 30 years. Many of us started at International Rectifier. I started there in 1988 and quickly joined forces with Dan Kinzer and then in the 1990s, Jason Zhang. Those two guys are as good as they get in device and circuit engineering and innovation. We started multiple businesses and new technologies and major innovations that continue today to drive billions of dollars in the market.
Notably, we started that first GaN program at IR back around 2000. That was very early days, very much in the early material development phase, but we could see that potential that early on. We've always been big GaN believers, but a lot of investment is about timing, not predicting the future, but about when is the right time to go all in, whether it's your career, building a business or making an investment. Too early, and you could spend a lot of time in kind of fundamental device physics, which happened throughout the 2000 and 2010 period. Too late, of course, and you could miss extraordinary opportunities. Dan Kinzer and I and Jason Zhang all reformed together, got back together in 2014, started Navitas because we felt the timing was perfect.
The basic manufacturing and physical challenges, material challenges had largely been solved, but the design and circuit challenges and the customer and commercialization challenges had not. That was the real focus in formation of our company, 8 years ago. We're an 8-year-old company that's actually been working together for 20-30 years.
Got it, maybe can you also talk a bit about, you know, in the last kind of, you know, 4 or 5 years you've come from being a startup to being number one in the Power GaN market. Can you talk a bit about that journey? H ow did you get there? You know, what was your strategy? Then we can talk about the future.
Yeah. I think there's one overarching benefit that I think set up the other three. I think the overarching one is although we're a young company, we're actually deeply experienced with some of the best in the business in power semiconductors, in my humble opinion. Best from the standpoint of innovation, but also in terms of go-to-market and commercialization. I think we bring together a really strong team that's done this many, many times before, and that's certainly helping us. In particular, I'd say there are three big drivers. One, we knew that GaN struggles with how to drive the thing. It's very powerful as a transistor, very efficient. It's got so much potential. Many people ask then, "With such great potential, why has it taken so long to commercialize?" There are a few reasons.
1, they're difficult to drive, difficult to control. They are lateral structures, and most silicon power devices are not lateral structures, which opens the door to integrate some of those circuit challenges that our customers have been struggling with. We squarely focused on this drive and control problem, ultimately figuring out how to integrate drive and control circuits that the customers were doing in silicon externally, we figured out how to integrate them directly into the GaN monolithically. Which might sound trivial because everybody integrates everything in semiconductors, but actually this is a brand-new material. You cannot use any of the proven circuits from silicon. There's no CMOS here, there's no P-channel. The basic building blocks are actually very difficult. If you get them right and take advantage of this really fast and efficient material, you can do it, and that's exactly what we did.
We figured out how to integrate, drive, control and protection into the power device to solve that customer circuit adoption problem. We went a step further and integrated and invested in system engineering capabilities to teach customers and co-develop with the customers how to actually get the best out of the system once you have this great GaN IC. Third, took a unique go-to-market strategy. Rather than building a transistor and putting it out there for all markets to figure out how to serve it, we went very application specific, choosing one market, one major application at a time, and really nailing how to extract that GaN value and help the customers with our system engineering capability to ramp it up.
This slide is a really good view of the blue areas that GaN has the potential, but if you try to tackle all of those simultaneously, the result can be a lack of any success at all, or at least mediocrity at best. We actually chose the mobile charger market, which are wall adapters, plugs into the wall, but delivers a ton of power very efficiently and very high density. You can pack a lot of power in a very short period of time to fast charge your phone, tablet, or laptop. Now, that is the one mainstream market for GaN of all markets. Now we're turning our attention while we're continuing that exciting new market growth to two or three others.
You can see on this slide on the right, it gives you some of those GaN advantages I talked about. Lateral structure, very first on the top there is what opened the door for Navitas to do the innovations we did. Analog circuits, even limited digital circuits we've now innovated. Once we started innovating the drive and control dynamics, and now we've gone even further, sensing, protection, almost every other building block circuit that's done in analog or even limited logic, as I said, we're now integrating into the GaN device, which only makes the GaN system performance, efficiency, frequency, density, simplicity, cost, reliability. Everything gets better as we continue to integrate more and more of these layers of value and different power analog and logic circuits.
Understood. Now you previously talked about, well, you recently said, mobile market is where you're operating today, but I understand you recently introduced a 20-year warranty with your GaN products. Maybe can you give us an overview of that guarantee or that warranty and which markets do you intend to expand into using that warranty?
Yeah, definitely. In fact, we started with the mobile charger market, not because it's the end all be all for GaN or even the biggest value, frankly, but it's a fast-moving market that has a lot of volume with a very simple value prop, fast charging, lightweight chargers. This is something everybody can relate to. We design these things with our roadmap in mind. That's the starting point, not the ending point. Now we're taking GaN into big, big power, industrial power applications, where actually the energy efficiency is a driver all by itself. A consumer won't pay for energy efficiency at low power in consumer application, but they certainly will in a data center or in solar or an EV and so many others. This is where we're headed.
