Hi, everyone, and welcome to the 27th Annual Needham Growth Conference. And thank you for joining me this afternoon. I'm here with Aehr Test Systems, and I'm thrilled to have Gayn Erickson, President and CEO, and Chris Siu, CFO, here with me. Gayn and Chris, thank you for joining us today. So first, I would like to start with a presentation, and then we can turn to some Q&A. Sounds good. All right. Hello, everyone. Thanks for joining in on the with a fair look at everything that we're seeing today and maybe provide some insight into the markets as well, and a picture of myself here. So just a little bit about Aehr Test. Aehr Test is in the semiconductor test equipment business. We do electrical tests of a wide range of semiconductors and have been doing this for a very long time. Within the test segment, which generally you think of testing as determining good from bad, all semiconductor devices go through not just some level of test, but many times multiple steps of tests along the way during their manufacturing process, step testing while they're still in wafer form, sometimes when they are singulated, always again when they're put into a package. And then even sometimes when they put multiple devices into a package, into multi-chip modules or systems, they'll be tested again. We actually specialize in a type of test normally referred to as reliability and burn-in, which is beyond good from bad. We'll actually take good devices that have previously been tested and gone through the manufacturing process, and we will do a stress test on them that is intended to weed out devices that have a higher level of what's called early life failures, infant mortality, before they're shipped to customers. All semiconductor devices have some level of failure rate at their inception. And interestingly, they follow a thing called a bathtub curve that allows them to, that actually increases the reliability or decreases their failure rates over time. What we're particularly good at is helping customers to weed out those failures. We have product lines that range from full production systems for wafer level package part burn-in systems as examples that are here within that are shown on the screen here. We have a large customer base over the years. This is only a partial list. We have a number of customers that ask that we do not put their names on here. Many of these customers have shown up as 10% customers and through SEC requirements required us to put their names on, or otherwise we've done white papers with and other public endorsements along the way. But it really covers the gamut across the entire globe where we have systems, infrastructure, people to support those tools as we have shipped thousands of systems worldwide over the years. Our market drivers are the kind of the key market drivers driving a lot of the semiconductor industry, although we really focus on high reliability requirements that are showing up in some of these key industries. One of the things driving the biggest growth right now this year is that, like a lot of folks related to artificial intelligence processors and the infrastructure around it, electric vehicles and electrification of the worldwide infrastructure are also key drivers. Those components tend to have reliability rates that are inconsistent or they have higher failure rates than the world can live with. And so we'll actually screen out those infant mortalities to provide them to meet the critical requirements of those industries. Data centers, the enormous growth in data storage also drives us as well as just overall communications infrastructure. The other thing is just the dynamics that are going on in semiconductors are also creating the need for our solutions. Kind of mega trends, if you will, kind of the highest level trend is that we're going to watch the semiconductor industry double. After taking about 40 years to get to $500 billion, it's going to take a little over 10 years to double from there. That is an enormous amount of capacity, and there are these multi-billion-dollar wafer fabrication facilities being put in all over the world with scheduled out through the end of the decade to be able to achieve this trillion dollars and continue on past that. That capacity is driven by artificial intelligence. A lot of this clean energy decarbonization, electrification, digital/ IO are some of the really key drivers. The key here is that reliability test or what we do is becoming more and more critical. Kind of big transitions are semiconductors themselves are actually not more reliable than before. They're getting less reliable, which doesn't seem intuitive, but the larger die size that is coming about from semiconductors that are just growing to meet the explosive growth of computing, for example, means that devices are more likely to have defects. Smaller geometries are more prone to defects, and things like compound semiconductors, different than silicon-based, are driving that are being driven by power and optical communication actually have just higher failure rates. That being said, more semiconductors are going into applications where quality, long-term reliability, safety, security are more critical. Obviously, things like electric vehicles, electrification of infrastructure that you put into a facility and expect it to