Welcome everyone to welcome and good afternoon. Here today I've got Russell, CEO of Tower Semiconductor. Russell, congratulations on your most recent quarter-
Thank you.
And, thank you for being here.
Pleasure.
Maybe, so what we'll do is, I have a bunch of prepared questions that we can go through, and maybe we'll leave, like, the last 10, 15 minutes for questions from the audience as you... We'll open it up for questions towards the end there, if that works.
Sure.
Maybe we can kick things off with, you know, would you like to just give us a view around how you view Tower Semi's place in the broader foundry industry ecosystem? And can you talk a little bit about differentiation? Maybe you could elaborate on that a little bit as well.
Thank you. Would love to. So I think Tower is relatively unique in the foundry landscape. We are focused solely on analog. I would say that, very close to 100% of our revenue is analog-derived. And what that means is, rather than chasing cutting-edge or slightly lagging-edge CMOS, our differentiation really deals in our engineering capabilities on the analog space, and coming up with solutions that are very unique for the platforms we're serving and for the customers that we serve with those platforms. What that drives is then, very much state-of-the-art unique capabilities within the silicon germanium optical that we do, within the RF SOI, within the silicon photonics, within power management, and within our imagers. And the result is, most everything, if not every product that's manufactured at Tower, is sole source to the customer.
That sole source nature drives extreme stickiness and also drives a relationship that, by definition, must be a relationship of trust, in that if the customer is giving you something that, by definition of the output parameters, cannot easily be made elsewhere, they have to trust you in the quality of what you're making, and the performance of what you're making, and in the stability that you'll be able to make it for them long term. That relationship as well drives a very unique multi-generation roadmap alignment. So I believe that that comes into the core of what we do and what differentiates us.
It also drives a very strong return within the company because rather than, again, chasing CMOS and spending, depending on what technology nodes one's going after, billions of dollars for the next node versus, hundreds of millions of dollars for trailing nodes, our model on capital is predominantly a reuse model, with our resources going into engineering. Several years back, when I was at Applied Materials, and, if you could believe, when I was a younger engineer, I met with the chairman of Applied, Jim Morgan was his name, and I had a question for him.
I said: "How do you balance between being so small picture that, and in the details, that you piss off everyone working for you, versus being so big picture that at the end of the day, you think you've done nothing?" And he wound up giving me a three-hour lecture on leadership. It was an incredible lecture. To his credit, the next morning, his assistant called me and said, "Jim Morgan is giving a presentation today outside the company at 3:00 o'clock. He'd like you to take all your notes and make a PowerPoint for him." So he definitely had an ulterior motive for what he gave. But regardless, one thing that he said to me was very, very interesting, and it is, I believe, one of the key differentiators that Tower has. He stated, "Don't ever spend your time on your poor performers.
Don't try to make them better." Okay, now, he said something here I disagree with. He said, "If they hurt you, fire them. If not, save them for the next layoff." I don't think you should ever have a bank of people for the next layoff. But maybe at the time, Applied did that because they did a peanut butter layoff every time there was a layoff. But the rest of what he said was really to the point and amazing. He said, "Don't spend your time on average performers. If you do, you'll just have an average group. Don't even spend your time on your good performers. If you do, you just have a good group. Identify stars. Determine how many stars you can mentor and work with, and if you do that, you'll have a group of stars." And it's really true.
At Tower, and this is something that I had learned from Rick Hill when I had worked for him at Novellus, an extremely powerful leader, one of the first leaders I ever had that just imbued energy to an extreme amount. But he would, on a yearly basis, and this is something that we do at Tower, he would meet with his management team, with the executive staff and they would make a list of the top 20 high-potential people in the company and the top 20 high-impact people in the company, being very strong in the definitions of both. And those that would be on the list of both the high impact and the high potential, sky was the limit. They were given an incredible amount of opportunities.