It's with that in mind, given our really strong track record, we've shipped 50 million units without a single GaN-related field failure. We've tested billions of device hours within our labs, achieving 20-year lifetime and beyond. With all of that track record, all of that confidence, and then the integrated protection and sensing circuits, which gives you added layers of protection, protect that transistor under all fault conditions and dangerous operating conditions. All of that combines together to give us confidence to set this new standard with a 20-year warranty when even silicon is typically 1-2 year warranty. Only the big guys usually can negotiate maybe 5 or 10 years or something. We're offering 20 years to everybody on all of our GaN ICs.
Understood. Now, I mean, this feels to me like if I'm a customer and I'm looking at a 20-year warranty, is there any reason why I would go with someone else other than Navitas? You know, maybe what are you seeing in terms of reception on that warranty? Is that helping you win a lot of share, especially against the big guys like Infineon, et cetera?
Well, I think these things work in combination. The warranty is very important. Reliability is a major consideration for power in general, and certainly for long life applications, 20-year expectations for solar, 20-year expectations for EV, 10-20 years for data center. It's a very conservative industry. I'd say that's the number one thing that has held customers back is just field track record. Nobody wants to see any trouble. People get nervous to be first. You need to give them a lot of confidence and a lot of data. We've done that with 50 million units shipped without a single field GaN failure with these billions of device hours that we've tested in our lab. Now we're adding another layer of confidence with the 20-year warranty.
I think you got to give them many layers of comfort to allow really big disruptive things to happen in the industry, especially as you bring them into the high reliability markets.
Got it. Are you seeing traction? Are you seeing more interest after you've introduced this 20-year warranty? Is that accelerating adoption?
Yeah. It was actually very timely because we've just started in the last couple of quarters, started sampling our brand new, even more protected GaN ICs that are going into these high reliability markets, so data center, solar and EV. I think that's where it's really paying off. It's obviously an opportune time as we're doing the sampling and customers are making decisions. Do they jump in with GaN? Is now the time? The reception has been fantastic, both to the warranty and all of our field testing, all of our lab testing, but also of course the devices themselves. To have the drive and control problem solved, to have the built-in protection, sensing capability, all of these things are building a lot of excitement.
We're really pleased with the response and see some big uptick coming in solar data center EV in particular.
Got it. Now on a lot of the investors I talk to specifically focus on the EV market because they see that as a big market. Y ou said this warranty is helping EV makers, you know, think about adopting GaN quicker than they would have otherwise done. Now the usual consensus seems to be that GaN and EVs, especially in onboard chargers, will probably not happen until the later part of this decade. Do you think your warranty could accelerate that?
The one thing about GaN, you can't drop it in late. You have to start from the beginning. Because you're really redesigning everything to really get that great system value, which is why we have these system design centers to help our customers. Everything is new in a GaN-based power supply, the control techniques, the EMI methods and filtering, the magnetics and transformer design, the thermal design, power density methods, everything is new, so there's a lot to do. Development times are shortening, especially in EV. It used to be back in the day, it was at least five years or six years. Things are getting more modular. They're happening faster. We're still looking at probably three years, so we started sampling earlier this year. Here we are in early 2022.
We think the first production ramp-up on EV is probably in 2025, which I think is consistent with your comment. There's nothing we can do other than giving them the confidence with these great samples and the devices, the innovation, the warranty, the field track record. They get going, but you can only go so fast, I think, with the development testing, the system integration and ultimately the field testing that they will still do, of course, for any of these new systems.
Understood. This EV market, is that specifically for onboard chargers, as I said, or is that for a different application?
Yeah, that is our focus. As I mentioned earlier, application-specific approach is a unique strategy of Navitas. It's for two reasons. One, there is a lot of system reengineering to do here and a lot of new skills to be brought to bear, and trying to do that for every application simultaneously is not practical, and you won't get a good result. Number two, our GaN chips are application-specific. We're looking at the application saying what integration makes sense that brings more performance, more reliability, simplify design, lower cost, smaller footprints, all of these, good things. The application-specific approach means that when we say EV, we have to pick. When we say data center, we have to pick where are we gonna go, and we pick the biggest ones, but also the highest value ones.
In this case it is OBC, but also DC-to-DC converter. The DC-to-DC is pulling the energy off of that battery, distributing around the car, where the OBC, of course, is delivering the fast charging capability. Traditionally, OBCs today, even on a Tesla or 6.7 kW, sounds like a lot of power, but it's actually really slow charging at your home, about 10 hours, which is overnight, fine in many cases, but many times you don't have that kind of time. We're gonna be changing that dramatically in terms of the power delivered, the efficiency of that energy delivered to the battery, and ultimately the fast charging time.
Got it. Now, you mentioned DC-to-DC. That's probably not the main inverter, right? Or is that?
That's right.
Main inverter.
That's right. We often talk about three main applications. There's lots of smaller applications, peripheral applications, but the three big ones for power chips, especially high voltage power chips in EVs, the OBC to charge it, the DC to DC to pull that energy off the battery distributor on the car, and then the traction control or the electric motor third, which we're not focused on yet, although I think it's time will come for GaN. In the short term, I think that's more of a silicon carbide play as a lot of people are talking about.
