Up next, we have Bob Daigle with Amtech Systems.
Well, good morning, everyone, and thank you for joining us. I see some familiar faces around the room, we'll do a general overview of the company before taking some questions. I call it Amtech 101, for those of you unfamiliar with the company. The exciting thing about our business is we're one of the fortunate companies to be benefiting a tremendous amount from the huge infrastructure investments being made for AI. We always had a leadership position in equipment used in the back-end processing for basically CPUs and automotive electronics, various applications. With what's happening with the AI investments, we've seen really a tremendous growth and demand for our equipment for advanced packaging. I'll walk you through that. There's a couple of different parts of the business that we'll talk to. Again, I'll jump over this. You can find our slides on Amtech's investor site.
Here's our forward-looking statement. Let me kind of give you a sense for the company. Overall revenue for Amtech, annualized basis, about $80 million. We have two reporting segments, the largest of which is our thermal processing solutions, about 70% of our revenue, and about 30% in semiconductor fabrication solutions. A way to characterize these two segments, primarily capital equipment for back-end processing, and by that I mean it's chip packaging, electronic assembly related. Then in our semiconductor fabrication solutions segment, it's really the front end. It's, for those of you familiar with the industry, things that are mostly related to chemical mechanical planarization for wafers. Pretty much, I'd say front-end, more mature node, where our involvement and our growth in our thermal process solutions has really been back-end and AI-driven. Just to put things in context, perspective.
This kind of equipment, which for those of you in the room here, pretty much, let's say, scale of this, here to that wall. This is pretty large scale, maybe 4 ft or 5 ft, actually 5 ft to 6 ft wide equipment. This type of equipment is utilized for chip packaging as well as for surface mount assembly for the enterprise boards. If you go back a little over a year ago, we produced these in our Shanghai factory. We were producing, let's say, four to five a week, if you go back a little over a year ago. As I mentioned in our last quarterly call, we ended the quarter at about nine a week. This type of equipment, where we have a, I call it semi-fabless model, where we have a series of subcontractors that provide things like the electrical panels, the frames, the structures.
We can build one of these in about two weeks. We test in two weeks, out to the customer. We've been able to maintain short lead times. As I mentioned, four to five a week a little over a year ago. We had six to eight-week lead times. Now, north of nine a week units, and we're still six to eight-week lead times. We've been able to scale very efficiently in supporting the demand. I'll talk a little bit more about the application in a second. As a matter of fact, let me jump to that, because I think it's easier to picture what we do if we look at this. I don't know if this pointer's going to. Yeah.
If you're familiar with an NVIDIA GPU, where you've got your High Bandwidth Memory, you've got the chipsets, which are assembled onto an interposer and then onto a chip package, often referred to as CoWoS, Chip-on-Wafer-on-Substrate. Where we play, and we actually, I'd say, own that market, is on substrate. What it is, it's down here you've got a series of, they plate up solder bumps. Then when you're producing the package, you put it through a piece of equipment that basically it melts the solder and it makes all the interconnections. The reason we own this space, and we're process of record across the industry, is True Flat t echnology. What that means is, what you might imagine is as you process these chip packages, you heat them up very high temperature to solder.
Materials tend to want to bow, twist, deform when you're heating them up. If you're trying to make these very. Oh, sorry. Wrong button. If you're trying to make these connections between the solder bumps, let's see if I can find that, you basically need to keep these things flat. What we're able to do, if I can get this button to work, there we go, is basically we have IP around what's called TrueFlat technologies, where we actually pull a vacuum. We actually pull air down to force the substrate flat on the belt. Why is that important? It's yield. If things are deforming while they're being reflowed, it's a yield problem, which is the reason we've been very successful at the high end traditionally.
