Morning, and thank you for joining Rosenblatt Securities' fifth annual Age of AI Scaling Technology Conference. My name is Kevin Cassidy. I'm one of the semiconductor analysts at Rosenblatt, and it's my sincere pleasure to introduce Gene Sheridan. Gene is the CEO and co-founder of Navitas. Navitas is a pure-play wide bandgap semiconductor company having both gallium nitride and silicon carbide technologies in volume production. Navitas GaN products have revolutionized the mobile charger market with higher power density, increased power efficiency, and smaller footprints than that's compared to silicon-based chargers. Now they've made significant inroads into the solar, automotive, and home appliance markets. This being an AI conference, we're going to focus on the much-needed data center that needs increased power density and power efficiency.
We now see Navitas is taking the combination of their gallium nitride and silicon carbide products to revolutionize the data center power supply market. The Navitas shares are up 125% year-to-date. Could be up a lot more even again today, but it's had a great run. The power semiconductor market is huge. It's in billions of dollars and plenty of room for Navitas to grow. I'll ask Gene to start off with an overview of Navitas, and then we'll get into some detailed questions around data centers. If you want to ask questions, you can click on the quote bubble that's in your graphic on your screen. That's on the upper right-hand corner, and I'll read the questions to Gene. With that, Gene, thanks for attending, and I'll hand it over to you to give an overview.
Awesome. Thank you, Kevin. Great to join you here for the big event and obviously very timely as we're hyper-focused on AI data centers. We started the company Navitas a little over 10 years ago, focused on gallium nitride. Nobody had adopted gallium nitride in any mainstream markets for power electronics. Despite many attempts in the prior years, we zeroed in on two things: solving the remaining problem of how to drive and control this really fast and efficient transistor. We did that by inventing the first GaN IC that's integrating the drive, control, sensing, and other functions into that chip to make it easier to harness the full value and get the full potential out of GaN. Secondly, we did it by focusing on mobile chargers, fast charging your phone, your tablet, your laptop.
That market, just a few years later, with those two decisions and inventions turned into the first mainstream adoption of GaN. We did our IPO, raising $300 million with the intent to bring GaN to higher power applications like data centers. That has worked well. We spent time investing into higher power versions of that GaN, which is now GaNSafe for high power, high reliability applications, and also bidirectional GaN, which we'll come back to. We also used that money to acquire fantastic silicon carbide technology, which is a great complement to GaN. They're actually both very similar compared to silicon. As you alluded to, they can make power systems far more efficient, higher power density, and smaller footprint. Gallium nitride doing that for what we call kind of medium voltage, medium power applications. Silicon carbide doing that perfectly for higher voltage, higher power.
You put them together, you actually need both for the future of AI data centers. Now we've turned our attention to AI data centers in the last two years. It's the number one priority for the company, and we're off to a great start, but it's early days in really building the next generation AI data center.
Great. Thanks, Gene. Yeah, you know, just last year at this conference, you announced a 4.5 kW power supply, and that was an industry-leading amount of density. But recently, you just announced a 12, maybe 12.5 kilowatt power supply, you know, and that's targeted at hyperscale AI data centers. You know, that's a 166% increase year over year, which is much more than Moore's Law would suggest. You know, can you describe how you're accomplishing this, and is there more to go, and what do you, you know, what's happening, and how does that compare to traditional power supplies?
Yeah, traditionally, they were typically with silicon around an average of 1,000 watt per power supply. You take these and you slide them in, six of them, across a tray, and you might have 40 trays, a bunch of them dedicated to power. While the form factors evolve over time, we've gone in 24 months from what was silicon at 1,000 watts, we've done 2.7, 3.5, 4.5, as you mentioned, 8.5, and now most recently 12 kilowatt. That's a massive change in 24 months. We're very excited about it, but it's a lot of change, and it's a combination of pushing gallium nitride, pushing silicon carbide. We actually use silicon carbide on the first stage called PFC, gallium nitride on that second stage to create the entire power system. We're even changing architectures.
IntelliWeave® is a new architecture for power factor correction on that first stage that makes it even better. You're doing a lot of things, pushing frequency up, changing architecture or topology, as we call it, and pushing the limits of gallium nitride and silicon carbide. All of that has come to bear to deliver the best we've ever done now at 12 kilowatts, and that's really where the customers are focused going forward.
