Gayn Erickson: Okay, two things. Okay. So if you’re going to take growth rate, be careful, because a small number of growth rate can seem large and it’s a small number, so are there — are those bigger? The optical IO market, we’ve said, we think can be as big or even potentially bigger than even silicon carbide simply because of the potential units and the test time, the dollars, it turns out it’s actually much more expensive for us to build, even though it’s extremely cost-effective. So it has the ability to be bigger. Memory absolutely is the biggest of all of them, right? And GaN is a TBD, I think, but it looks promising. So if I try and give you relative sizes, okay?
Matt Smith: All right. Thanks. I appreciate the color.
Gayn Erickson: Okay.
Operator: Thank you. The next question is coming from Larry Chlebina from Chelbina Capital. Larry, your line live.
Larry Chlebina: Hey, Gayn. I have several questions and we get through them fairly quickly. We are getting late. Do you expect follow-on orders and shipments this fiscal year to the fully automated silicon carbide customer that you just booked the revenue last quarter. Do you expect shipments and revenues, additional follow-on with revenues this fiscal year from that customer?
Gayn Erickson: Yes. We — I am looking down to the whole list here. I think we are expecting some level of revenue from most customers including them.
Larry Chlebina: That was a significant order and since they ordered two of them, I would expect there — they have a significant ramp that they’re going to need some more capacity.
Gayn Erickson: Hi. All of these — all of the customers need significant more capacity to meet their revenue forecast. That’s correct.
Larry Chlebina: Could it be that maybe the slower uptake of some of these customers, particularly the one you’ve been working on forever, is simply the fact that they’re building 200-millimeter fabs and they’re waiting for the 200-millimeter wafers before they commit to your full system so that they don’t have to buy two sets of wafer packs? Why start at 150 if you wait a little bit you’re going to be on 200. Maybe they have a bunch of single wafer probers that they can get by on. It’s not efficient but could it be something along those lines?
Gayn Erickson: I don’t think that — it could be, Larry. I mean I don’t think so. I look at each individual customer. Most of them are all planning to go from 150 to 200. Now, it’s not — so if you — let’s say the 200 is a new line completely. Very often people may want to change their strategy and go from discrete to die level on a new line, let’s say. In that case, maybe you’re right, they would time it with the 200-millimeter ramps. But that hasn’t been how it has felt to us. People have — I think the 200-millimeter is coming. I mean, we’re testing 200-millimeter wafers today from multiple customers, right? We see it. I think it’s kind of consistent with everything that everybody’s hearing. We don’t have any particular great insight.
I just don’t — it’s not — I think it’s slower than people have originally thought. I don’t think it’s going to be 100% of the market, it’s going to — through the end of the decade, there’s still going to be a big chunk of 150-millimeter wafers being built. One of the big discussions that’s going on is will, a 200-millimeter wafer yield higher than 150 at a lower cost per yielded die? And right now I think everything I’ve read is not yet, and so — but that’s also very true with most of these transitions. It’s sort of chicken and egg, you have to lead with it and then eventually your costs come under control. But one thing with us is that, the system is fully compatible between 150 and 200. You can build our auto aligner and it can work with both.
We have current customers that in the same day are doing both. And so — and the wafer packs, there are some things we can do with wafer packs to actually leverage and help with that too. But yes, I don’t think that’s the majority of it, but it might be on some customers.
Larry Chlebina: Okay. On the optical IO customer, it’s interesting that they started with [NPs] (ph), if I recall, four of them that is scattered around the world, and then now they’re pushing you to pull in and complete as soon as possible the production XP for that application. It would seem that in a fairly short time, maybe sooner than later, they might be needing quantities of production tools. As fast as that market is growing on the AI front, possibly it would give them a terrific leg up on their competitors. Do you think that could be something that might be at work here?
Gayn Erickson: Maybe, yes. I mean, we’re preparing the tool. It will allow us to duplicate and build multiple of the same flavors. One of the nice things that we did with this one is the system is actually in the new chamber configuration, so it bolts right onto the front or the back, if you will, of the aligner. So they could do it either with offline or integrated. It was not purchased with the integrated aligner, but we could easily upgrade it in the field if they wanted to do that. And we think — I think that makes sense for that market. I think the aligner is the right answer. And as high volume, that would totally make sense.
Larry Chlebina: Exactly. So along the high volume. Last question on the — you mentioned many times that every DRAM is burned in, I’m assuming some kind of exotic package part of burn-in or whatever. But the — the large DRAM producer in the US is planning on building actually several fabs in the US. And they’re also introducing high bandwidth memory this quarter, I believe, with 12 stacks of DRAM on top of it. It just seems like a layup. I know it’s more complicated than that, but an application that would lend itself towards wafer-level burn-in versus doing testing on that stack and finding you got a bad die in there somewhere. Would it make sense to try to get an evaluation tool on a fully automated XP over that memory customer before they set their plans on clean room space and so on and so forth, particularly before they commit to building those fabs.