We're not the low-cost provider of this type of equipment, where we've historically done well are challenging applications like CPUs for computers. We've done well in automotive electronics because of the cost of quality. Basically, the recall risk is high, so they'll pay up for equipment. We do well in defense. We do well in the medical field. With AI, all of a sudden, we've seen tremendous growth in the need for this kind of equipment for demanding applications. That's key summary for today. I would say the tailwinds we have in the industry, as everybody reads, and the amount of capital investment being made, there's tons of new capacity going into Taiwan, places like Malaysia, Thailand, Vietnam, India, where they're putting new packaging facilities in place. Very nice tailwind.
The other thing we're seeing is tremendous demand for the enterprise board assembly as well, where you've got these large 2 ft by 3 ft boards, in some cases, that are going to these data centers. Again, these are such expensive assemblies that we're seeing very robust demand for equipment used in enterprise board because, again, they're willing to pay up for performance, and that's what we have in our product offering. The other thing that I'll note is if you look at basically what's happening in the industry and the amount of investment, you might imagine probably the next two to three years, there's some very nice secular tailwinds. One of the things that I think is exciting for us in terms of, we have the wave today of demand that's just driven by continuing to expand existing capabilities.
One of the things that is happening is if you look at, for example, like the H100 out of NVIDIA, H200, you go to Blackwell, you go to Rubin, things are getting larger and more complex. When you think about, many of you may have read this, where even two years ago, chip packaging was this thing, was an afterthought. It's something you did to encapsulate the semiconductors and protect them from the environment. Now everything you're seeing is people are starting to see Moore's Law is kind of reaching some limits in terms of how much capability you can add. Chip packaging is now becoming key enabler, not just for protecting devices, but for providing performance. That's a key tailwind we believe that will continue to drive this industry. What's that mean for us?
I think there's an opportunity, and we're pursuing opportunities really for higher value equipment. If you think about Blackwell's probably twice the size of the package of the previous generation, and if you look at Rubin, it's almost twice the size of Blackwell. The other thing is if you look at the roadmaps for interconnects, things are getting tighter and denser. What I mean by that are where things might have been 40 micron between the bumps, they're going to go to 25 micron, 20 micron, and less. Those more demanding requirements are driving the need for, we believe, next generation equipment. What's important about that is this next generation equipment should have a multiple in terms of price point because of the technology that we need to bring to the market. We see that as a tailwind.
I've talked about it in our conference call. Probably doesn't meaningfully impact demand till late 2027 into 2028, we see that as really another driver for our business as you look forward. With that, let me go back and talk a little bit about semiconductor fabrication solutions. Back to this business, back end, AI driven, if you follow the semiconductor industry, it's not homogenous, right? You've got a tale of two worlds. If you're NVIDIA, Taiwan Semi, AMD, Micron, life's been good for a couple of years, driven by AI. If you're in the mature node world, the 28 nanometer automotive industrial applications, that industry's been soft for a couple of years now. Partly, just reduced overall demand, you also have emerging Chinese competition where China has put in, I think, about as much capacity for mature node semis as the Western providers combined.
You have that potential headwind in the mature node world. With that backdrop, like everybody else, we've faced soft demand in our mature node business, which is our semi-fab solutions. About nine months ago, we made the decision, we stepped back and said, "Let's assume the industry doesn't become robust again. Let's say with the headwinds out of China, demand remains soft. How do we control our own destiny? How do we grow double digits even in that environment?" Two initiatives which are starting to play well for us, one of which is we are in a unique situation. We have a small specialty chemicals business that does wafer cleaners, lubricants, slicing, dicing, lubricants, as well as some CMP slurries.
We have something that is pretty unique as well, where we have a CMP foundry service, and we do contract work for startups, DoD, emerging substrates, where we actually have clean rooms, we have polishing equipment. What we did was we combined those two parts of our business, and we said we want to prioritize work in our foundry with applications where we think we can build a consumables reoccurring revenue business. An example I used at a recent conference call is, a medical device company in Massachusetts gets notified by their semiconductor supplier that their CMP slurry is being discontinued because the volumes are too low. It's a niche-y low volume application. They come to us because we do contract qual development work, and we basically get paid for this.
