Gayn Erickson: Yes. So we’ve been kind of holding our cards to our chest for several years on this thing and just recently have started to talk about it. So with the announcements by some major suppliers, the two largest microprocessor suppliers in the world, the main graphics processors companies in the world, even some of the large fabs like TSMC and GlobalFoundries have created these consortiums to talk about heterogeneous integration, which is a fancy word for multiple chips in one package that include a fiber optic transceiver port on it. And what they’re saying is servers first are going to start having chipsets that are in communication with processors and disk drives and data storage through fiber optic ports directly.
That is a huge deal, okay? Because the fiber optic transceiver itself will still require the stabilization in the burn-in that we have now been doing for years. It’s really what all the hub up has been about and why there are so many companies and so much investment that’s been in there. Now having said that, I think there’s going to be probably fewer big players than there can be even in silicon carbide, but there’ll be lots of smaller players certainly over the next, say, three years or something along those lines. But any one of those big players could be larger than our biggest silicon carbide customer. So the total available market for the silicon photonics when you start talking about it being embedded in servers and chipsets and processors and GPUs, I think is bigger than the silicon carbide business.
And so as we start to look at the second half of the decade, when that will really kick in, in particular, that’s a huge opportunity, and we are all in on that. So a number of investments that we’re already making some we haven’t told you about are directed directly at that space.
Christian Schwab: That’s fantastic. No other questions. Thanks guys.
Gayn Erickson: Thanks, Christian,
Operator: The next question comes from Jed Dorsheimer with William Blair. Please go ahead.
Jed Dorsheimer: Yeah. Thanks. First off, congratulations on a spectacular quarter. So nice work.
Gayn Erickson: Thanks, Jed.
Jed Dorsheimer: I guess my first question on the — your first new silicon carbide customer, the one that’s purchased two XPs with the automated handler. Is the automated handler going to be shipping at the same time as the shipment of these two tools? So will you have both available to ship with that automated handler at the March time frame?
Gayn Erickson: That’s a pretty good specific question. Let me leave it a bit — so yeah, — well, I’m going to just go out and say, the customers actually requested that we shipped the XP system with the first one, shift the second one integrated with the aligner and then upgrade the first one with the aligner. That’s what we are doing. And so there’s a little bit of timing, but they’re all about the same time frame. So there’s like a — but it’s sort of a risk reduction thing for everybody on doing that.
Jed Dorsheimer: Got it. And from a rev rec perspective, is — or do you expect that these will fall into this fiscal year’s revenue recognition or is it
Gayn Erickson: Another good question. Yes, no, that’s the advanced question, Jed. Folks, we did not see, Jed, with these questions, okay? So — no, Jed, that’s dead on. So as you understand, we seldom if ever get into rev rec things. But our policy, which is extremely conservative for a hardware company, is that if we have a new product going to a new customer until that customer says, I’ve accepted it, we do not score revenue, even if they pay us for it entirely. So the interesting thing here and embedded in our kind of weird range of forecast for the $60 million to $70 million includes the timing of when one or both of those systems would get rev rec. So — but it could very well happen that we don’t get revenue recognition for those systems until Q1 when they’re finally accepted even though we ship them in Q4.
Jed Dorsheimer: Got it. That’s helpful. Thank you. And I guess, I was wondering if you could just outline if the material quality of silicon carbide milliohm (ph) resistance, it’s coming down in the material, it enables for a multiple shrinks. Every time you would have a shrink or your customer would have a shrink in terms of chip design, that will trigger new consumables from a WaferPak .
Gayn Erickson: That’s right. Yeah, WaferPak — full wafer contactors are unique to the wafer design, which is unique to the device design. So the very nature of the word that shrink means that it’s going to — the X by Y square area would get smaller. And as such, the pads would change and it would require to add new WaferPak. This is very similar to what the test business is referred to as a probe card. And so the probe cards become the consumables. And even if they don’t need more capacity, if they simply change all their wafer patterns, they would buy only probe cards or WaferPaks in our case.
Jed Dorsheimer: Got it. So when you look at and you talk about capacity needs simply for silicon carbide from a wafer start perspective, there’s a third that presumably if this industry gets on a similar to a — I hesitate to say it more as well, but a shrink scale, you would have a 30% recurring revenue stream associated with that, not to mention sort of the movement over of additional machines for different diameters.
Gayn Erickson: Yeah. So let me put better numbers around that. So in a typical purchase upfront for our tool, we’ve talked about these ASPs in the like $2 million, $2.5 million or so for a tool and then a set of WaferPaks might be $1.5 million or somewhere in there. So there’s sort of this two-third, one-third rule. So if you use the one-third rule and call that 30%, I’m with you, okay? So upfront, they would buy 67% tester and 33% what consumable or WaferPaks. But in some point in time, they would then — those WaferPaks would be no longer valuable or useful because they no longer sell those wafers. They would then buy all new WaferPaks from us. So the one thing that’s still going on with us is trying to get our arms around what is that cycle going to be.
On one hand, automotive devices tend to last longer. So they might be the ones that might last multiple years, whereas something like a consumer device or memory might only have a 1.5- or two year life, meaning every other year, all of the probe cards are replaced on a fab, okay? In our case, though, with silicon carbide, there’s multiple things that are going on. Everyone is talking about Gen 2, Gen 3, Gen 4 shrinks. They’re talking about a planar versus trench. They’re talking about going from 150 millimeter to 200 millimeter, each of those create dynamic transitions that would actually accelerate the obsolescence of the previous generation device. So we’ve been using maybe every four years right now, and they’re still yet to see it. But there’s no doubt that when you look out four, five years, a significant amount of our revenue is going to be coming from WaferPaks of the installed base.
And we’re already seeing that with our silicon photonics customers, for example. So — and as our installed base grows, that number would grow as an absolute value as well.