I just want to make sure it's clear that this is the exact same footprint that was the 1,000 watt supply. You're just coming in and now giving that rack 10 times more power.
Yeah, there's two different configurations. You have CRPS, which is traditional, where about half of that tray is dedicated to power, and half of it does the processing like the NVIDIA processing. The new trend now is to dedicate the entire tray, called an OCP or ORv3, ORv3 form factor, which dedicates more of the power. Admittedly, we're creating more space or more footprint so we can increase that power. Even with that, we've gone from 1,000 watts to 12,000 watts. We've made more space for power, double the space in some cases, but we've increased the total power by over 10x. As you alluded to earlier, a big improvement in power density.
Yeah, and maybe if you could just elaborate a little more on the role of gallium nitride and silicon carbide, what these technologies are doing to increase higher efficiency also and power density.
Yeah, kind of the key to the power electronics world, and it certainly applies in spades to data center, is speed and efficiency. Efficiency is more obvious. Of course, we want to make the power supply as efficient as possible. As we're converting and delivering that power, we do not want to waste it because wasting that power or inefficiency of a power supply actually translates to heat. We are going to burn up that energy as heat. Now you have got a thermal management problem. Most people know heat is sort of the enemy of data centers. The second most costly element of operating a data center is the cost of thermal management. The number one cost is cost of electricity. Actually, efficiency is helping you on both of those problems: cost of electricity, cost of thermal management. That is just efficiency.
Speed, operating the power supply faster, allows you to deliver more power in smaller size, the power density you talked about. Gallium nitride and silicon carbide both can operate 3, 5, 10 times, even 20 times faster than silicon. We are generally pushing up frequencies from tens of kilohertz to hundreds of kilohertz and moving into the megahertz range.
Yeah, maybe, you know, yesterday at our conference, we had NVIDIA, their Senior Vice President of Networking, but he emphasized how the compute is now the entire data center and that they have to network together all of these GPUs. You know, the new GPUs are using, you know, I think it used to be 300 watts, and then it went to 700 watts, and now their latest is 1,000 watts. What's that doing to the, you know, rack power that's needed and even the power coming into a data center?
Yeah, everything is pretty much proportional. It starts with that GPU. You're right, 100-300 watts was very common for years, and that was going up. It was already a good power challenge. It's always a good opportunity for us, and we were focused on it. We went from an average of 100-300 watts per CPU to Blackwell® is at 1,000 W+. Rubin® will be 2–3 kW, and that's an exponential rise. You put clusters of them, many of them in parallel, and that translates into the rack power. Rack power has lots of these trays. As I said, it's got a lot of GPUs. Rack power with those 100-300 watt CPUs used to be maybe 10,000 watts in total that we had to power from the power supplies. With Blackwell®, that's headed quickly to 100,000 watts.
Most recently, NVIDIA announced a plan to go to a megawatt rack power to be ready for Rubin® and beyond. I mean, it's crazy, crazy numbers, but we love a really good technical challenge. Obviously, it's a great commercial opportunity.
Yeah, and before I want to dig in more on the NVIDIA, and most of my questions are around NVIDIA that are coming up. Yeah, just you had mentioned IntelliWeave®, and maybe if you could go into the advantages of what the IntelliWeave® is and, you know, how does that contribute to the power savings?
Yeah, and it's kind of neat because it goes to show that we're working not just on making that power device better, which makes the whole system more energy efficient and faster. How do we change the control architecture, the algorithm? So we're really kind of getting in the software business. It's a first step in that direction. And IntelliWeave® is a software or a control algorithm for the first stage of the AC-to-DC converter. I mentioned it earlier, PFC, power factor correction. Traditionally, that operates at a slow frequency. Even with GaN or silicon carbide, it's hard to push that frequency higher, and its energy efficiency is somewhat limited, again, even with GaN or silicon carbide. And IntelliWeave® unlocks more of that potential.
By changing the switching algorithm or control algorithm, we can use gallium nitride or silicon carbide to operate in what we call a soft switching mode, which allows that frequency to go way, way up, kind of unlock that full potential of GaN or silicon carbide on the PFC, and it can save up to 30% of energy savings. Whatever we did to go from silicon to GaN or SiC without IntelliWeave®, we're pushing that at 30% more energy efficiency. That's a control algorithm we basically offer to our customers for free when they design with our chips on that first stage.