We said, "Sure, we'll do the work, but can we include our CMP slurry in the evaluation?" They say, "Fine." Ours works great. We're now their process of record for that device, and they tell us every new product they're developing is going to use our slurry. We're trying to build this pipeline. In the most recent quarter, year-over-year growth, we're still in the pipeline building mode. It was up 15%, though, year-over-year for our specialty chemicals part of the business. That's encouraging sign. The other area we're focused on is, if you, again, you step back and you say, if the Western semiconductor fabs are facing increased competition and cost pressures from China, and my understanding is the wafers out of China can be 30% cheaper than the guys in the West. How do we help them succeed?
We've had a refurb business where we do aftermarket AMAT tools, Mirra, older generation Reflexion tools, where we do some service, and we do parts. We built out more of that library of parts in our service team, and we're basically offering is, these are usually tools that are fully depreciated in these fabs, older generation, and we basically are going in and saying, "We can provide parts and service more cost effectively." That grew 40% year-over-year. Again, we're trying to face up to the reality. Obviously, if the tide comes back in and everybody's doing better, we'll do better, we want to grow regardless. That's kind of an overview there. Let me jump ahead.
I think I'm going to give you a little bit of history because I think if you go back and look at our financials, I always get these questions as, you guys shrank. I give you a little bit of history. I joined Amtech's board in August of 2021, after 35 years in the electronic materials industry. I had built our circuit materials business, built a power electronics business, did corp dev. I was recruited to Amtech's board because back to the CMP polishing, a lot of the target business was silicon carbide for EV, HEV, and since I had managed a power electronics business, substrate business, I was recruited to the board. Fast-forward from August 2021 to 2023, it became apparent to the board the company was in trouble financially.
We were burning $2 million-$3 million a quarter of cash, had $11 million of bank debt, violating bank covenants, we needed to take some action. I stepped in as CEO in August of 2023, and I call it the three phases to the Amtech journey. Fortunately, the past year, we're in the positive side of things, but it was survive the first six months, optimize for about 18 months, and then revitalize, which is really around positioning the company for growth. Back in the survive mode, there were really two big things that we dealt with. One was the realization that there was pretty significant investment being made in next generation capital equipment for silicon carbide batch processing. When you peeled the onion, you realized what we were developing wasn't going to be all that differentiated.
The margin profile wasn't going to be all that great, yet we were investing all this money. We pulled the plug on some, I think, questionable return on invested capital, in cash burn. The other realization, there were a bunch of little smaller product lines, 500,000 , 2 million legacy, that you looked at the margin profile, 15%-20% gross margins. Hadn't grown in a decade. Again, there wasn't any real product differentiation, we did some pruning of the portfolio. About six months, and we got ourselves to basically cash flow, break even or better in the various pieces of the business. The heavy lift was really optimize. Maybe it's because I came from more of a reoccurring revenue business, but I struggle with the idea that these capital equipment businesses can be highly cyclical.
In particular for Amtech, we had some parts of our products where when demand was high, the backlog went through the roof. We couldn't seem to get it out the door. When we did get it out the door, the margins were low, and realization that we just weren't very good at making equipment. We moved to a contract manufacturing model. Where we had seven factories, we now have four. Only one facility is now making a significant amount of equipment, and that's our Shanghai factory, where even that's semi-fab. We assemble sub-assemblies and test two weeks, two weeks out the door. We basically went to a contract manufacturing model. There were parts of the portfolio that when we were making it ourselves, 15%-20% gross margins that are north of 40 now. People keep saying, "Well, how can that be?
A contract manufacturer is making money and now, you outsource it and you're saying more. We were just that bad at it, frankly. This model also creates some other benefits when you think about it, right? We're now an asset-light company. Back to the scalability, we were able to double output without having bricks and mortar. We were able to scale our supply chain. What it also means is our free cash flow is pretty close to our operating cash flow. What I've talked about in our investor calls, my estimate is probably capital needs less than $1 million a year, even as we scale, and that's mostly things like analytical equipment for a specialty chemicals business or ERP systems because our backend's too expensive right now on G&A, so we'll invest in better systems to streamline our costs in G&A.