It's something that we do at Tower, and it's not just a single meeting, it's multiple, multiple meetings to get this done. And it's an interesting thing, even with the management of the company, the definition of high impact, the understanding that being in a high-impact job doesn't necessarily mean the person is high impact, the understanding of what it means to be high potential. So we do this on a yearly basis, and we really insist that there's a certain amount of people within this high-impact, high-potential group that are young employees, less than five years in the company. If we focus on our employees and get those that are extremely high potential and high impact, and you put those in customer-facing positions, they excite the customers. And that is a differentiation that is insurmountable.
When you have the best of the best, those that are inspired, capable, and making a difference, working with your key customers, amazing things happen. It's something that I have enjoyed for the most part of my career, and that is the ability to interact with extremely intelligent people, many of which of extremely high character, and that is inspiring.
When you put those people in front of customers, the customers become inspired, and they wish to work with you, and you have then multi-generation roadmap alignment, and you're there with them for the long haul. So that really is our differentiation, having the best of the best, working on extremely relevant technologies that the customer trusts, and they trust to put their future in our hands. And I cannot understate how important that is. Or I'm sorry, I cannot overstate how important that is. Thank you.
No, that was great and definitely true from an organizational perspective. Maybe switching gears a little bit. I'm sure you've heard this through the course of your day in investor meetings. There's been a significant investor focus on AI over the last 12-24 months. Could you talk a little bit about how, about the implications of that on your business, and essentially, how are you addressing it?
Okay, so within the needs for AI, there's certainly very, very high-speed data transfer. Within the data center, we've served that area for a long time within the space of optical transceivers by having silicon germanium-based CDRs, TIAs, drivers. CDRs, over time, were replaced more or less by DSP, which is now somewhat being replaced with LPO to where the DSP is taken out. So we've had a very good position with silicon germanium, with optical transceivers, and that has given us a very strong position within infrastructure on this ecosystem with direct customers and the end hyperscalers themselves. Based off of those, of that space and those capabilities, we got very involved in silicon photonics. The acceleration of AI has also driven a very, very big acceleration of silicon photonics.
I think if you were to look at an analyst report from two years ago, they would all be stating that for 2024, 100G would be probably 90% of the market and those technologies below it. I believe that probably 800G is at least 50% of the market in 2024. A lot of that is being driven and being accelerated with silicon photonics. So within silicon photonics now, we have press released with InnoLight, we've press released with Coherent, both having given us technology achievement awards. In the case of InnoLight, talking about long-term strategic partnership, and InnoLight and Coherent being the number one and number two optical transceiver integrators in the world. In addition to InnoLight and Coherent, we serve in silicon photonics right now, six of the top 10 integrators with active silicon photonics programs.
So our capabilities there are very, very strong, and this is where we now have two different platforms: a passive platform, where we have a germanium modulator put onto the SiPho PIC, and as well, where we have a lot of silicon passives within that same PIC itself. So you're dealing then by having capabilities that are offsetting for older technologies, where you would have external discretes being made with indium phosphide that were extremely expensive. So now, if you're going to a four by 200 for an 800 G, and you have a modulator and an emitter that are both intrinsic in the monolithic SiPho PIC versus having these discrete indium phosphide components, you have a tremendous cost benefit and a huge benefit in form factor.
We had released last quarter that 5% of our revenue right now is silicon photonics itself, and growing quite rapidly. We see the movement to 1.6 T being very real, that going initially to an eight by 200, which will only drive a greater need for replacing these discretes with the passive components in the SiPho itself. As well as we have a program with OpenLight, to where we integrate the indium phosphide laser itself onto the PIC. This was an activity that we started years back with a company by the name of Aurrion, that was purchased by Juniper, and then most recently, the majority owner is Synopsys.