Okay, got it. Now you briefly talked about high voltage. Can you give us an overview of the different voltages you operate in? Do you see yourself as having an advantage at any particular part of the power semi market, high voltage, low voltage with your GaN products?
Yeah. Yeah, it is an important point, Adi. This slide here gives you the sort of voltage across the X-axis and the power level across the Y-axis. There's a loose relationship between the two. It can be a little confusing, but as you have higher power systems, you tend to want to operate at higher voltages to move that current around as efficiently as possible to keep that explanation a little bit simple. The sweet spot today for Navitas and for GaN is 600-800 volts.
That voltage range, as you can see in blue, covers a whole lot of ground because the grid, the power that we're pulling from the grid around the world is operating from 110-220 volts, and you want a lot of voltage margin from that voltage to the power device that sees that voltage and converts it and distribute it efficiently into low voltage DC power to power things, whether it's 5G base stations, solar, consumer applications, LED lighting. You can see them all listed there, including a big part of the electric vehicle market and the data center market is also included in that category.
Got it. Now one of the things that, you know, we've noticed historically with the silicon-based power semi market is the module capabilities. W ith silicon carbide as well, there is a lot of discussion around the module capabilities of the different vendors. Do you think the GaN power market will be a discrete market or will modules provide an advantage? If modules do provide an advantage, what is your strategy there?
Yeah. Number one, I think in low to medium power, which is a relative term, but let's say sub 20,000 watts, which you can see on the Y-axis, covers a lot of ground here. I think you're gonna see discrete packaging, if you will, but we believe GaN IC, even with simple drive integration. I should add that the integration we're talking about is extremely cost-effective. We're adding drive, control, sensing, protection and so much more, level shifting, bootstrap, all sorts of things in at pennies, in terms of our cost of a small fraction of that GaN chip, almost free compared to a GaN discrete. I think a GaN IC compared to a GaN discrete is very, very compelling and will therefore dominate in large parts of the market.
Above 20,000 watts, I think power modules, whether it's IGBT, silicon carbide, or GaN are very popular. What I think is exciting for a GaN power module or a smart module, today power modules usually demand a big price premium because there's a lot of complexity inside that module. While it's great for convenience, great for manufacturability, right? The simplicity of all of that, you just bolt it into your system, create a good heat sink and a good thermal path, and you're sort of done. You're paying a big premium today for that. Traditionally, with IGBT and even silicon carbide, I think GaN has the potential to dramatically simplify the interior construction and the number of components needed inside the module because of GaN ICs and the built-in protection that we could probably change that equation.
We don't have any announcements to make, but I think that's a real exciting promise of getting GaN smart, intelligent, highly integrated power module, and eliminate that premium that people are used to and frankly accelerate module adoption above 20,000 watts.
Got it. If I were to summarize that basically above 20,000 watts, you'd still need modules, but those modules would be far less complicated.
Yes. Even with that said, there's a split out there. There's no perfect cut-off line. Not everybody goes immediately to modules. I think you'll also see discretes in power packages with GaN ICs inside, highly integrated, a single version of it, a half-bridge version of it, sort of small integration levels that will also be popular. I do think kind of changing the game on power modules so they don't have that big premium, price premium, is a pretty exciting possibility.
Got it. How will you tackle the module opportunity, given historically at least, the view is that modules are hard to make. P eople like Infineon, Mitsubishi Electric have a you know, long-standing advantage there. How would you want to target that market? Would it be positive-
Yeah, I think like we've done in discrete packages, if you will, there's a chance to reinnovate and rethink how that internal construction, not only in the number of components, as I mentioned, because of our GaN ICs integrating many of them, we can reduce that premium of modules, but it's also a lateral shift. Traditionally, you have vertical power devices also being constructed with passive components and other lateral structures side by side. This creates a lot of that internal construction and added cost challenge of power modules. With a GaN IC that operates a very high frequency, it shrinks dramatically the size and even the need for some of the passive components. It eliminates the need for the secondary lateral shifts that are in there.
You can actually reinnovate, kind of create disruptive, new, innovative, new packaging techniques that not only change that cost structure, might also change the supply chain, where that traditional strength has been, over the years. I think that kind of innovation opens the door to new entrants like Navitas and possibly new supply chain partners that can make that happen away from where we're seeing traditional strength with the bigger players.
Got it. Now, when I look at the market share data in the power GaN space, I mean, the market itself is relatively small, but it feels like the top five players were all relatively smallish players, startups, really. I just wondered, you know, how do you see competition ramping in space from the, big power semiconductor incumbents? Specifically, how do you view STMicroelectronics and Infineon as competitors in this space?
Yeah. I like our chances because power is a very big market. It is dominated by big players, but with size comes some challenges. It's hard to be nimble. It's hard to make fast decisions. It's hard to take a lot of chances. A lot of people in semiconductors are more focused on M&A than they are organic investment and innovation. I think those are fundamental industry observations that create big opportunities for fast-moving visionary, disruptive companies like Navitas that can really change the game. We actually went back. The last time we saw something this disruptive in power was back in the seventies and eighties when we had bipolar transistors. It was actually very analogous to what we're seeing today. Bipolars dominated being used in linear regulators, very inefficient power supply. It's very expensive, very big and bulky.