last for 20 or 30 years. And then just in general, the content of semiconductors across so many applications is increasing, which is driving the need for equipment like ours to be able to weed out the infant mortality shown in the first bullet in these applications. The last thing that's kind of some of the most dramatic is that with the advent of the sort of slowing or stoppage of Moore's Law, which was the observation that semiconductors' density were doubling every 18 months, it stopped doing that several years ago. So people who are consuming more and more semiconductors aren't able to do that within the same constraints. In order to get more capability, you're actually not just making the semiconductor larger, but you're actually putting two semiconductors together or putting them in the same package. So there's all kinds of applications for that, whether it be for power that's just simply needing more devices in parallel or mixing communication protocols like Intel has said in their co-packaged optics where electrical semiconductors are now going to be communicating between each other with an optical interface. Devices like NAND and now DRAM with HBM stacks, they're stacking many, many die on top of each other to get the density because they just can't make the semiconductors as dense as the applications are requiring. And then you come to something like the AI processors where you're mixing all of the above with the processors that are getting really large, HBM stacks, co-packaged with optics as we go forward on these very complex substrates is causing opportunities for equipment like ours. This just summarizes those sort of main key bullets. So in addition to semiconductors doubling, there's a need for tests because of decreasing reliability. The applications need higher reliability. And there's more need for known good die and other things because of this heterogeneous integration. We've actually published this for our shareholders to understand and for our customers to look at our fiscal year strategic focus. FY24 was through last May. FY25 is through this May. And FY26 going forward starts from June. In 2024, our business was really dominated by silicon carbide, dominated itself by electric vehicles, chargers, and power conversion, and also the transition going to silicon photonics for optical I/O and data transceivers. This year, we said at the beginning of the year, we're going to see a shift to wafer level test and burn-in of the addition of GaN semiconductors for power, hard disk drives, AI processors, both at package and wafer level. Those we would expect for revenue. In fiscal 2025, in addition to those developments to meet those expansions, we're also working on NAND and DRAM memory expansion. This is the bathtub curve I alluded to before. I just want to point out for people that are following us, we also did a strategic acquisition of Incal adding to our family in July. I think we closed it August 1st. This bathtub curve represents the decreasing failure rate in time until things start to wear out. This is basically a semiconductor last 20 years and then they'll all fail, if you will. Intel is extremely good at this curve. Their software is very well known and loved in the industry. They have a massive installed base of tools used in the engineering qualification of these devices, whereas Aehr Test historically was known for the production burn-in or the stage by which you would test the devices to weed out infant mortality. Together, we actually combine to make a very good combinational opportunity for the customers to buy from us. Just a couple of things on silicon carbide real quick, and I'll go through this a little faster. This is an example where multi-die packages are really evolving. A critical application of silicon carbide is for electric vehicle traction inverters, and as the devices are going from discrete to modules, the need for our tools is shifting from a package part to a wafer level where Aehr is particularly or exceptionally good at is the wafer level test and burn-in of devices like silicon carbide, GaN nitride devices. Why this becomes important is basically as you put multiple die in there, if you fail one die during the burn-in, you actually lose all the rest of the die, and so in this device here, it has 10 die inside of it. You have 10 times as much yield loss or 10 times the impact of those failure rates, and so as yield loss increases, the module loss goes down linearly, if you will, at 10x the rate because of the 10 die. These are real numbers, and it's what's really been driving our business. Another key example, which is a lot more topical these days, is the multi-chip modules associated with AI processors, and this AMD 300 series accelerator is a great example. There's actually eight processors on here and eight stacks of multi-die MCMs. Then the system has the package on top of it, but just this package itself has 16 die, if you will, of which each of these has 12 die stacks. So you're literally looking at, what, 96-104 die on a single package. And all of these devices need to be burned in. So the implications of the quality and the yield loss associated with burn-in and where you do it is really critical. This was an interesting article that was put out by Meta last summer that talked about the failures