Okay, good afternoon. I would like to introduce our next presenter, Bob Daigle, Executive Chairman of Amtech Systems. Welcome.
Well, thank you, everyone. Looking around the room, I see some familiar faces. Some of you I may not have met. Hopefully, you know a little bit about Amtech. I am going to try to quickly run through key highlights, and in particular, I am going to talk about where we play in AI and where we want to play in AI. I think that is usually the area people want to focus on. I do want to leave quite a bit of time for Q&A. Please stay tuned. Quick rundown of, I call it Amtech 101. Our annualized revenue is about $90 million, with 80% of it in our Thermal Processing Solutions segment and about 20% in our Semiconductor Fabrication Solutions. The breakdown, think of the Thermal Processing Solutions as really, it is thermal processing. It is furnaces that are used for mostly electronics applications.
Where the big play for us there has been advanced packaging for AI infrastructure. I will give you a little bit more of a view of where we play in a minute here. It is this kind of equipment, and just to put things in perspective of the scale, maybe the width of this room is the length of one of these pieces of equipment. We manufacture this kind of equipment primarily in our Shanghai factory, but we also have a contractor in the Malaysia-Singapore area that serves the U.S. market because of tariffs. The other part of the business is really around its parts services, consumables that are servicing mostly tools in the CMP, chemical mechanical planarization area, for semiconductor production. The big growth area here that I will focus on is really what we are doing in the reflow area for AI.
This is an example of what we call our TrueFlat equipment, and I think it is helpful to understand what this equipment is actually doing by taking a look at, this is unfortunately a little bit small picture, but an NVIDIA GPU. So you have high bandwidth memory, you have the chipsets. Those get assembled typically on an interposer, silicon or in some cases glass interposer. Then you have a high-density printed circuit board they call the chip package that do the interconnects. Where our equipment is used, this high-temperature furnace here, again, big piece of equipment, is melting these ball grid arrays for assembling the packages. We own this space. The majority of GPUs, TPUs, use our equipment. The reason we own this space is it goes back to a little bit of ancient history when CPUs were a big thing, the Intels, the AMDs.
The problems they were running into is when you do the soldering process and you've got these plastic substrates, they tend to want to potato chip, bow and twist when you heat them up. There's stresses. What we developed and patented is called TrueFlat technology. The simplest way to think about that is it's a reverse air hockey table. Instead of blowing up air through a plenum for the hockey puck, we're pulling suction down. The vacuum holds down the substrates as they're going down this long tunnel furnace. We ended up being process of record for all what's in your laptops and what's in your home computers. Then AI shows up, and that's really the inflection point for this business.
Where if you go back about 18 months ago, not that long ago, we were producing about four of these systems per week in our Shanghai factory. Just to give you a perspective, we're semi-fabless, right? We have our own factory, but what we're doing in Shanghai is assemble, test, ship. We have a very robust supply chain. We were shipping about four of these a week. Lead time's six weeks, two weeks build, two weeks test, out the door. Fast-forward to last quarter, producing about 10-12 of these a week, so about threefold number of units. Lead times are still six weeks. We can still build in two, test in two, out the door. We've been able to scale very rapidly with the demands of our customers, and they are the who's who. It's the OSATs you all know.
It's the memory guys you all know. It's all the processor guys you know are using our equipment for this process. That's been a huge secular driver and, again, back to chip on wafer on substrate, we're the on-substrate assembly guys down here with this kind of equipment. A little bit about the history going back a little bit further. I moved to the executive chair role in August. I joined Amtech's board about five and a half years ago. Three years ago I took on the CEO role because the company was in trouble at the time. We owed a bunch of money to the banks. We were burning $3 million a quarter of cash, losing money. I stepped into that role and spent about two and a half years restructuring.
What we did is that's when we migrated to this, I call it semi-fabless, in some cases, fabless model, where we had seven factories building equipment and products. We went down to four facilities. We only have one out of the four that it still builds equipment. That really changed the cost structure for the business. We were able to pay off the debt. Operating cash got us to $26 million in cash. We now have $83 million in cash, no debt, with a raise we did. Basically got ourselves in position last quarter, 50% gross margins, 15% EBITDA margins. We got the company to be very healthy.
I like to tell people, I think it is true, if we and our team sat on our hands for the next couple of years, we would probably take this $90 million a year revenue company and probably just with the tailwinds in the industry, we could be $150. But we have ambitions to do much more, and we have a team and a balance sheet to do it, which is why I transitioned the CEO role to another gentleman in August. He comes, his history, 30 years in the semiconductor industry. His name is Guy Shechter. He worked for Veeco for 20+ years, most recently with YES, which is a chip packaging company. Because our ambitions are really, there are eight different operations that go into making these packages. We participate in one right now. We need to get our TAM up.