The chairman of the company is Mario Paniccia, an extremely well-known person in the area of silicon photonics, having been an Intel fellow, and the person I think that initiated all of the silicon photonics work at Intel. So, an incredible platform program, many things going on there, both being targeted very strongly in data center for AI, as well as for LiDAR and other various applications. So, I believe with AI, we have a very strong position, both by serving it with the components that we've served in the past, by accelerating our Ft/Fmax roadmap with silicon germanium, and as well by the adoption of a silicon photonics platform.
T o where we right now are using a germanium modulator with activities going to more advanced modulators, such as lithium niobate and barium titanate, and active programs with the leaders on those developments. Believe we're in a very good position now, have over 50 active customers, the majority of which with active silicon programs, and look forward to seeing this adoption rate only increase over time and our position and importance within that sector continue to grow.
Thanks. Yeah, so it sounds like, yeah, based on your last earnings call as well, you'd mentioned, you know, you've got over 50 silicon photonics customers with partnerships with InnoLight and, and Marvell. I think you mentioned Coherent right now as well. And, and-
Yeah, Marvell, we've had a long-term partnership with-
Yeah
... and that dealt with Inphi-
Yeah
... who I think was the very first, person into the market with SiPho.
Yeah.
In the case of, of that application, it's predominantly on long haul right now, but with a direction to bring it into the data center as well.
Yeah. Sounds like a lot of, like, future growth coming from silicon photonics. Is that, would that be an accurate statement to make based on your.?
Extremely accurate.
Yeah. Extreme, yeah. Are there other verticals and markets that you're focused on, where you're potentially, where you're looking to gain share?
So one of the big areas driving our growth right now, what we had guided when we went into the year or set as a target, and what we reiterated with our Q1 release, is that we see notable quarter-over-quarter growth throughout the year. I'm not sure that there's so many semis that are seeing that. An area of very big tailwind for us is that of RF-SOI with mobile platforms for front-end module. If you look at that, it's probably a derivative of two things.
Firstly, you have a continued adoption of 5G with the increased content of 5G, although the background market itself still doesn't appear to be overly strong. So one would have to assume that we're growing market share with new customers that we've brought on, as well as that our existing customers themselves are growing market share. The vertical that we see there that's very important for us is that we are running presently at 100% capacity, 100% utilization for the RF SOI.
I think as a result of good planning, but maybe also a bit fortuitous, we entered into a partnership several years back with ST, to where we took on a portion of their factory in Agrate, Italy. That is in final stages of qualification and should be ramping in RF SOI in Q3 and Q4. So the growth that we see and the demand that now is at the limit of what we can ship, will continue to be able to be met by capacity that we're bringing online in Agrate, and under a model that I think is a very, very interesting model.
It's a greenfield-type model, if you will, in that it's a new facility, but it's a model that for both Tower and ST is a very interesting one, in that it would be very hard to have a greenfield activity, where with the first three, four quarters of a production ramp, you would expect it to be net profit accretive. But here, because a lot of the cost and the utilization levels are shared between two companies, we believe that come the second quarter of 2025, according to what we see on forecasts and the volume of what we'll be shipping then, will be EPS accretive.
And so it's a model that twofold: number one, allows us to grow in revenue, in meeting customers' needs in this market that remains very strong, at least for us, and, is a content-enriching market, as well as the fact that, the model of capacity that we've grown is a very smart model that allows a greenfield-type activity to certainly have much better financials than a greenfield-type activity.
Yeah.
But that's only one vertical. We're also seeing extreme benefit and growth in our activities in power management.
Right.
So we were very close to being acquired a few years ago by Intel. In February of 2022, we signed a DA with them. There were a series of customers who were very interested in our 65-nanometer BCD power platform, but our capacity at the time was somewhat restricted. Under the auspices of the Intel manufacturing footprint, we were able to work with them, and a specific end customer that many of them were beating was very happy to be working with us on that platform, that was an advanced platform, due to the fact of now having the capacity to meet their long-term needs. At the onset of this deal with Intel, we worked to have an arm's length capacity agreement, capacity corridor, ultimately in the Albuquerque, New Mexico, facility of Intel. We signed that deal.