they were seeing of their GPUs when they were doing a Llama 3 LLM development. And they were showing that GPUs and the HBM memory on those GPUs were accounting for 47% of their interruptions related to the failure and reliability of those. So those are those same parts here that are seeing infant mortality rates because they are not screened well enough. One of the reasons the tools just fail to exist out there to be able to do that, which is one of the critical opportunities that's driving our business. This just talks to the patents that we have in our wafer level burn-in and the requirement to actually have a full turnkey solution. Something that Aehr Test has is we not only develop the tester, we develop the handling mechanism, but we also develop the consumable contactor as a sale as part of the total solution. And so there's no finger pointing whatsoever. This is an example and a photo of our multi-wafer system. There's literally nothing like it in the world as we introduced this thing four years ago, five years ago to the market. It's very flexible in its ability to address a wide range of devices, power levels, voltage levels, and so far, we've proven that across a wide range of devices from optical to power to AI and even memory type devices on our FOX S ystems that will allow you to cost-effectively do wafer level test and burn-in. In the world, people think about testing a wafer with a tester that might test across the wafer, stepping across the wafer to test all the devices. We're not only testing all the devices on the wafer, but we're testing, in this case, 18 wafers at a time. We introduced a couple of years ago this new fully automated aligner that we're now shipping in volume as of about a year and a year plus ago. We started to do this across multiple applications. That includes silicon carbide, gallium nitride, capable for AI processors, and other devices. This is the platform we're doing the memory development on as well. But this allows people to do full hands-free operation of their wafer handling, ranging from 4-inch to 12-inch or 100-millimeter, 300-millimeter wafers. This is a picture of our new Sonoma system through the acquisition of Incal Technology. It's already shipping in volume, badged with our colors, etc. Aehr Test's manufacturing group is pulled in. We're probably at four plus times the amount of manufacturing capacity that they had, and we're already shipping in volume on these to both qualification and production customers. One of our keys is around testing without compromise. It sounds like a bit of a shtick, but in reality, we do things differently to absolutely positively confirm the traceability of every single device that's tested. Because if you don't test a device or you don't burn it in and you thought you did, in many cases, you have greater than a 1% failure that it'll fail in the field. When you start thinking about that in an automotive application or in an engine controller, if you had 1% of the die fail, you're going to have one out of 100 cars that you're going to have to replace the entire inverter. There's no way you can think you tested it and didn't. This is one of the key differentiations with our tools. We have just in the last few weeks announced a couple of critical wins for us. At the beginning of the year, we said we thought we were going to be successful in AI processors for wafer level burn-in. We just announced our first order. It was over $10 million of orders for multiple systems for an AI processor application at wafer level. This is a really big deal in our industry. This is going to have a very visible footprint as it's installed at one of the world's largest OSATs, and we'll be able to demonstrate feasibility of having done this and cost-effectiveness in a production environment, and we think that this has an opportunity to have a big impact on the market and have other customers drive to this wafer level test and burn-in of the processors before they put them into package or do it at a system level test. We also announced just a few days ago our first production win for GaN power semiconductors, Gallium Nitride, similar to Silicon Carbide in that it's a high-voltage semiconductor process. It's aimed at kind of medium power applications, which are a much wider market opportunity than silicon carbide in terms of segments. Silicon carbide, very high power, and particularly at high voltages above 800 volts is where its real sweet spot is. It's found its application widely in automotive electrification or the EV infrastructure. GaN is applicable, can be up to those voltages, but tends to be more in the maybe 50-80 volts up to 700-800 volts. There's actually a lot more applications for that. And so it has a much more diversified customer base. There's some companies that have gone out and said that's a really attractive part of it. There have even been comments that state that it's going to displace all of the silicon-based power semiconductor, IGBT. We'll see if that happens. If it does, the market's going to be a lot bigger than we currently think it is. We are shipping this fully with a fully integrated handler as it's shown here to test 6-inch, even all the way up to 12-inch wafers that have been introduced by GaN companies. One thing I just want to point out, we've had this discussion