If we want to be much, much bigger than $150 million, we need to participate in more process steps, and we have the right to win because we have been a very. We get kudos from all the customers about the ability we were able to scale. We never have had a quality issue, and we did not do anything offensive like double the price of our equipment because the demand was high. So we have been very good to our customers. They need to scale these other layers more cost effectively. Some of the equipment, most of the equipment being used in these other operations are batch tools, very low throughput, very expensive tools.
We are being asked, can we move up, do other things, and maybe we can get three, four times the price we are getting now, but it is not the 10 times they are paying to others for doing those operations. We can handle more volume. So I brought Guy on board because he is in. He knows the players at the key customers. He knows the supply chain. He is a capital equipment guy. I was an electronic materials guy, so it is like I was not the right person to take the next step here and drive future growth. So Guy stepped in, and those of you interested, he is actually traveling internationally, which is why he is not here. Anyway, I encourage you to talk to him. Again, I think we are very well positioned in the AI space. I am going to jump ahead.
This just shows what we have been able to do from an operation standpoint. Our free cash flow is pretty. Now that we are asset light, fabless for the most part, our free cash flow is pretty much our operating cash flow. So we are generating very solid profitability. I was pleased the last quarter, 15% EBITDA margins. We took a lot of cost out. I think this actually underestimates where we were at. Yeah, this is the most recent quarter. I will pause here because I think it is important. I would rather leave more time for questions you may have. Yes.
You are talking about bringing more processing steps, right? Getting more customer on that front. Does that imply by acquisition or from the ground up?
Yeah. It is possible we will do small acquisitions, and I will give you an example. Right now, our equipment is used in mostly Class 10,000 clean room, some cases 1,000. Some of the process steps we are looking at are Class 1,000, more mainstream, sometimes 100. We could, with time, make or develop, it is the motor drives, it is the vacuum systems, it is components, and become more clean room compatible. There may be some opportunities to acquire to get there quicker. I would look at opportunities where if we could accelerate our ability to get there, reduce risk, that we would do some acquisitions. But I do not see a big target out there that we are going to go and just buy our way into another step. It is not a meaningful way to create value in my view.
But enabling technology that would accelerate our path to get there, and support our customers, I think would be interesting. But again, it has to make sense. I also did corp dev for about a decade. I am not a big believer in investing unless there is a clear path to good return on investment, so it has to meet that criteria. Yes.
On that same topic, on acquisitions
Yes
Do you have parameters that you live by for when you do acquisitions? Will you buy a company that's a fixer-upper? Will you do a dilutive deal? Will you favor equity? A related question on acquisitions, can you talk about the market out there? Are there opportunities out there? Are people expecting unreasonably high prices?
Yeah. Let me start with, probably don't need to be dilutive. We've got a strong balance sheet. I would say from that perspective, financially, we've got the war chest to do something. I would say what we're looking at tend to be more likely to be privately held. By that, I don't mean private equity. I would mean private owner and pretty small in scale. As a matter of fact, this will sound bad, but I'm not all that unhappy that at least for the past week and a half, everyone seems to be upset about AI with CoreWeave, et cetera. That's not necessarily a bad thing. I'm also not all that upset that the private debt markets seem to have frozen up.
If you're looking at some of the sources of capital for small companies with the private debt markets, a lot of that is very hard to come by right now. I think timing-wise, things are getting better in terms of being able to potentially pick things up at reasonable valuations. I don't know if that answered your question or-
It answers both questions.
Okay. Yes.
Can you talk about the competitive landscape? How much market share do you have within your segment?
We're very in what we do. When I talk about TAM, that's I think an important point I want to stress. We kind of own what we do, right? This is what we do, and the TAM will support growth of the company, 20%, 30% tailwinds, AI. That's nice, right, in this space. But if you're trying to build something of much larger scale, you've got to increase your TAM. We believe, and good reason to believe, that as we move up the layer stack, the TAM or the addressable market for each of those additional steps actually goes up as a multiple, right? So if it's X, it could be 3x for the next step up, and it's because tighter clean room requirements. You're not going to put something this big, like I said, it's the width of this room.
You're not putting that in a Class 100 clean room, right? The fabs, basically, it's precious. Its throughput per square meter of clean room space is a key metric for most of these. So you're going to have a different profile. What I'm getting at is they're lower throughput, so there's more units, right? I think as we move up the different steps in the process, the TAM goes up quite a bit. That's really our approach as well. I do think also margin-wise, we did 50% gross margin last quarter. It's reasonable, but I think those other steps also come potentially with even better margins associated with them. Next question. Yes.
You talked about the book-to-bill over the last several quarters and
Yeah, no, that's an interesting thing. The good thing about being short cycle time, short lead time, is you're very responsive. The bad thing is you usually don't get a lot of visibility. If a customer can call you and get equipment in six weeks, they usually don't book well ahead. That dynamic has changed. Our book-to-bill for this segment was just about 1.4 to 1 the last quarter, and it was 1.2 the prior quarter. What's happening is the big guys that we all know, the Nvidias of the world, are basically concerned about the supply chain and being able to support growth. So there's now pressure for everybody to place orders ahead of time to make sure there's visibility. We didn't necessarily need it. We're glad we have it.
