We're going to get started. We have a very exciting company, SkyWater , overweight-rated stock. I'm Harsh Kumar. I cover semis at Piper, and we have the pleasure of hosting Tom Sonderman, the CEO of SkyWater. I've got a handful of questions, but I did want to keep it fairly interactive, so every now and then, I'll poll you guys to see if anybody's got a question, but Tom, let me ask you about ATS business and the drivers of your ATS business. That's kind of like the engine for your growth. So maybe you could tell me, what is the ATS business? What are you trying to accomplish with that service or that business? T hen how is it a tailwind for your overall company?
Yeah. A great question and great to be here. When we created SkyWater back in 2017, we had inherited a legacy fab that Cypress Semiconductor had been using with technologies that were mature and running in volume, but not a lot of innovation, and so what we did to transform the fab is we created ATS, Advanced Technology Services, and it's essentially technology as a service where we partnered with customers to bring in their innovations, but instead of us kind of funding the technology creation and then recouping that investment when they went to volume, we said, "You as a customer are going to pay for that," and so ATS is a high-margin, 50%+ type engineering services business that we essentially leveraged to have our customers fund our technology roadmap, so now, as you look eight years later, we have a fab full of ATS capabilities.
Starting last year, and it continues to ramp this year, we're now moving those development programs into wafer services, which is our volume business. O f course, we get good margins while we're developing the technologies. T hen we not only own the IP for that development, but as we move to production, we get much more competitive margins because we are the company that created the technology and the sole fabricator. I t's a way to take a fab like we had in Minnesota, which is a lower-scale fab. It requires kind of a high-mix, lower-volume model. O ne of the ways we leveraged this model was with the DoD, the Department of Defense. They like investing upfront for capability. T hey have invested upwards of $300 million in SkyWater to create this kind of foundry model for the defense industrial base.
Okay. Great. So just a follow-up question on that. So a customer comes to you, they would work with you on how to make the chips, how to mass-produce the chips, and maybe some aspects of technology, packaging. Is that a fair game also?
Oh, absolutely. I think part of our movement into advanced packaging, maybe before everyone else started thinking about it, was because our customers don't just make a single chip. They're all putting things in packages today. Heterogeneous integration is really where a lot of the innovation occurs. W hen a customer, if you look at PsiQuantum computing, for example, there's a lot of innovation in creating the quantum processing unit, but also how you use interposer technology, chiplet technology to create a comprehensive package. T he ability for us to offer that total solution, offer it domestically, offer it in a trusted, secure environment, is something for a technology like that where there is just tons of innovation, tons of scrutiny around protecting IP. We have a very unique model that we can offer.
Does the ATS customer put the money for some of the initial tools, or do you pitch in, or is it shared? How does that work?
Yeah. Another nice question. So CapEx-light is our approach and very different from a normal foundry model where the fabricator, the foundry, would do all the upfront investment and then recoup that investment when you went to volume. What we do is we say we want the customer to co-invest not only in the NRE, but also in the equipment.
So if a customer comes in with a new idea, and we're all about customization, so most foundries don't embrace that, we would look at our normal flow and say, "We can do 80% of that, but there's 20% that's going to require some NRE, and we may need another tool." W e would go to the customer and say, "You have to invest in that tool." T hen because we don't offer exclusivity, we would say, "You obviously get to use that tool primarily, but if we have excess capacity on that tool, we could use it for other customers." T he customers tend to like that approach because it makes the tool run more efficiently and faster and higher yields and all that. So it's really a co-investment model.
What it does, especially for a company of SkyWater size, is it puts skin in the game on both sides. It's hard for them to walk away. Once you create a technology, it's very hard to move these technologies, but it forces them to invest in the early stages where there is a lot of risk and whether the technology will reach maturity. T hat's why a lot of foundries shy away from that approach because they end up having to take that risk. N ot all those programs pan out as expected.
Yeah. We started to get into it earlier with quantum computing. Some of these companies, three years ago, we could be talking about small balance sheets. N ow the amount of money and capability that these quantum companies have is quite enormous. Y our name comes up as a company, not yours, but SkyWater's name comes up as a company as a premier fab for a lot of the quantum folks. Maybe you could talk about if that is the case indeed, you're working with several of them and to what level of engagement. It's all cutting-edge stuff. Nobody else sensitive to on top of that?
Yeah. I mean, it's interesting with quantum because Claire's in the room, but 18 months ago, we brought up quantum. People were like, "Oh, that's still way out in the future." W e kept silently working on it with our partners. I think we all see now it's very real. T here is a lot of money going into it. T hat money flows into our ATS business.