Great. Yeah, and you know, a couple of weeks ago, investors took notice of Navitas's data center progress with NVIDIA announcing partnering with Navitas, you know, among other power semiconductor players. Can you still walk through with us what this new power structure looks like for the data centers that NVIDIA is defining?
Yeah, it gets a little technical, but the most important thing to know is Ohm's Law. There's Moore's Law, and in power, we have Ohm's Law. That's power equals current times voltage. If you want to deliver a lot of power and the voltage is kept low, it means you have to deliver a lot of current, right? Current times voltage to make that big power. The power is going way up. The question is, how do we deliver it? Traditional data centers using silicon distributed that power at a pretty low voltage, 12 volts. That means a lot of current. Now, if you're 100, 300 watts per processor, it's not terrible. 1,000 watts a processor, 3,000 watts a processor, this is a problem. We had been working on 48-volt data centers. The industry had. That was in the works. AI came along.
We moved very quickly to 48 volts. The other cool thing about it is that the voltage goes up four times. The current will go down four times to deliver the same power, but the power distribution losses related to that current are proportional to the square of the current. Instead of being four times less current, it is 16 times less distribution losses. There is a lot of leverage in pushing that voltage up to get the current down. Now, what did NVIDIA announce? They said 48 volts is a good step. It is not enough. We got to go big, really big. They are going to 800-volt data centers. 800 volts is almost 20 times higher voltage, almost 400 times less current. Really, really big advantages and benefits are coming from that step up. What does it mean to power chips?
It changes a lot of the architecture. Now you need really high voltage silicon carbide, high voltage GaN, and you still need low voltage GaN to ultimately deliver that power to a low voltage GPU. We love it because it actually played perfectly into the strength of our silicon carbide and our GaN, allowing us to address the entire power delivery really efficiently, really well.
Just to be clear, you can't just bring in new voltage to an old data center, right? This would have to be brand new data centers bringing in new power supplies, new everything.
That's right. This is a big change. It's not quick. We've been very clear. There's a lot of development to be done here. It's still early days, prototyping, hardware testing, refining. We're hooking up to the grid in a completely different way. Traditional data centers hook up to the grid in a traditional way, 110 or 220 volts AC. If you're going to then bump that back up to 800 volts, that's very inefficient. The grid actually starts at tens of thousands of volts. Why would you take it all the way down to 110 or 220 just to then rise it back up to this 800 volts? That's wasteful. One of the key dimensions of the 800-volt data center, as NVIDIA described it, is hooking up to the grid at really high voltages.
Get deep into that grid, start at 10,000, 13.8 is the specific voltage, step that down to 800 volts. We're not going to do it with a traditional low-frequency transformer that has no semiconductor content. We're going to do it with something called solid-state transformers, SST, which are semiconductor-based switching, basically like a power supply, to make that step down far more efficient, probably more reliable, smaller footprint. That is going to require super high voltage, ultra high voltage silicon carbide, which happens to be one of our great strengths coming from the GeneSiC® acquisition.
That's great. That's even one of the long poles in the tent right now for data centers is the trying to get those transformers just outside to, you know, bring the power in.
That's right. It's a big, I mean, even Musk said he thought this was going to be one of the biggest shortages after the semiconductor shortage, you know, a few years ago. I think, and this is not specific to data centers, it's actually an exciting new trend. Everybody knows we've got to upgrade this grid to connect to solar, to connect to renewable energy, to power cities. Data centers are becoming like cities. It's a great place to start.
Yeah, that's great. Lee, you know, you mentioned all the AI workloads is adding on all this extra power that's needed. How are you shooting ahead of the duck, I guess? You mentioned a higher voltage with silicon carbide. Is your gallium nitride going to increase in the capabilities of power?
Yeah, on silicon carbide, we can already go, we call it ultra high voltage, anything above 2,000 volts. We can go 3,000, 4,000, 5,000, even 6,000 volts. We're the only company in the world that can get to 6,000, 6.5 kV, actually 6,500 kV. We're really ready for that. We're actually working on 10,000 volt silicon carbide. Crazy numbers, but if you hook up to that really high voltage grid and you want to efficiently step it down to 800 volts, whether it's 800 volts to power a data center or 800 volts to fast charge your car, these are common voltages, common architectures, common technologies. Gallium nitride, we're also looking to push up that voltage. It's good at 650 volts. We're looking at 850 volt gallium nitride.