It was done. The acquisition did not go through. We never ultimately got Chinese regulatory approval, but the capacity is there, and it's a very high-end capability in 65 nanometer BCD that right now is having customer prototypes starting to go through Q3, Q4, and we should be having our first production ramp in the second half of 2025. And that deal that we did with Intel is a capacity corridor of 22,000 wafer per month. That has a potential to go into other deals, but that deal itself gives us a very strong capacity.
So when you talk about the verticals, if we look at what we're doing with the RFSOI in Agrate, what we're doing with the power management in Albuquerque, as well as what we're doing with silicon photonics, both in 200- and 300-millimeter, and the silicon germanium in both 200- and 300-millimeter, we have right now a capacity footprint that will allow $1 billion of growth, and more or less, according to our long-term model, a doubling of our net profit to $500 million off of capacity that has already been established or is being established but has already been agreed to.
Yeah. Any impact from the recent earthquakes on capacity and utilization, specifically in Japan?
Yeah. So the earthquake was an incredible New Year's present. It happened exactly on the first. Fortunately, and the foremost, we did not suffer any harm to human life, which was a very, very big thing. We had two factories that are fairly close to the epicenter of the earthquake, and also, due to very good Japanese-based building regulations, neither factory suffered any material damage. There was damage in both factories, in particular in quartz, in the quartz furnaces, and it took a little bit of time to get that back up and running. So on a utilization standpoint, both factories suffered somewhat in utilization in the first quarter, and in particular, in the first three weeks of January.
By the fourth week of January, beginning of February, both factories were up and running as if there had been no impact, other than the fact that the work in progress became nonlinear, so it took a period of time to get the factories running as per the amount of moves that they would normally do. As stated, the 300-millimeter RF-SOI is running at full demand, so for the entire quarter, the utilization of the 300-millimeter factory, given the 3+ weeks of lower utilization or no utilization for several weeks, was at 75%. Our model is 85%. The Tonami 8-inch factory also had an impact on utilization but is ramping right now very strong in utilization. From the earthquake, a short answer, there is no impact whatsoever at this point. There was some small impact-
Beginning of the year.
... during-
Yeah
... January that I think has been taken up at this point.
Great. Outside of earthquakes, what do you see as some of the biggest threats to your business today?
I don't know. I think threat is something maybe overstated. The biggest impetus of any business is to make sure that you keep your key employees excited and motivated, that you have an organization that rewards excellence... and within that, that the best people are given the best tools and the support to be successful. I think as long as you maintain an environment that promotes and inspires excellence, there's not many competitive threats that really can hurt you. If we were to look at specific businesses, we have a fairly good discrete business that we've had for many years. This is a business that it's not a foundry flow that we ourselves own. It's a customer flow that we transferred into the company, and we manufacture for those specific customers.
Our press-released relationships are, Infineon and Vishay, and there's others as well. That is probably one area that I would say, we do not differentiate on directly 'cause it's not our technology. We, in many instances, did co-developments in our factories, for next generations, which was a good thing for us. But, probably in the area of discretes, which are, very low layer count, power devices, as far as a geographic threat, probably China would be the biggest threat for that because of the ability for them to not necessarily be lower cost, but have a much lower cost margin model than we would want to be having here, and, maybe to have government subsidies that are more than 15%.
Fair. Fair. I did promise I was gonna leave some time for questions from the audience. Maybe I'll open it up and see if there are any at this point in time. Otherwise, I have a couple more that I can run through. Okay. Doesn't sound like there are a ton of questions. So maybe it'll be helpful to kinda hear your thoughts on essentially... We talked about AI a little bit, and we talked about how AI is kind of has the potential and is currently driving growth, right, across the industry. How do you see-- What are your thoughts on, like, the current state and the future state of the industry, and how do you see the next few years unfolding?