So you were looking at levers of how ATS comes together. Quantum is a great example where the fab we have is a CMOS, an advanced 200-millimeter CMOS fab. T here's a lot of similarities with how you create a quantum processing unit versus a microprocessor. T he fact that we have other capabilities like interposers, advanced packaging means we can really offer a comprehensive solution. W hat's happened is we've had some seed customers.
One we'd like to talk about is D-Wave. We've been working with D-Wave for many years. We also talk about Quantum, who isn't publicly traded but just raised $1 billion a couple of days ago, and all that money is coming—not all of it, of course, but a lot of it—comes into our business to create these capabilities. O ver time, we've turned Minnesota into a center of excellence for quantum technology. There's also a focus around super-cooled capabilities.
A lot of the things we do for the DoD, our thermal imaging cameras, these all have to operate in space, and these ultra extreme environments. T hat's very similar with quantum, so we see quantum this year as being our strongest driver. I think we said in the last earnings call, 30% growth we expect to see in our quantum business this year. We see it really being a major growth driver for us in our ATS segment as we exit this year and get into next year.
Yeah. So one more question on that. With quantum, there's the bits, the qubits themselves. T hen all the companies talk about growing the qubits from whether 50 or 100 to like 100,000 to be practical. T hey talk about chiplets. Would you have that kind of packaging capability for chiplets and all of the other technology that goes with it?
Yeah. Absolutely. A gain, think of qubits as now almost same thing as we're creating more transistors, right, and traditional silicon. So there's going to be a race to build more qubits. T hat's literally what we're doing. D epending upon your application, not all qubits are created equal. So we don't have a common nomenclature. So that's one thing that you have to keep in mind. When people talk about qubits, it's all about the functionality on the other side.
The quantum processing unit is very analogous to CMOS processing. T hen when it comes to scaling qubits, instead of just making one chip with a bunch of qubits, now they're focusing on making one chip with a certain amount of qubits and combining with other chips through the chiplet strategy. T hat's where our 2D-to-advanced packaging capability comes into play.
It's really, when you look at any quantum customers, and we have three that we are working with in the public domain too. We talk about there's several more we intend to announce this year. We have a whole quantum development platform that's going to be coming out. A ll this is just accelerating the time to market. H aving the ability for a single customer to go to us and get a comprehensive solution is just going to speed that innovation.
So we see it as an exciting area. Y ou mentioned trust. I want to elaborate on that too. The fact that we're a trusted fab for the DoD and we do trusted work means a lot of these customers have confidence because we compartmentalize everything and treat this community of customers just like we do the DoD community. So we protect that IP.
What about tool revenues? You mentioned earlier tool revenues. What is the part that tool revenues can play in your overall strategy for growth?
Yeah. Yeah. Tools revenue for us is an interesting and sometimes confusing topic. Our customers sometimes will invest in us to where we own the tools, in which case they would flow through kind of a normal investment that we would make in like a normal CapEx investment that they would compensate us through the program. In other situations, especially in DoD programs, they want to retain ownership of those tools. When that happens, it flows to the top line of the business. It appears we have tools revenue. We don't sell tools, but we buy the tools for the customers. What that does for us, though, is think of it as an investment in the future.
So when we talk about $200 million of tools revenue over a three-year period, that's our customers buying tools for our fab that will eventually generate ATS revenue. So think of it as a precursor to ATS. I t's also a precursor sometimes to wafer services, to where in certain cases where the customer is actually saying, "Hey, I need additional capacity."
I think you're seeing all the foundries do this now. O ur model is very much, "We will do the manufacturing for you, but you have to make these investments." T he foundry world has learned that you don't make investments without customer commitment. O ne way to get customer commitment is to have them make the investment.
Once they invest, they're committed. They're like pot-committed. Okay. This was later down on my list, but I'll move it up. The importance of being an analog fab based in the United States with the blessing of the U.S. government and willing to take on not massive projects. Now, we're not talking about large-cap, multi-hundred-million-dollar chips sort of units, but these quantum-like projects. What is the importance, in your opinion, of being a fab that's willing to take on these cutting-edge projects?
Yeah. I mean, it's interesting because you get a lot of focus on extreme node 300-millimeter. W ithin 200-millimeter, there's also a lot of innovation and, like you said, cutting-edge technology, quantum being one. The reason we acquired the fab in Austin from Infineon is foundational semiconductors are used everywhere.