We're also taking gallium nitride to lower voltages because you still need the 48 volt converter and ultimately step that down to a processor that runs at 1 volt. Taking GaN into higher voltages and lower voltages is a clear trend and roadmap item. Taking that silicon carbide, we've already got it at 6.5 kV, 10 kV, and even higher. It is a pretty clear roadmap, but it is a lot to do. It is a lot, it is an exciting engineering challenge for our teams.
Yeah, and that's, you know, as I've talked to investors, you know, a lot of them maybe are used to the digital world where things happen very quickly and, you know, you'd replace DDR4 DRAM with DDR5 and it happens in a year. With power supplies, you know, things move a little slower. I think you've been very public about all the design wins you have, but, you know, it takes a long time for this industry to switch over from silicon-based power supplies to gallium nitride and silicon carbide. You know, what is that? What, you know, can it help investors understand why does it take so long?
Yeah, typical data center development times. We started with mobile chargers. Those are pretty quick, can be six to nine months. That's fast in our world of power electronics and in the world of hardware and electronics in general. Data centers are typically 12-18 months designing existing architecture ones. If you want to throw a whole new architecture into the mix, like an 800 volt data center, you're often looking at 24 months or longer. Even NVIDIA in their announcement for the 800 volt was clear that's really focused on Rubin-class processors, which is really a 2027. That's two years from now. That fits the model, right? We'll be doing prototypes this year, maybe some early production ramps late 2026, but this is a lot of work to do to solve that problem for 2027 and beyond.
Maybe, you know, the way I looked at it too is, you know, of course, it's not till 2027, but you do have designs today and having NVIDIA endorsing the new types of power supplies maybe accelerates some of the qualifications you have already. Have you seen anything related to that or is that just my imagination?
No, it's a good point. We've already started, as you know, on the 48 volt data center is happening today. We started with AC-to-DC converters using high voltage GaN or silicon carbide. That work continues. We have 40 customer projects that are ramping throughout this year and into next year. We've just started sampling low voltage GaN for the 48 volt DC-to-DC converter that will start ramping again in 48 volt data centers next year and continue even with the 800 volt data center into 2027. That work is happening. We're ramping 2026, we're ramping, or sorry, 2025, 2026, but obviously the numbers and the opportunities get bigger and bigger as we get into 2027 with the 800 volt data center.
Yeah, and you know, have you received feedback from, you know, the major cloud providers, you know, like AWS, Google, Microsoft? Are you hearing from them or is it just through the power supply customers?
Yeah, no, we work directly with the data center hyperscalers. We really need to understand just how could work closely with NVIDIA to understand what's that roadmap, what do you need? NVIDIA challenged the whole industry to double power every year. That's a lot of what we're all talking about here and working on. AWS, Google, Microsoft, Azure, there's many, many, many people that we're working with, as I said, 40 customer projects in development. We thought we were pretty well connected with most of this industry, but when the announcement came out, we certainly got a lot of inbound calls from people saying, "Hey, solid-state transformers, let's get going. 800 volt power supplies, let's get going." It's definitely added some credibility and visibility for Navitas to create even more opportunities. We're excited about that.
When you say your design wins, you know, I guess how many power supply manufacturers are there in the market and, you know, how many are you engaged with? I know we had a press release earlier about Great Wall power supply company as a win, but then you've got all the Taiwanese manufacturers, but it's, you know, what's the obstacle of them moving to, you know, the gallium nitride-based designs rather than silicon?
Yeah, there's probably, and you can kind of break it down to these different power conversion steps. The DC-to-DC converter of 48 volt has maybe 10-20 customers that work on that. It's pretty concentrated. It's not like it's 100. So let's say 10-20 broadly serve that market. Then the AC-to-DC converter, another 10-20, some are in common, some are different. Now we got that ultra high voltage solid-state transformer tends to be a different group of customers. That might be 10, 15. So all of those are the guys that we engage with and we're doing pretty well, but, you know, they're all at different stages of development, product concept, feasibility. Obviously, there's different stages to their maturity.
Yeah, yeah. You know, there's other consortiums, like in particular, we attend the Open Compute Project where they define the next rack of compute and get the whole industry behind it. Are you working with consortiums like that, the Open Compute Project?