Where we typically would book, ship same quarter, maybe straddle a quarter, we now last quarter got bookings that go out a quarter or two. You probably have read this, but like an ASE, an OSAT, like an ASE, they're building four fabs in the next year to two. They're talking about building 15 through 2030. These fabs, this isn't one or two pieces of equipment. Some of these fabs can have dozens, even up to 100 pieces of equipment in a fab. These are big numbers as they expand capacity to support this AI growth. But we're getting better visibility, which is nice. We are getting backlog that historically for this segment we have not had. Yes.
Do you see revenue disruption from thermal compression bonding or hybrid bonding?
Yeah. The question is, are we seeing demand disruption from thermal compression bonding, hybrid bonding? That is just for those of you not familiar, the technology for building the high bandwidth memory. You can either do solder bumps or you can do thermal compression bonding. Thermal compression bonding is where they're going. It's very expensive, but that's where they're going. Not for us because we don't play there. Quick answer, right? We don't play there and probably wouldn't. When we say other layers, I think that's a leap in terms of it's a higher risk, higher cost leap for us than some of the things that are closer to home. I should have mentioned this earlier. I think other areas that are going to be nice tailwinds for us where we have good process of record experience.
When everyone talks about co-packaged optics, that's a very nice. It's just starting, but that's a nice secular tailwind. I would say panel-level packaging is the other area that's a nice tailwind in the industry. It's not just AI panel-level packaging. It's analogous to the semiconductor industry, right? Because it used to be a long, long time ago, right? You had 4 in wafers, they went to 6 in, they went to 8 in, they went to 12 in.
Why'd that happen? It was economically driven, right? Because if you process a 12 in wafer, the unit process cost is more than an 8 in or a 6 in, but it's not twice the cost, right? Yet you get twice the number of die. That's the same thing with panel-level packaging. When you're producing packages, small packages, or even like if you get to Blackwell or Rubin, they get bigger, right?
But when you're producing discrete packages and you go to a panel level and I can get 100, 200, it's a lot cheaper than producing them individually. Now you're starting to see panel-level packaging migrate to, I'll call it more mainstream. Like an RF module for your cell phone or an LCD module or things that are more mainstream applications. Now why does that matter to us? Again, I'll elaborate a little bit. We aren't back to the technology. We win because we have differentiated performance. We have TrueFlat. We have better temperature uniformity. We can handle higher thermal mass. Historically, before AI showed up, where did we win? We won in CPUs because demanding. We would win in automotive because of the cost to quality issue. So recall risk, they'll pay up for equipment. We sell ours for more.
They'll pay up for equipment because they can't take a chance of a quality problem. We win in DoD applications. They'll usually pay for quality and some medical. In the mainstream electronics areas, we usually didn't win because we were more expensive, and they didn't need the requirements. The nice thing about co-packaged optics and panel-level packaging is it's hard. Yield matters a lot more. So where we win is where it's more demanding and panel-level packaging plays into that. Co-packaged optics play into that. Obviously, all the AI plays into that. The industry has also been migrating towards the tougher things get, the more you need better equipment, and you're willing to pay more. So that's been very positive for us. Yes.
You talked a lot about the build-out of AI data centers as a macro driver of demand. The other macro driver I want to ask you about is what we read is the huge onshoring and increase of fab
Yeah
production capability in the U.S., driven by the U.S., which is huge.
Yeah.
Are you seeing that? Are you seeing demand? You are in Arizona.
Yes. I think it is coming. It has come on the wafer level. Packaging, Amkor has announced. It is starting, but they are still probably 18, 24 months out. My first job was at IBM. I will date myself. It was probably 40 years ago. At the time, there were 3,000 circuit facilities, packaging circuit facilities in the U.S. I ended up working for a Ford subcontract. I did electronic packaging for quite a few times. Then it all went overseas. Anybody who knows how to make a circuit in this country is my age. The problem we have is really in the U.S., it is going to take a while because we have got to skill up. When you look at what an ASE is primarily doing, they are not like the four.
They are doing it in Taiwan, but they are also going to other parts of Southeast Asia because they are still easier to get the skill sets in Vietnam, Philippines, Thailand, Indonesia, even India. Labor is a much smaller component of the cost structure, of cost of sales. If you are looking at the semis themselves, you go to packaging, even with automation, labor is a bigger component of the cost structure. So it is going to take longer. I think it is coming to the U.S., but it is going to take longer. Part of its skillset, we have got to get people trained. We are going to have to do a lot of back to when the entire industry. I had so many people I knew who went to live in China, in Taiwan, other places when the industry went away from the U.S.
A lot of people are going to have to come from overseas to come help us bring it back, I think. Well, I think we are out of time. A stop sign and everything. They are not kidding. Well, thank you everybody. I appreciate your time.