for a while, this advent of optical IO and silicon photonics is coming. I updated this slide that has been similar to it out on our deck just recently to now include some of the most recent comments, including multiple coming out of CES last week. But TSMC is talking about samples that will be coming in this year. Ayar Labs CEO talked about optical chiplets are coming soon. Adding that to the Intel AMD comments and historical comments out of Broadcom and others, we think that this is something that is coming. Aehr Test is particularly well suited for doing the test and burn-in of silicon photonics devices at wafer level, and this is an area that we are working with some of the market leaders with systems that started to ship, including our systems that have been shipping for several years, including a new high-power system we shipped last year. We think that we're in a really good position for when this, not if this, but when this market starts to take off. One last thing on just overall opportunities, and I'll kind of wrap it up. Flash memory is another area that we are investing in right now. We've talked about we are engaging with one of the suppliers that we're talking with in earnest to actually do a full qualification of our test cells, so a new WaferPak, aligner, and infrastructure that they see as an opportunity to be able to lower their cost of test and address some of the new NAND devices that are coming down the market. It's going to take us through the rest of this fiscal year, so a few more months for us to validate out the WaferPak and get some of the data that we're hoping for that we believe can lead us to the next phase, which is optimizing of the system around their new next-generation test requirements. That hopefully we can secure some more perhaps even revenue in the following fiscal year, which seems like a long way out, but it's a really, really enormous opportunity. The next slide is something there's been a lot of questions related to, okay, how big are all these markets? I don't understand. And to some extent, that's still being worked on and worked through. And it's hard to put numbers around. And sometimes we want to hold that close to our chest a little bit too. But this is a fair way of looking at it. It turns out that customers normally think about their capital spend and their cost of test as a percentage of revenue. Historically, that's been 2%-5% of device revenue allocated to test. Now, if you look at the markets over the last few years as aggregate, about 2.5% of the entire semiconductor revenue was spent on semiconductor test. But that includes not very much on analog, way more on AI processors. Last year, not very much on flash, more on DRAM. So there's gives and takes. So you can actually have 10% spurts at times, but that's not sustainable. So if you use somewhere in the 2%-5%, that's probably a fair way of looking at it. These are simply revenues. If you go out and you Google what's the revenue of silicon photonics, silicon carbide, gallium nitride, flash DRAM, and AI processors in 2024, it'll give you numbers. It'll actually give you a range. This is a snapshot of a reasonable number. What is it for 2030? It'll give you another range. And that's the combined stack of orange and gray. This is a reasonable assumption of the size of the dollars being spent by our customers' customers, what their revenue is. And if you look at a 2%-5% spend of it, you can understand when we talk about why flash is so attractive to us and ultimately DRAM will be bigger. And as we try and get our arms around the size of the AI processor market, it's very obvious where we should be spending our time. This is a pretty good indicator of where we spend our time every day, where our R&D dollars are being spent, and where we're putting our efforts. And what I also show is, in the ones with the wafer level, a picture of our Fox system or Sonoma system. We have announced production wins in these segments. These are the ones we've talked about working on, the ones that are grayed out. I hope that helps and puts it in perspective. We're going to open it up for questions now. And I'll send it back to you, Ross, to lead us through that. Great. Thank you, Gayn. And thank you for the presentation. So now we have some time for Q&A. If anyone on the call would like to submit a question, you can do it through the chat, and I will read it anonymously. You can also email it to me at rcole@needhamco.com, and I can read the question from there. I think while we wait for the questions to come in, maybe I'll kick us off. I know recently on January 13th, you reported some results and guidance for your fiscal 2Q25. I was wondering if maybe you could walk through that a little bit and provide us maybe some insight into your results and guidance. Okay. Well, one of the big news for sure is that we did not, our numbers came in less than the street had us for our Q2. We actually don't provide quarterly guidance, so I guess you could have a discussion, did we miss our expectations? But we maintained our guidance for the year. We try to be really specific, and if people are just looking and saying, "Oh, they missed," that's one of the reasons we think, or the primary reason our stock took a hit and it was trying to recover yesterday, is we have stated in the