Yeah, we follow the standard very closely because OCP is the one that advocated for dedicating entire shelves, expanding that form factor so we could push that power level up as we talked about before. We have to really track carefully how the form factors and interoperability or configuration specs are changing and evolving with standards like OCP and then talk directly to the hyperscalers to know what's that future processor roadmap so we know what power requirements are coming over the next few years and kind of piece that all together, then directly back to the power supply companies, DC-to-DC, AC-to-DC and greater solid-state transformers. It's kind of a complicated ecosystem, but that's the world we live in and how we try to triangulate and make sure we've got the right roadmap at the right time.
Great. You know, maybe just one other thing, with all this power savings that gallium nitride can deliver and silicon carbide, you know, governments are worrying, you know, have a lot of green programs, and, you know, are you getting any pull, any carbon footprint type of advantages by bringing your products out?
Direct, you know, direct energy efficiency is going to directly translate to CO2 savings. NVIDIA announced at least a 5% savings in their announcement about 800 volt data centers. We believe if you zoom out, overall power efficiency of an entire data center when it was silicon and 12 volts was maybe 70-something percent. But GaN and silicon carbide and a 48 volt data center were pushing that well into the 80s overall. That means 30% of the energy going in never got to do its work, never got to the GPU. It got burned up as heat and wasted CO2, contributing to the CO2 problem. Ultimately, the 800 volt data center, gallium nitride, ultra high voltage silicon carbide, low voltage GaN, all of those will push us well into the 90s.
Those are big steps forward, but these are multi-year efforts as we talked about earlier to get there.
Now, the key question I get from investors is, you know, what can you put as the TAM? You know, I know in your original presentation you said how big the power management market is, but, you know, with this change of going to even higher voltage in data centers, do you have a stab at what that does to the total available market?
Yeah, we're actually rolling that up now. Because it's so new, there's still a lot of moving parts. So we're kind of doing some thorough third-party estimates, understanding for each block what we could expect. Roughly, if you wanted to look at it, it's about $5-$10 per power conversion stage, or $5 million-$10 million per gigawatt of power delivered for each power conversion stage. And there's up to five power conversion stages that we're working on. You know, what does that translate to in the overall world? You know, it's certainly hundreds of millions of dollars headed to a billion dollars over the coming years. And we're really scrubbing that now and we're probably rolling out a more formal and thoughtful TAM estimate in coming weeks.
Okay, great. Be interested to see that. And, you know, as I said, when NVIDIA made the announcement, they of course named, you know, Monolithic Power, Texas Instruments, Infineon, and, you know, maybe focus on Infineon. I think of them as your cousins because of their acquisition of International Rectifier, which is your heritage. And, but you have agreements with them where you have second sourcing. Can you talk about what's important about second sourcing and why Infineon?
Yeah, we saw this even in mobile chargers. Ultimately, the technology is exciting and you can definitely get a lot of sole source business on the high end of probably any market. Once you want to go for the mainstream, once people want to jump in with two feet and bet their entire business on it with their mainstream volumes going to new technology, you know, they naturally get a little nervous about being sole source. Two sources becomes really valuable. I'd rather get, you know, 50% of a really big number than 100% of being sort of confined to that high end of any market. We saw it in mobile chargers, which is where we started thinking, well, maybe we need a partner, maybe we need a second source, at least for select cases.
You know, there's leaders in solar that are going to go really big in GaN in the next 12 months. There's major EV players that are going to go really big with GaN. And certainly it's true here with data centers and it's true with NVIDIA. The data center guys want to go big with GaN and silicon carbide. We could really benefit from having a second source, at least in select cases. Infineon turned out to be a really good option for us. They have the range of GaN. They also have a range of silicon carbide. They're not quite matching our generational position, but capable and competent. You put it all together, it made good sense for us.
I think it helped us a lot to solidify our position in collaboration with NVIDIA where they're looking for two solid sources across a lot of these applications.
Great. Yeah. You know, there are plenty of other competitors and some still sticking with silicon, saying you can squeeze more juice out of that lemon. There might be some new guys playing, coming into the gallium nitride. One of the questions from the audience is, how do you differentiate from Power Integrations with their solutions for both silicon, but then gallium nitride that they're coming into?