They're kind of the bread and butter in many ways, but also have high degrees of differentiation. T hey're the necessary ingredients that allow the AI chips and some of the more advanced technologies to come together. W hat's happening in the U.S., there's what's called the 232 analysis, which is looking at the entire supply chain. T hat's identifying vulnerabilities and things like foundational semiconductors. S o as a pure-play foundry, and I like to remind everyone, we're the only U.S. investor-owned pure-play foundry in the U.S. today. That puts us in a unique position.
We don't have foreign contamination from an ownership perspective. T here are certain entities that want to make sure that silicon is getting fabricated at a SkyWater or someone like a SkyWater. S o the mixed signal, and it's interesting, that whole space is going through the fabless foundry dynamic as well, including Infineon. Infineon runs their own fabs, and they also use outsourced, if you go down the list, Microchip, ADI, TI, onsemi.
They're all going this whole kind of analog mixed signal hybrid semi-sector is trying to figure out NXP is another one, what fabs they want to own versus the ones they don't. W hat's happening is most of them are deciding, "I have to invest in 300 millimeter." So they're looking at sub-optimized 200-millimeter fabs.
That's where SkyWater can come in as a consolidator of 200-millimeter foundational silicon with the backdrop, for example, 100% tariffs if you make it in the U.S. I f you invest in the U.S., then you don't have to pay the tariff. S o there's an incentive now for some of these hybrid manufacturers to not only move to the foundry model, but do it in a way where they could invest to bring certain capabilities that maybe they were planning on or do get outside the U.S. because there's a cost to transfer.
There's a cost to bring these capabilities in. T he idea is matching things that need to be made in the U.S. with capabilities in the U.S. T oday, there's a lot of 200-millimeter capacity, but it's owned by IBMs. As you migrate that to the foundry model, you can really secure the foundational capability. 40%-50% of all the semiconductors used in every tank, missile, and jet come from Taiwan and China.
Wait, wait, wait. Can you repeat that again?
Yeah. 40%-50% of all the semiconductors that are used in every missile, tank, and jet come from Taiwan and China. I know it's a very sobering statistic, and that has to change.
I didn't know that. That's an astounding fab.
Yeah. T hat's just for the defense industrial base. So you can imagine there's other situations like that. T hat's the driving force behind a lot of this.
So I'm going to poll to see if any of you folks in the room have a question. Please do not feel shy. If you have one, just step in and ask away. Okay. If not, then I wanted to follow up on the chain of revenue.
Oh, there's one thing.
Sorry. Please. Sorry.
So the IDM community in the hybrid space, which again are the analog mixed-signal community, they're all invested in 300-millimeter today. I'll use Infineon as an example. They are building a 300-millimeter fab themselves in Dresden. They're part of the joint venture with TSMC also in Dresden. They have a new 300-millimeter fab in Villach. They're also invested in 200-millimeter silicon carbide, 300-millimeter for GaN, by the way, in Villach. That's where their investment's going.
They still need a lot of 200-millimeter capacity. What they're doing is moving their 200-millimeter capacity to more of the foundry model, investing where they want to differentiate in 300-millimeter. If you look at NXP and other examples, they have three 200-millimeter fabs in the U.S. They said over the next five years, they want to divest themselves because they're also invested in 300-millimeter.
Onsemi has a new 300-millimeter fab, the old IBM fab in East Fishkill. They're trying to ramp that. They're looking at ways. The challenge is you want a lot of these 200-millimeter products still need to exist. I t's a question of, do I want to keep running my 200-millimeter fab where maybe I can't completely keep it full? Or can I leverage the fab for what I need it for and then have someone like a SkyWater run it to bring another business?
So to add to that, that's called flexible manufacturing. M ost of my logic cap guys are going to flex manufacturing because they don't want the headache of a semi-empty fab when things go bad. T hey'd rather have 50% share. I believe there might have been another question.
Do you guys have the balance sheet to get some of these 200-millimeter fabs? Or who's going to grab it? Do you have access to capital at a reasonable rate kind of thing?
That's why we're here. No. Obviously, so when we did the deal, and part of it, again, is buying the right assets at the right price. So we bought the fab in Texas for $73 million. It's over a $300 million asset. S o because of that, we were able to significantly increase our line of credit. W e have a $350 million line of credit. W e were able to do this transaction, again, with our existing lending facility.
As we look at doing, say, other M&A, that would require different scenarios. O ne of the things that is clear is when you're in the situation that we're talking about, the seller has an interest in coming out with a good solution, like Infineon did. T hey're most interested in the supply, making sure they have longevity of supply because their alternative typically is shutting down the fab.