past, and it's still true, that our business is quite lumpy. An ASP of our wafer level burn-in systems is north of a few million dollars. When you're talking about quarters that might have $15 to $20 million, plus or minus, one of those shipments can have a material effect, if you will, on the dollars. One of the key reasons we simply just don't provide a quarterly guidance. Our two deals we just announced, both the AI and the GaN customer, both customers asked us to ship in November during our Q2. We pre-built those systems, had them ready to go. They were literally in crates, and we were scheduled for shipment pickups on those before the end of our fiscal year per their request. Neither one of them got us the PO in time. Challenge with a public company is, I jokingly, we do have a policy. We don't ship before a PO. In this case, it bit us. We have since gotten the orders, and we're shipping them in January. I mean, we're shipping them now. But in fact, they didn't even come in the next day. We were thinking they were still coming in that week, and it took till after the holidays for them to get us the orders. It's one of the challenges we're still working through, trying to work through that with our customers. Interestingly, both of those customers are screaming to get them shipped and installed immediately. That's a different problem. The reality is that there was no change in what was going on with that. The forecast for the year maintains at what we believe will be at least $70 million. Of course, there's risks. There's also upside to that. We always try and balance that when we communicate that. But our current snapshot is believed that we will be able to do those numbers. The quarter to quarter is always fun. But we start off stronger this quarter as a result of shipping some of the systems we thought we were going to ship last quarter. Great. Thank you for providing some color there. Always exciting to hear about your new shipments and tools. And I know you have a lot going on there. So I was wondering for another question, maybe could you provide some insight into what you're most excited about coming up for your new tools going out, any kind of milestones you're expecting? Well, I mean, we had been working with the GaN customer for a year to prove out the capability. It was great. I mean, we knew it was coming for a little bit, but it's always good to get the first order in and celebrate that. It's a leading indicator. It's a key supplier in the GaN market. I would maybe argue one of the, if not the leaders in the GaN area, and them proving out the wafer level burn-in and the commitment to moving to production on that, on burn-in of all the GaN devices is critical. That's a big deal for us, and it's going to drive business for a long time. To be fair, it's kind of hard to beat AI, though. The AI stuff, it's obviously exciting in the market, but there's just so much growth going on. There's so much money going into that space. We've now captured not only a production burn-in of package part, the first one for AI for our Incal group under Aehr Test. We were able to convince the customer that Aehr can actually support both the production requirements to build and support and the quantity and have the resources to do it. We're already shipping in quantity those systems at a level much higher than Incal could have done. Our infrastructure and people are much larger than Incal, so we've been able to fulfill that ahead of schedule. That bodes very well from a package part burn-in perspective. On the wafer level side, the reality is I think most people thought that there was no way you would ever be able to achieve a wafer level burn-in. It's really a big deal because these devices are very complicated. If you look at this device as an example, this NVIDIA, and we've actually said it is not NVIDIA. We have not said who it is, but we did take the time to make sure people knew that. But you look at this processor surrounded by HBM memories, and it sort of and then if you say you're testing this at a system level, it becomes intuitively obvious why you might want to burn this device in and weed it out before you take out all of the rest of these devices. I'd say that we had two or three customers in just the last week that are clear-named people in the AI space that are working with us on their next-generation devices to ensure that we have both qualification and production capability to meet those needs. It's very exciting. There are people that simply do not have the capability in-house to address their next generation. They come to us, and we have a roadmap or capability to do it for them. That makes for very exciting times, and it's something that is awesome. At the same time, the work that we're working on to address the memory market is something we had three meetings on in the last week. So it's something else that's really exciting because we know that this is something that should be able to benefit from our multi-wafer systems, particularly now that we have the full automation on the front end of it. Great. Well, thank you. I really appreciate that. And so I just want to give a reminder to everyone in the audience. You can submit a question through the chat or email it to me. So I'll give a few moments if anyone else wants to get some questions in.