Yeah, yeah, that's right. We're the only ones creating true GaN ICs. And by GaN ICs, I mean monolithic integration of critical functions like the drive function. How do you drive and control a super fast and efficient transistor of gallium nitride is very difficult to do. Trying to do that with external silicon drivers, even co-packaged in the package, is a challenge. That's our unique advantage, whether it's Power Integrations or anyone else. We've also pushed our GaN into high power GaN. I haven't seen that from Power Integrations or most others. Our GaNSafe product line that I talked about or a bidirectional GaN, those are really squarely focused on applications like tens of thousands of watts, even hundreds of thousands of watts as we're seeing with the data center.
I don't see Power Integrations there in the data center, but we see Infineon as a good partner to help us make it happen.
Yeah. I think Power Integrations may have even announced getting out of the mobile charger market. It just got to be too low priced for them. I think if you have a higher end, you know, maybe you can address that, of, you know, where the spread is in mobile chargers, of, you know, there are five watt chargers to 140 watt or maybe higher now. Yeah, where do you see that market?
Yeah, you know, as the power goes up, GaN brings more value. If you want a slow charger, slow speed of charging is proportional to the power. Five, 10, 15, 20 watts, relatively slow chargers. Silicon does the job well. If you want to use GaN, it'll give you a little benefit, but not a lot. You're going to be more price sensitive in those markets. If you really want a fast charger, you really want to be talking 65 watts and higher. The more the power, if you try to do it with silicon, you can, but that adapter is going to get really big. It's not going to fit in your pocket, big and heavy. The term brick is there for a reason, right? You look at a GaN charger, nobody's calling them bricks anymore.
These things are tiny, miniaturized, slide in your pocket, but pack a real punch, a lot of power. As you go up in power, gallium nitride brings a lot of value. Our flavor of gallium nitride with the integrated driver and integrated functions allows us to go up in power and up in frequency and up in efficiency better than anybody else. We will always do better at that leading edge, bleeding edge of fast charging and high power. Of course, that is the trend of the whole industry. It is sort of moving in our direction, which is good.
Yeah. It comes in extremely handy when you're on cross-country flights when you only have one power outlet.
Yeah.
With a GaN charger, you can charge three different devices at the same time.
Yeah. You and I are both road warriors, so we can really appreciate it.
Make your neighbor next to you a little friendlier than stealing his place.
I know. I'm always tempted to just say, you know, here, take one. Just have one for free. I'm sort of a walking GaN charger advertisement. Yeah.
We had one other question from the audience about Tesla news and regarding any future investments with Tesla and maybe even on their, you know, maybe exposure into the onboard charging and even Tesla.
Was there something on the investment? Was that a Tesla announcement on investment or otherwise?
Yeah. They're asking about the Tesla news regarding future investments.
Okay. I'm not sure what the investment is, but certainly we see a lot of, you know, Tesla is sort of the perfect customer for us over time. I don't have anything specific to announce, but obviously they're in solar, they're in energy storage, they're pushing the limits of EVs, they're pushing the limits of fast chargers. All four of those are really interesting and they're creating their own data centers and their own AI processors to power their fleet of self-driving cars. They have it all, they have it all going on. We love Tesla, we love where they're going from a power electronics perspective. It's a nice fit for us. Nothing specific to announce today, but there's a lot of great opportunity there that we're certainly hustling on.
Yeah. One of the things that impressed me when I first met you and met Navitas is even for a small company and, you know, it's a breakthrough technology that not a lot of engineers coming out of college knew how to use. You came out to the market and started putting in design centers. Can you talk about the expansion of your design centers or what those centers are bringing to the market?
Yeah. We found early on you can't just make a better transistor. If you make a better transistor, the rest of the system has to get redesigned. This is an industry, like you said in the beginning, pretty conservative. They've been designing power supplies with silicon the same way for 30, 40 years. Low frequency, that means low frequency controllers, magnetics, capacitors, EMI filters, and transistors and silicon. Everything's changing. If you just make a better transistor, you're going to be waiting for the entire ecosystem to develop. Where are the high frequency controllers? Where's the high frequency magnetics? Where's the training for the power supply engineers to design at this higher frequency? We knew we had to take a more proactive effort.
In mobile chargers, we have a team that can really design the entire mobile charger or the guts of it anyway, all that power electronics for the customer. Frankly, that not only speeds it up for the customer and we worked on high frequency controllers, now we do our own high frequency magnetics. We're very good at that. High frequency EMI, we're very good at that. It teaches us how to make the GaN better for the next generation. What should we integrate? How should we tune it? How does it need to evolve? Higher voltage, higher power, whatever that is. It's kind of a win-win. For a small company, as you alluded to, we can't do it all. We have to be really selective. Power is a huge market, which is exciting, but that's also the problem.