A power MOSFET came along, opening the door to a switching power supply. Everything in the power supply changed. New controllers, new architectures, new EMI techniques. Everything I mentioned earlier, you had to redo it. It created the door for analog ICs that created powerhouse companies like Maxim and ADI and Linear Technology, right? Didn't exist in the seventies. They created application-specific ICs to go with those power MOSFETs. It opened the door for new leaders like International Rectifier to create their power MOSFET leadership throughout the eighties and nineties. If you look at who the top 10 power semiconductor companies and power supply companies were in the seventies, and then fast-forward just a decade later, almost no company within the top 10 of power semiconductors and power electronics that was also in that top 10 list just a decade later.
That's a classic scenario where disruptions come and the big guys are saddled with their traditional technologies, are not seeing the disruptions come, are not aggressive enough to obsolete. They're protecting their traditional assets and capabilities. Major disruptions can come. All new losers, all new winners. I think that's a pretty exciting thing. We have not seen this before until this decade, and I'd argue we're only two or three years into a decade-long major disruption. Of course, Navitas plans to be that next-generation power semiconductor leader when that decade is over.
Got it. Specifically, I mean, you know, there's been a lot of talk about China building its own semiconductor industry. I noticed Innoscience is one of the players in the GaN market. I just wondered how do you see them as a competitor? Yeah, any thoughts on whether, how the Chinese will fare in the GaN market?
Yeah, I think, of course, it's a big market, it's no secret, and it's happening finally after years of development. We're gonna see many players, but they're all in the category of GaN discretes from our perspective. None of the others have solved this fundamental problem with how to drive and control and protect these things, to give them the reliability and ultimately get the speed and efficiency at a GaN at the system level, which is so important ultimately to the broad adoption. Innoscience is like others in that discretes are gonna be there. I think they'll have a role in the market, but I think they'll be limited in the broad mainstream adoption. You can't price your way into performance. You can't price your way into fast charging and high efficiency.
In the end, that's what people want from GaN. We're changing the profile of what you can do with power electronics in major new markets like solar renewables, broadly, data center, and EV. Ultimately low price only gets you so far when you're trying to redefine the power density, the power efficiency, the frequency, the fast charging capabilities of these power systems.
Got it. Just briefly on my previous question. O n you said the large guys are, you know, will struggle to be nimble, et cetera. Do you think Infineon and ST will be relevant in ten years' time or are you planning to eat their lunch?
Well, I think they're capable, respectable companies. I'm not here to criticize the competition as much as state that there is a really extraordinary opportunity for new leaders to emerge and become major players in the world of power semiconductors, and that's certainly our intent. Starting with GaN, and GaN will always be critical. We're also suggesting, when we talk about being the next generation power semiconductor company, GaN's not even in that kind of mission statement, if you will. We are looking more broadly at other complementary technologies that will make us the mainline next gen power semiconductor company that I described.
Understood. Now since you talked about other technologies, can you talk a bit about your plans in the silicon and silicon carbide space, and how you intend to build up your capabilities there, how you intend to differentiate?
Yeah. We've been pretty open that last year with our IPO, we raised nearly $300 million to reach cash flow positive. With our current GaN business, we shouldn't need more than $100 million. We had $253 million on our balance sheet at the end of Q1. That leaves us a lot of dry powder. We can put that dry powder to work in lots of different ways. Additional organic investments, joint venture, ventures, partnerships, or of course, acquisitions. We could do more in GaN, even though we love our position. Obviously, I wouldn't rule out more investments or more acquisitions in the field of GaN. Every next generation power semiconductor or power system that uses GaN still uses silicon for the controllers.
Low voltage brain, low voltage chips, as that chart earlier showed, will likely stay with silicon for a long time to come. There's an important place for silicon in these next generation power systems. Above a certain power and voltage range, as this chart shows, silicon carbide is still gonna be the preference, even though GaN's kind of the new kid on the block, and we're coming on fast, especially in these new markets we move to data center, solar and EV. At some point, the vertical structure actually of silicon carbide shines through. It can handle really high voltages. It can handle really high temperatures very well because of its temperature coefficient. Above a certain power and voltage range, we still think silicon carbide is the winner, as we're already seeing happening in the market today.
That's a very interesting area for us because it's largely a perfect complement. While there's some overlap, and there'll be some kind of interesting debates about GaN versus silicon carbide, around 1,000 volts or less, when you go above 1,000 volts, there is no commercially viable, GaN. We think silicon carbide is a very interesting investment area for us going forward.
Got it. Are you able to say any more on silicon carbide and what you might do there? You know, any further detail or?
No specifics at this time. It is a big market already today, much bigger than GaN because we are the new kid on the block. It does take time to develop. If we did it organically, it will take some years. If we did it through acquisition, that obviously could speed it up, but we won't have anything to announce until we've got one of those things to give some specifics around.
