Jed Dorsheimer: Got it. That’s helpful. Last question for me, I’ll jump back in the queue. The — did I hear you correctly when you talked about your second new customer, did they have not announced products in silicon carbide? I was wondering if you could just help clarify that.
Gayn Erickson: Yeah. I’m choosing my words very specifically. They have not introduced silicon carbide MOSFETs, and I looked at their website again last night just to double check. They still have not — they still have said it. So they have figured out a program to come stealthily at this, they have multiple devices I know because we have their wafers. And so I — we’re being pretty elusive. I realize someone — I think someone asked to why you guys can’t never mention about customers? I’m telling you that customers do not want to be mentioned. It’s the biggest secret around. And we’re always trying to be very careful. I don’t even mention the name of the customer outside of the context anymore because they will get bad at me if I mention them, right.
So it’s a difficult thing, I realize. We’ve said we have two of the big guys. I will be bold and I mean, I think because of them being a 10-K — 10% customer in our 10Ks, I think people understood that on Semiconductor was one of our biggest customers. Everybody is trying to guess who the next big guy was — is. And we’re — I’m not sure exactly how that’s all going to play out. Right now, burden is very clear. I’m just going to win them all, and then we can say we have all of them. So — but in the meantime, I apologize that you guys don’t have better insight as to who they are.
Jed Dorsheimer: No apologies is job security for us. We have to do some real research. So I like it. Thank you.
Gayn Erickson: Thank you, Jed.
Operator: The next question comes from Dylan Patel with SemiAnalysis. Please go ahead.
Dylan Patel: Hey, Gayn. I wanted to ask the WaferPaks, the various burn-in equipment on both a silicon carbide side and photonics side, is that going to be in the wafer fab or can it be in the test fab? Because I was just trying to think about it from my perspective, and I think it could be in either.
Gayn Erickson: Yeah. You know what, it absolutely can be an either. People — the closest analogy is what’s referred to as known good die, where people sell die sales, okay? In that case, the final test of the die often happens in what we would refer to in the test community is the back-end operation where packages are, okay, versus the front end where the wafers are. In this case, the wafer level burn-in process step feels like a known good die step. So it would be reasonable for somebody to put it in their back end. Having said that, I — most of our equipment today, if not all of it, is in the front end. It naturally fits in the wafer fab right next to the fab. Our systems are rated for clean room specifications. They’re intended to go into the clean rooms. And so we’ll see both for sure. But I’ve heard both. I think we’ll have systems both places.
Dylan Patel: It’s burn-in and then stimulation next, right or is there anything else in the middle?
Gayn Erickson: No, there’s test after burn-in on your wafer always. So our recommended process is take a raw wafer, no need to test it. We will test it and tell you which are the good and bad die, we will burn it in and we’ll let you know which ones die during the process and exactly what time it did. And then you would correlate that with a functional final test of that singulated die because you can get tighter accuracies and some other things that you might want to do. But you would do a single insertion test with a test system like the likes of a Teradyne, there’s a handful of little companies that are out there that test one, two, three, four devices at a time. And then on a wafer with a 1,000 devices, they might test 250 to 500 insertions of about one sec in a piece.
And then they would singulated the good die — singulate, separate the good die and oftentimes then that die, so they will pick and choose die with matching parametrics. One of the things that people — I’ve spent more time on in the past is part of the process of silicon photonics today and silicon carbide. And I did this in a white paper in Munich in a presentation of it. There’s also a bunch of technical white papers out there. Silicon carbide devices age and actually their parametric voltage threshold, the threshold at which the device turns on, changes in time over multiple hours and then stabilizes. If you put that into a module like you would do in, say, a module that goes into even from Tesla, but Lucid or Ford’s or any of the automotive modules that are going on.
You — and you put, say, eight devices in there. The devices would all be turning on at different times, if you do not cherry pick them. So let’s say, you cherry pick them and put them in, during the first 24 hours of use, the threshold voltage varies. So then the one that turns on the fastest will wear out the fastest and can become a reliability problem. So when people use our tools, they’re not just being good from bad or weeding out what we call infant mortality or early life failures, we’re actually stabilizing the threshold voltage, which takes a period of time such that they can cherry pick them and then sell them to the likes of the VWs of the world or Dan Foss or BorgWarner, who are going to be purchasing known good die from all of the suppliers with specified threshold voltages and RDS on voltages or RDS on impedances or resistances to catch all that.