A small company tries to do it all, we're going to be overwhelmed and we're already overwhelmed. We are really pivoting now. We'll keep going on mobile chargers. EV is a good longer-term play, but it takes even more time, obviously longer design cycles. AI data centers now are number one priority for all the reasons we talked about. We've got great system design capability in that space. We are going to grow that even faster to cover each of these areas we talked about earlier.
Yeah. Yeah. That's where, you know, I think again, the NVIDIA announcement probably helped pull some of the data centers into the upgrades right now. Maybe as we talk just a little bit about the other markets that you had, industrial, you know, appliances, just lots of home appliances that have inefficient power supplies. Can you talk just a little bit about that, of how that market is developing?
It is. Yeah. It's not another, it's a pretty conservative one, as you said. The development times there are even longer. Two years is typical. Home appliances all are consuming more and more power. Powerful motors, compressors, refrigerators, and they all want energy efficiency for the same reasons as everybody else. Upgrading to gallium nitride or silicon carbide on the higher power is a natural next step. A lot of them have big heat sinks, which are trying to manage the thermals. Again, back to thermal management. If I can run more efficiently, you're going to reduce that cost of electricity. You're going to get rid of thermal management, like the cost of manufacturing and placing down large heat sinks.
In many cases, we're shrinking down the size, especially for more home appliances that might be smaller and power density driven, like hair dryers and home appliances like smaller home appliance sizes. That's a nice market. It does take time. It's not the top focus for us because we're so busy with AI data centers. Mobile and EV is sort of our top three focus areas.
Okay. Great. And the, you know, silicon carbide, the conversations have come up a lot with Wolfspeed having some financial issues more than technology issues. What do you see in your supply chain for silicon carbide? Are you seeing prices coming down or availability improving? You know, how does the supply chain look?
Yeah. Our fabless model, or what we really call fab light, because we're very hands-on with bringing up the technology and owning and managing the technology. We use TSMC for gallium nitride and we use X-FAB in Texas actually for silicon carbide. I think that fab light model has actually paid off really nicely for us in what is really a semiconductor downturn. AI data centers is the only real growth sector right now. That fabless model means we don't have the burden of a lot of overhead, a lot of depreciation. Our foundry partners, TSMC and X-FAB, their loading is pretty low right now with most markets pretty soft. They're hungry. We see costs coming down the supply chain, cycle times shortening, and lots of upside capacity to apply that to the growth engine of our industry, which is now AI data centers.
It really has helped us with our fab light model on costs and supply and capacity availability.
Great. I'll poll the audience for any other questions. We're not getting any more from the audience. Is there any other closing remarks you'd like to make, Gene? We're down to five minutes left.
Yeah, sure. Sounds good, Kevin. I think in summary, as I said, the market is huge. Gallium nitride and silicon carbide. Silicon carbide is already a larger market, a multi-billion dollar market. We're a relatively small player, but we've got the leading edge technology, especially in that ultra high voltage. Very excited about upgrading the grid to power data centers and the grid in general with that ultra high voltage technology. Gallium nitride is now getting adopted not only in AI data centers, but also going into mainstream solar in the next 12 months and mainstream EV in the next 12 months. Those are the big waves that we're really pushing now, expanding beyond the mobile charger base where we started. That's where we're focused over the next 12-18 months.
Great. We did get one more question in, and I don't know if you can answer it directly, but you know, what's the microinverter from Enphase, their ninth generation? This was based on Navitas. Can you talk about that design?
Oh yeah. Enphase has been pretty public about their intent to go to gallium nitride with IQ9, which is their next generation. They've been pretty public that they're working with Navitas and others. It's not an exclusive thing, but we're excited about it. What I just alluded to, gallium nitride going mainstream into solar, that's obviously one place we see that happening. They're tracking, as they publicly explained, to launch IQ9 later this year. We'll see that ramp at the very tail end of this year and then ramping nicely into next year.
Okay. Great. Thanks.
Yeah.
I'll give it one more poll. Otherwise, I'll say thank you, everyone. Gene, thank you for your time and getting up early for us.
Of course. Thanks for having me, Kevin. Always great to chat with you. Glad to participate. Thank you.