Okay, got it. Then moving to your strategy with manufacturing GaN products. Can you talk about you know, how you manufacture them? You know, what gives you an advantage if you're using a foundry to manufacture them and yeah.
Yeah. We are working with TSMC exclusively at this time. GaN has another really interesting angle. While the material is super advanced, as we talked about, Navitas' design is highly differentiated and very advanced. GaN can be manufactured on very low-tech tools from a semiconductor perspective. You can basically retrofit older 6-inch or 8-inch fabs with very old geometry technology. The one at TSMC is Fab Two. Fab One's not even in production anymore. Fab Two is the oldest one in production. I think it was built in the 1980s. These are highly depreciated, highly utilized fabs. Silicon is largely moving on to 12-inch, the FinFET or 65-nanometer, whatever it might be, opening up these fabs that otherwise might get mothballed. For pennies on the dollar, very limited capital, you can upgrade an old silicon factory to be excellent for GaN manufacturing.
That's exactly what TSMC did. That gives you a great cost structure, and it can run GaN and silicon simultaneously. I think the fabless model is actually really attractive, especially at this phase, because I can leverage older factories. You don't wanna go build a billion-dollar factory. That's what Innoscience did, actually. It doesn't really make a lot of sense when GaN has this wonderful ability to retrofit older investments and run right alongside of silicon, which means you want. T he key for economic advantage in semiconductors is to have fabs 80% full or more. The GaN can run right alongside the silicon. As the GaN ramps, silicon tends to ramp down because it'll move on to the smaller nodes in the more advanced factory.
I think that fabless model and this capital light model makes a lot of sense for the GaN industry in general, and certainly for Navitas and TSMC.
Got it. Now if you're using TSMC, I think some of your GaN peers may also be using TSMC. I think EPC is, if I'm not mistaken or maybe it's not EPC, it's someone else. How do you intend to differentiate versus competitors? What would your strategy be if you were to move to silicon carbide, which you said earlier?
Yep. The big differentiation certainly for us are in the three things I talked about at the device level is the high level of integration. We're really the only one. Certainly the first, I would say the only one in high voltage GaN. When we announced our GaN IC 3-4 years ago, almost every GaN company said, "Sure, we're going to make GaN IC." You know, logically, everybody said, "That's so compelling. That makes so much sense. Why doesn't everybody do it?" It was on everybody's roadmap. Here we are, 3-4 years later, nobody's producing even samples, let alone production availability of the IC. The one exception would be, as you alluded to, in low voltage GaN, where we're not participating today, Efficient Power Conversion or EPC has done some good work in GaN IC.
It's a little bit easier to do some of the integration we talked about at lower voltages, and they've made good progress in that category. That's a big differentiation for us on top of the system design centers and the system systems and co-development we offer to our customers. Finally, that unique application-specific go-to-market strategy that kind of ties it all together, one major application at a time. Together, we bring a whole lot of system and semiconductor or circuit innovation and value to our customers in each of those targeted areas. As you look at silicon carbide, that is a vertical structure, so it's not gonna be as conducive to integrating other circuits, analog circuits that are traditionally done in analog. That will be a problem.
I do believe the system capability we're bringing together and the circuit integration innovations that we're doing in GaN, we're also very capable of doing those in silicon. I think there's a parallel to the market approach, to the application specific approach, and to the system circuit innovations that we can bring to silicon carbide, even though it may not be monolithically integrated in the silicon carbide transistor itself.
How would you manufacture it? Would you then have to use a foundry, and would you still have advantages versus IDMs?
Yeah, good question. There's not much to announce yet. Obviously, we're not in that business yet, so this is all a little bit of a theoretical discussion. I think that market is maturing as well. There's IDMs, which are vertically integrating, but there's multiple fabless options, foundry options out there. Here again, in the relatively early stages, and even in silicon carbide, I would argue we're relatively early. You want to leverage foundries that are already running very full. You get the economic advantage of a full fab, in that case, running silicon carbide with other device technologies. I think can outweigh the additional margin you would pay to that outside foundry to leverage it.
Where you need to be later on, you know, IDMs and vertically integrated may start to make more sense.
Got it. Maybe at this stage, Alex, are there any questions from the audience?
Just as a reminder, if you'd like to ask a question on Zoom, please use the Raise Hand button, or you may use the Q&A box. If you have joined us on the telephone lines, please press star one on your telephone keypad. We have a written question from Hudson Hoville, who asks, "Is built-in driving protection, etc., possible with vertical devices on silicon?
Yep. I think that's a similar question. I think when it's vertical, silicon carbide or others, it's gonna be more challenging, if not impossible. You got to be creative, and it gets more expensive to try to integrate lateral circuits, which analog circuits or logic circuits, of course, traditionally are with that vertical device. You have to think about other ways to create the integrated system benefits and circuit benefits that we've talked about.
Got it. Any other questions, Alex? Oh, there's a question from Gunther.
Oh, yeah, a question from Gunther. Gunther, please go ahead and state your company name.
Can you hear me now? Hello?
Yes. Yes.