So I think in many segments, the overall semi industry is still a bit weak. Certainly in power, it's a bit weak. If you were to look at the, you know, real blue-chip U.S. IDMs, I think none of them are giving a very rosy story about revenue from power management. But that will rebound, you know, with anything within this industry, especially something of power that, you know, there's hard to find a device application that doesn't require some degree of power management, be it from discretes or be it from PMICs itself. So that being, I think right now, the weakest segment, it will certainly recover. As we go forward and look at everything that's being driven within AI, and, you know, you started with AI, I don't think that that can be discounted right now.
From a social moral standpoint, it's a very, very different story if it's the right direction for a country to take or not. But whether it is or it isn't, the train has well left the station, so it is there. Within all of the speed that's required with AI, both if you're looking at edge devices, which ultimately a lot of AI will be done on, you have to look at battery consumption and hence battery conservation. That's all power management.
If you look at what's happening in data centers, the burden on a power grid of a nation already is so taxed by data center to be able to have very good power management within that and continuing a power management benefit on reductions, it becomes something that will continue to be very, very critical. So although I think many people, and not incorrectly, focus on GPU development and, you know, 3 nanometer, 2 nanometer, sub-2 nanometer digital roadmap, and you look at, you know, sub-5 nanometer just for the switches, what's going on within the data center, all of that's critical. But the role of analog within the space becomes even more critical. And, you know, power consumption, the ability to keep it reduced so that energy stays under control.
If you look at electrification of cars, power management there becomes unbelievably critical. You know, the battery management and the ability to continue to benefit in that direction. So the power management roadmap will stay very, very real for a long period of time. The need for advancement is definitely there. If you look at high speed, what we do with everything with optical transceivers, I believe that it is the most critical thing, if anything, you're dealing with AI. And as stated, again, the silicon photonics for both cost and form factor becomes I think intrinsic to the success of AI. So over the next years, you know, semiconductor will remain unbelievably strong, must be strong.
At whatever specific day, power management or this business or that business rebounds, may be less significant than the fact that it must rebound. Data center itself must become stronger. The ability of doing a lot of AI with edge computing becomes needed, essential, will happen. So, you know, that's the relevance that I see for an analog company such as Tower only accelerates quite, quite drastically. For us, the important thing is to make sure that we stay relevant on customer relationships, driving next generation roadmaps, understanding specifically from lead customers what the figures of merits are that they need, driving to make it done, and doing very smart deals that allow investors to enjoy our capacity growth. What we did in Agrate, what we did with Intel, I think were very smart deals for investors. There's other activities happening now.
There are certain countries that are very desirous to get involved in semiconductor, that are offering very, very large incentives. You know, and I'm, I'm not saying that we would or we wouldn't be involved here, but if you look at India right now, between the central government and the state government, they're offering 75% incentive. That's a, a pretty amazing thing. If you're going to go ahead and build an 80,000 wafer per month analog factory for a cost of $10 billion, to be able to start off with a $7.5 billion grant, probably a reasonable place to start. So, you know, there's, there's many things going on in the area right now that, we need to focus on to make sure that we stay not just relevant, but become more relevant as time goes on.
As stated, we have right now at hand, $1 billion of revenue that we can be using against a $1.5 billion run rate. So we have a good 40% growth that we can have within the next three to four years. But to make sure that we have additional capacity that will be online after that, that takes activities within the next couple of years to make sure that that's at hand. And, you know, capacity that's filled with very advanced technologies, that's, you know, I, I think, the name of success in what we do. So if that answered your question, I-
It did. It did, and, and more. Appreciate-
Too much more or?
No, it was great. It was fantastic. No, those were all the questions I had, Russell, for you today. I really appreciate you taking the time to be here, and appreciate everyone for sitting in on the session as well. And if you have any other questions, I'm sure you can find Russell on his way out. But appreciate it. Thank you so much.
Thank you. Appreciate it.