With this transformation, we basically were able to take our break-even point, pull north of $13 million of fixed costs out of the company, took our break-even from $120 million down to $75 million, obviously paid off all the debt, and ended the last quarter with $24 million of cash on the balance sheet. The past year has really been nine months, let's say, we've really been focused on the growth side of things. We've got our house in order. We've got great operating leverage across the segments. We're still sub-scale. Last quarter, we did about $20.5 million in revenue. I think of healthy companies as being 15%-20% EBITDA margins. We only did 12%. We're making progress. We did almost 48% gross margin in the last quarter. I think there's room for that to grow as well as we scale.
Really, right now, it's capitalize on the growth. As we drive revenue, I think we're going to be in a position where we're not that far off, I believe, from getting into that 15%-20% EBITDA margin range for the company. With that, just a couple of additions, because some of you may be aware, we did raise capital a few weeks ago, actually. We put $60 million on the balance sheet with the idea that we want to scale the company, but want to do it in a very investor-friendly approach, where I spent 10 years doing corp dev in my prior job, and a lot of lessons learned about what worked financially and what didn't. When there's a whole bunch of hand-waving about synergies on revenue, usually they didn't work out. Usually, you overpaid.
When there were true hard synergies and you really could tuck things in or really leverage cost structure, they worked if you didn't overpay. We positioned ourselves to be a credible acquirer. We've built a stronger team. Recent adds are Guy Shechter, where I said we play on Chip-on-Wafer-on-Substrate. We're in the on-substrate part of Chip-on-Wafer-on-Substrate. Guy comes from the Chip-on-Wafer world. Then Tom Sabol, who has a lot of contract manufacturing background, has done north of 25 M&A deals in his career, so very experienced. We've built a team which I think positions us where, again, people say, "Are you going to acquire?" I always say maybe. It's like we've got great secular tailwinds. We've got a great organic growth business. To the extent we can scale the company through M&A, we'll do it.
We're not going to do for the sake of scale. With that, I think just quick summary. I think we've got great operating leverage. We've seen it in our most recent results. We're generating cash, we're in some pretty good growth markets. Where we're not, I think we've positioned ourselves to control our own destiny for growth. I think we only have a couple of minutes for questions. For those of you, our booth number is 221. I will head over there after this session if anybody would like to stop by. Any questions before we have to wrap up? Yes. Yep. The question, if I understood, is a lot of announcements, Taiwan Semi and others, about panel-level packaging. I didn't really touch on that.
Yes, we play in panel-level packaging, we're shipping equipment for panel-level packaging. There's a little bit of a caveat to what I'll say is because AI will drive panel-level packaging, but something that's actually very interesting for us. It goes back to the fact that we've played at the high end and in some of the more mainstream consumer industrial applications, there's cheaper equipment out there. What's interesting about panel-level packaging is it's hard. Now you're starting also to see, in addition to AI, you're seeing things like RF modules, LCD displays, more mainstream electronics go to panel level. What's good for us is that's challenging, which has created a bunch of opportunities for equipment in more mainstream applications. Yes, definitely panel-level packaging is a very nice tailwind for us as well. Yes.
I don't know the business that well, your back end, you got H1, it's kind of written into the process for H1. Every time a new chip comes out, does Heller or one of your competitors have a chance to say, "We got this wrong?
Yeah, we've generally owned the high end because of our True Flat t echnology, we've been plugged in with the OSATs and the OEMs. I'll tell you, everyone seems to be scrambling at same capacity. Can't say that's forever, but I would say right now we're not giving anybody any reason to look elsewhere. Back to the scalability, we've doubled our equipment deliveries without having to increase lead times, we've positioned ourselves, I think, with even, let's hope it all doubles again, that we can take care of our customers well. I think we're feeling pretty good about our market position. Again, next gen is key because when you look out 2028, having equipment that potentially is a multiple of the current ASP, I think could be exciting in terms of driving more scale here.
That hits our time, Bob.
Oh, okay.
Again, thank you everyone for your interest and we're in-