Hi, it's Gunther Holzheuer from Polar Capital. I mean, you recently announced the hiring of CFO, also mentioning the M&A experience. Now in this call, you're talking about M&A, about your excess cash and also in relation to silicon carbide. I was just wondering, I mean, in the current market environment where, you know, some, a lot of tech companies are struggling to survive, and you're in the comfortable position with a strong balance sheet and with cash, and cash probably is gonna be king in a very difficult market environment, potentially in the coming months and maybe quarters.
At the same time, the silicon carbide, I mean, in contrast to gallium nitride, I've seen silicon carbide, but the big power semi companies, they're already very strongly involved in this technology. It's also a more mature market, I would say. If you could maybe talk a little bit why you think it makes sense at this time to think about silicon carbide with and the silicon carbide entry.
Sure, yeah. Gunther, good questions and good topics. Certainly, we're not making any announcements now, talking about the future and the future potential as we see it. One perspective as we talked about, again, is silicon carbide are extremely complementary more than they're competitive, and allow a company with the aspirations like we do of becoming that next generation power semiconductor company to have the tools in the toolbox that can address broader markets. That appeal is sort of obvious. I also mentioned our unique go-to-market strategy, which I don't think even the big guys are applying in these markets to bring a lot of system value and circuit innovation rather than just offering a better transistor and allowing the market to sort of figure it out from there. I think that can also bring us good value.
I don't wanna put too much weight on this topic because we don't have anything to announce at this time. It's a little bit of a theoretical discussion about the possibilities for the future, and I agree with your thought that cash is king. We're gonna be awfully careful with our money. We're not gonna gamble with that money. We're not gonna take big chances and put the company at risk. We're gonna focus on cash flow and cash flow to positive cash flow, runway to positive cash flow and positive EBITDA over the next couple of years, and make sure we use that cash, you know, very judiciously to get us there.
Thank you.
Yeah. Thank you, Gunther.
Gene Sheridan, maybe could you remind us, did you say you were going to be cash flow positive this year?
No. In a couple of years, actually. We're estimating 2024. That's why I mentioned sort of reserving $100 million for organic business to make sure we're on track to that. As I'm implying with, you know, any M&A things we might contemplate will be certainly very sensitive to cash and that runway to cash flow positive. Frankly, trying to accelerate it, not doing things that would increase our risk on runway and cash conservation to profitability.
Got it. Now, moving to substrates, there's been , a lot of the investors on this call will know silicon carbide and the issues with getting silicon carbide substrates. Can you talk about how easy it is to get substrates within GaN? Is that an issue? You know, are you having to... Is that a bottleneck at this stage? Can you talk a bit about the cost, who your suppliers are, et cetera?
Yeah. In fact, we're quick to say GaN, but we should take a step back. Actually, when we say GaN at Navitas and virtually all of the GaN power players, it's actually a very thin GaN epi layer on a silicon substrate. Since all the circuitry, the power, the current flow, the voltage, that's all along the surface, lateral, as we said. That silicon substrate is really more like a mechanical carrier. You don't have the circuits in the silicon, you don't have the power flowing through it, but it's also very cheap and inexpensive, which is why it's a perfect mechanical carrier. In fact, it's relatively low quality silicon, so the cost of that for 6-inch is pretty small, you know, $30, $40, $50. Really, really inexpensive. That's one of the challenges with silicon carbide.
It is vertical, which means the entire power device from top to bottom needs to be in this high quality, really efficient material, silicon carbide. You have silicon carbide epi on top of silicon carbide substrates. The challenge is growing those wafers or substrates is time consuming and very expensive, and the quality of them is very demanding. I think the cost of a silicon carbide substrate is something like 20-50 times more expensive than that of silicon today. Of course, it will come down over time. I think those are some of the differences. People are playing with different substrates on GaN, mainly to see if you can take GaN into the higher voltage and higher power regime of silicon carbide. That's certainly something we're investigating and monitoring.
People are looking at SOI and GaN-on-GaN and GaN-on-silicon carbide. All of these things have their pluses and minuses. I don't think anything has near-term viability or commercial potential from our perspective, but it's something we're investigating and monitoring going forward.
Got it. Just, you mentioned GaN on SOI. Can you talk a bit about what are the positives and the negatives of GaN on SOI? W hat would you need to see for you to adopt GaN on SOI?
Yeah. I'm no expert on SOI or even all of the different substrate options, but I think in general, you're trying to figure out methods that can handle that higher voltage in a tiny chip. At some point, handling, you know, 2,000 volts, 3,000 volts, 4,000 volts along the surface of the device is a challenge, and you wanna look at vertical structures, where you need to then have the power flowing through that structure. In the case of SOI, it's more about insulation. You can get some challenges in how some of the bias currents or leakages flow through that silicon substrate. SOI will give you an inherent insulation element to minimize some of those challenges.
It's expensive, and so you've got to really think about what are the unique applications where that's really a problem, and you're willing to pay the premium for something like an SOI substrate.
Got it. Would using SOI substrates, GaN on SOI provide any advantage in terms of integrating logic onto the device, or are you already able to do that really well with silicon?