Of course, we're in a down market now. You buy depressed assets in down markets so you can take advantage when the market recovers. Let's just say any additional situations would be evaluated in that context. It's more, again, sticking with our CapEx-like model. We would expect our partners to come in in a way. That's what Infineon did. We said, "Hey, we can't afford to pay the $300 million for this fab. W e can ensure you a supply for a period of time." Then we worked out a price that we could afford. I think when you have a willing seller, then you can find the economics to make it work.
I mean, it's exactly what happened when you guys were formed with the Cypress fab. I believe there was a three-year or five-year supply agreement.
Yep. The same way.
You sell back to the seller the products that are made, and you kind of cowork to find an efficient solution.
Yeah, and in most of these cases, including Infineon, they want to keep the fab open. They just can't keep it full, and so we expect that the fab will maybe have 50% Infineon over time, 50% new stuff that we bring in, but the mix of Infineon may also change because we expect them to want to take products that maybe they're making in other 200-millimeter fabs, move them to the U.S. supply base.
That's one thing that's unique about the window that we're in right now, is there is a desire to make more things in the U.S. Our goal is to capture that momentum, bring as much stuff in so that five years from now, if the pendulum swings and people aren't as concerned, it's just very hard to move silicon technology. So having excess foundry capacity at a time where people are looking for it puts us in a very nice position, at least for the Austin fab. Any subsequent fab would have to have the same value proposition.
I love it. So we can keep it going if you guys have any more questions. If not, then please don't be shy. We've got about four minutes left. So that's your window to ask. I want to follow up on revenue. So you talked about ATS. You talked about tool revenues. Then where the rubber meets the road is the wafer services. Can you talk about how that works and how long does a customer spend years in ATS and then?
Yeah. So the one thing that happened by acquiring the Austin fab, we diversified our revenue to a much more healthy mix, we'll call it, between wafer services and ATS. So obviously, the fab in Austin is all wafer services. It's all Infineon today. W e get to more of a, you call it 60/40 kind of balance between ATS and the volume portion of our business.
So what we will be doing is leveraging that and kind of that runway that we have in Austin as we bring in new business so that we can continue to transition the business in Minnesota. T hat's where anywhere like the ThermaVue program, that was a four-year program. So four years in ATS. T his year, we started moving it to production.
We have others like tape-outs that are now happening on ThermaVue where you tape it out, you run a shuttle, and you could be, say, in a two-year window moving the Wafer Services. So our whole strategy is really to leverage ATS for new innovative platforms that require a lot of R&D but also have a path to production, and we've talked about this year, we expect to generate more revenue in Minnesota from ATS converts than we do from legacy.
Wow.
A t least 50% more. S o that engine is working. T hen in Texas, we have a three-and-a-half-year take-or-pay, which kind of secures the Infineon revenue. W e also have 180 engineers. So we're going to start bringing in ATS programs into Texas. We license high-voltage IP on 130. So we just add that to our design enablement kit. Now we can start doing tape-outs, same process technology, just bringing in new customers. That's an immediate revenue driver.
T hen we also intend to have a 65-nanometer mixed-signal, highly differentiated, high-performance capability in the fab. That will require investment. W e're formulating that. T here is no 65-nanometer capability in the U.S. today. So if you're in a 130 TSMC fab and you now want to move to the U.S., and we can offer you a 65-nanometer solution, that helps pay for that cost of transition. That, to me, is just like on the ThermaVue, we took customers out of power and, on a 150-nanometer, moved them to 90. If you can offer a technology differentiation, then people, they will pay for that transition.
All right. We have about 54 seconds left. I'm going to ask you one final question. Is the U.S. government aware that like 40%-50% of chips and jets and missiles are coming from Taiwan?
Yeah. That's why I'm going to DC tomorrow. No, yeah. W hat's interesting is that when you say a statistic like that, you would hope they would look back and go, "Oh, that's not true. Let me show you." T hey nod their head and go, "Yeah, that's why we're in the problem." T he issue is because a lot of it is customer off the shop, right? So it's hard to trace where it's coming from.
That's the 232 analysis. L et's just say no one's comfortable with that number. I t's going to be a forcing function to kind of what we talked before. We need to look at the CHIPS investment and all that. Where are we going to solve the problem as opposed to just where we're building these random fabs? T hink that's kind of the focus right now.
Tom, thank you so much. Thanks, everybody, for coming in.
Yeah. Thank you.
This was great.
Good.