Yeah. Not that I'm aware of. We've really nailed that. That was the first thing we did with generation one, was the driver integration. That's still fundamental to the value because they're so difficult to drive these very sensitive inputs or gates of GaN, and that's true for everybody's GaN. Integrating the driver is something we really perfected with low cost GaN on silicon. Now, of course, we've taken that to a whole nother level. Not just driver, but level shifters and bootstrap, the sensing, the control, DV/DT or EMI is a tricky thing, and actually being able to control how fast you switch, not too fast, not too slow, is a tricky balance. We've got all that integrated in.
The levels of integration continue at a pretty rapid pace with each generation that we develop, and we're doing all of that. We see a lot of potential to keep that innovation going and that integration going with GaN on silicon.
Understood. Now, one of the players I spoke to, one of your competitors, said that the driver that you've integrated is not the same as, you know, it's a relatively low level driver. You know, is that understanding correct? Would you be able to integrate more into these monolithic ICs going forward?
Yeah, we've actually done a variety of them. Some are simple, some are complex, depending upon the application need. We have a half-bridge integrated, which has not only the two power devices, high and low side power devices and their drivers, but also level shift functionality, bootstrap functionality, DV/DT control and protection. It goes far beyond actually a simple driver in many of the implementations. The half-bridge is an example of that. That's a lot of layers of driver elements, kind of like a smart driver. A gain, not just one device, but two. Even how you do the level shifting between the high side and low side device is pretty tricky. We have pretty advanced driver integration now in GaN, and I think that will continue.
Got it. It feels like that's one of the key things why you guys went, from number two to number one. Correct me if I'm wrong, but are you seeing your competitors also adopt it? You know, how will you stay ahead of the game in terms of integration?
When we started the company and we knew this driver was a big problem holding everybody back, we actually did a little patent search, and it was a real greenfield. There was nothing out there. That was just indicative of the opportunity in front of us. Not only did we innovate all of these driver related building blocks, we also modeled them, we patented them, and now we've built up a patent portfolio that's over 150 patents issued or pending. Anything and everything you can think of that are common in silicon but hadn't been innovated in GaN. You really do have to innovate them, not only how to make them functional in GaN, but fast and efficient using that awesome material of GaN. That's really been a key strength.
That was just generation 1. Well, we're sampling generation 4 this quarter, starting production next quarter. I think if anybody's gonna be out there trying to copy us or follow in those footsteps, obviously you've got to navigate the patent portfolio. There's an incredible amount of know-how, not just our last 8 years, but our whole careers sort of leading up to this point. Then ultimately, you know, you gotta move like the wind, because if you copy our Gen 1 or our Gen 2, we're introducing a new generation every year. That's a pretty rapid pace. We know we've got to keep that up to keep this number 1 position that we've acquired and keep that going and frankly strengthen our lead.
Got it. Now on the patent side, you know, I remember a slide from when Infineon bought International Rectifier, where they showed International Rectifier having, you know, I think the number one patent portfolio in GaN. I don't know if that was specifically on the transistor side or if it involved associated things around the transistor. I just wonder, I mean, you know, despite that you've seen, you know, a number of smaller companies come up in the power GaN space recently, including yourselves. Is that patent portfolio really going to prevent somebody like an Infineon or ST or, you know, some of the bigger guys from monolithically integrating driver ICs?
Well, I think you always have to have a multilevel competitive strategy. No one dimension is adequate, just like even no one patent invention is adequate. I think you have to have lots of layers to these things. Some things you don't even patent, they're not public. You don't want them out in the public. Our PDK, for example, the entire know-how of how to create all of these popular analog and power building blocks in GaN, we put into our PDK, which is our device and circuit library that's characterizing everything, modeling everything, and that's how we actually build up these circuits. None of that is in the patent. We want to protect that and keep it as one very large trade secret. All the system know-how we've done. How do you...
Once you have this amazing GaN IC, you still have a lot of work to do to get the full system value. Like I said, new controllers, new magnetic, new EMI, new filtering, so many new techniques. There's a lot of know-how there that's not in patent, that's not easy to duplicate. That ultimately creates a lot of stickiness with customers. I think our business is really all about trust, and a lot of that trust is technical trust. Not just showing up with a great device, but showing up with a lot of know-how that can help the customer navigate this exciting but disruptive time of GaN or silicon carbide-based power systems.
I think all of that has to come together, and I think any large competitor or small competitor has got to navigate all those different layers of value more in competitive advantage before they're, you know, copying exactly what we do.
Got it. Now, you mentioned a lot of the advantages of GaN. One of the things, you know, when we talk to somebody like an EPC, they're also very vocal about the cost benefits, the power benefits, you know, or power or reduced power consumption, more power efficient products. Yet, adoption seems to be relatively slow. What is holding up adoption of GaN? Is it mainly customers wanting more comfort around reliability? Is it , just the longer lead times, the longer cycle, you know, production times in some of the markets you operate in? What exactly is holding it up?
I would say three or four years ago, everybody was excited about GaN transistor and frustrated they could not get the commercial value, certainly not economic, commercial significant value at the system level. That was largely this drive and control challenge. You ended up adding so much extra circuitry and silicon around the GaN, and the silicon drivers would slow you down, add cost, add complexity, and degrade performance. I think that was the big thing. Of course, we solved that with the GaN IC. The next big thing became, well, reliability. It's a very conservative industry. They don't believe PowerPoint. You have to have data. Lab data is good, but field data is ten times better. On that, again, it's kind of why we started with the mobile market, 'cause we could get data very quickly, a lot of volume quickly.
Now, it's not 10-year life and 20-year life applications, but you need to get that field track record. Now we've got 4 years under about 50 million units, not a single GaN failure. I think that's quickly going away. We added more protection circuits. We added the 20-year warranty, and now we have the GaN IC to go into those high reliability, bigger, more industrial markets like data center, solar, and EV. I think you're left now with just 2 things. The system know-how on how to take a great GaN IC and make a great GaN-based power system. That's why we have a mobile design center that helps all of our customers do the system design for mobile chargers.
Last year, we opened one for data center, so we're in the middle now of designing data center power supplies with our customers, for our customers, using our GaN IC. This year, we announced one for EV. Again, I say broadly EV, but it's squarely focused on the OBC and DC-to-DC, like we said. They're already active now designing onboard chargers with GaN, with our customers concurrently to make this happen. I think system know-how is gonna continue to be a challenge. There's gonna be limited resources in the world on people who know how to do this, and this is why our go-to-market application-specific strategy is so key. Our system design centers are so key. The final one is cost. It's still a very price-sensitive market.
You're bringing a lot of value, but at what price? That is important, and we're equally aggressive on driving system cost reduction and component cost reduction. There's a lot of pieces to that, not just the GaN chip or GaN wafer price coming down. Each generation, more integration, lower system cost. Each generation, higher operating frequency, lower system cost. By reducing the cost of the passive components in the system, all of these are working together. Higher efficiency means you have less thermals to manage and reducing the cost of thermal management. All those work together to where we confidently predict by next year, we're at system cost parity with silicon, and in 2024, it's a system cost reduction. I think that's the last layer.
When you nail the GaN IC to drive problems, you nail the system reliability concerns and the warranty and all that, you nail the application-specific know-how with the design centers, and now you get that system cost premium to be very small, if not zero, or even a cost advantage. Now there seems to be nothing holding back this whole market from taking off. Although I still think it takes off one major application at a time, because it does take that concerted effort in each of those segments.
To essentially keep an eye out for 2024.
Well, I think, the fact is you start designing today in many of these applications for 2023 production, for 2024 production. D ata centers will start ramping in 2023, solar will ramp in 2024, and we think EV, as we said earlier, probably ramps in 2025. T hat's not sitting and waiting. It's actually doing the system work. They just take more time with a lot of their system integration and field testing.
Got it. Just to confirm on the EV comment, that's the OBC ramping in 2025, not DC-to-DC.
Probably a combination. In fact, there's a trend to create hybrids where you actually enclose both of them in one system. A lot of those projects are running concurrently. We'll probably see both OBC and DC-to-DC in 2025.
Got it. Maybe at this stage, we've got another 5 minutes left. Alex, are there any questions on the call?
As a reminder, if you'd like to ask a question on Zoom, please use the Raise Hand button or you may use the Q&A box. If you have joined us via telephone lines, please press star one on your telephone keypad.
Maybe while we're waiting for any questions to come through. Now, your focus so far has been on the Power GaN market. Do you have any plans to be in the RF GaN market? You know, that currently is a bigger market than Power GaN. I f you do intend to go into that, what would your strategy be? What would you need to build in terms of capabilities, et cetera?
Yeah. We're really all power guys. You know, power is a big market, but also unique. There's not a lot of universities that teach it. A lot of it is dedicating your career around power electronics. Even though the material sounds like it's all the same stuff, power for GaN for power, different substrate, by the way. Usually GaN on silicon is what's dominant here. You're gonna tend to use a different substrate. More importantly, I think the device design, the circuit know-how, the system capabilities, all very different and very unique. We plan to stay squarely focused on the power space. When that's a $20 billion industry that's probably going to $40 billion or $50 billion in the next couple of decades, I think there's plenty of opportunity to do what we wanna do in the power space.
Got it. I was just gonna bring up that slide where you showed $13 billion opportunity.
That's right. That 13 is just target markets. It's not the entire thing, the entire market, but rather the five big markets that we're targeting.
Got it. Okay, that's very clear. Maybe, operator, are there any questions at this stage?
We currently have no further questions.
Okay. In that case, we're almost at the end of time. Let's end it here. Firstly, you know, thank you very much, Gene, for making the time. That was very informative. All the best in growing Navitas. I'm sure we'll reconnect again. A big thank you to our clients as well for your patronage. We'll end the call here. Yeah, thanks everyone, and have a good day.
Thank you, Adi. Thanks, everyone.
Thank you all for joining today's webinar. You may now disconnect.