Aehr Test Systems - Earnings Call - Q2 2025

Executive Summary

• Q2 FY25 revenue was $13.5M, down 37% year-over-year and modestly up quarter-over-quarter; GAAP EPS was $(0.03) and non-GAAP EPS was $0.02, reflecting soft SiC demand and timing delays of expected system orders into Q3.
• Results missed S&P Global consensus: revenue $15.0M* vs actual $13.5M and EPS $0.03* vs non-GAAP $0.02; bookings were $9.2M and effective backlog rose to $26.6M, mitigating near-term visibility concerns.
• Management reaffirmed FY25 guidance for at least $70M revenue and ≥10% non-GAAP PBT margin, citing acceleration in AI processor burn-in (first wafer-level AI customer and volume Sonoma packaged burn-in orders) and first GaN production order.
• Stock-relevant catalysts: shipment timing from prebuilt AI/GaN systems, multi-market diversification (AI, GaN, HDD, silicon photonics), and China IP litigation risks impacting SiC bookings cadence.

What Went Well and What Went Wrong

What Went Well

• First wafer-level AI processor customer selected FOX-XP for production burn-in; initial orders for multiple systems and WaferPaks secured, a “technological and commercial breakthrough” expanding TAM in AI.
• First volume production Sonoma ultra-high-power packaged burn-in orders from a hyperscaler; shipments commenced to an Asian OSAT, positioning for AI packaged burn-in growth.
• First GaN production order using FOX-XP, broadening beyond SiC with >40% CAGR market outlook and >10% power semi mix by 2028, per Yole/Frost & Sullivan references; management emphasized multi-market diversification.

What Went Wrong

• YoY revenue fell to $13.5M from $21.4M; non-GAAP gross margin contracted to 45.3% vs 51.6% last year, driven by lower revenue and mix, while non-GAAP OpEx rose on Incal and legal costs.
• Timing: prebuilt AI/GaN systems missed quarter-end due to PO finalization after Q2, depressing the quarter; management flagged volatility from high ASP systems and does not provide quarterly guidance.
• China risks: IP suit filed against a local supplier; trade and competitive dynamics heighten risk to contemplated SiC orders and revenue from Chinese customers in FY25.

Transcript

Operator (participant)

Please note, this conference is being recorded. I will now turn the conference over to your host, Jim Byers of PondelWilkinson Investor Relations. You may begin.

Jim Byers (Head of Investor Relations)

Thank you, Operator. Good afternoon and welcome to Aehr Test Systems' second quarter fiscal 2025 financial results conference call. With me on today's call are Aehr Test Systems President and Chief Executive Officer Gayn Erickson and CFO Chris Siu. Before I turn the call over to Gayn and Chris, I'd like to cover a few items. This afternoon, after market close, Aehr Test issued a press release announcing its second quarter fiscal 2025 results. That release is available on the company's website at aehr.com. This call is being broadcast live over the internet for all interested parties, and the webcast will be archived in the investor relations page of the Aehr Test website.

I'd like to remind everyone that on today's call, management will be making forward-looking statements today that are based on current information and estimates and are subject to a number of risks and uncertainties that could cause actual results to differ materially from those in the forward-looking statements. These factors that may cause results to differ materially from those in the forward-looking statements are discussed in the company's most recent periodic and current reports filed with the SEC. These forward-looking statements, including guidance provided during today's call, are only valid as of this date, and Aehr Test Systems undertakes no obligation to update the forward-looking statements. Now, with that, I'd like to turn the conference call over to Gayn Erickson, President and CEO.

Gayn Erickson (President and CEO)

Thanks, Jim. Good afternoon, everyone, and welcome to our second quarter fiscal 2025 earnings conference call. Thanks for joining us today. I actually have a number of topics that I'm going to go through. I'll try and get through them quickly. We've had a lot of inbound requests for information and to clarify all the different markets that we have been expanding into, particularly a lot of questions around the AI side. So I thought I would spend a little bit of time just making sure to bring everybody up to speed and kind of normalize what everybody understands about that. So bear with me as I go through this. Chris will then cover some details related to our financials, a little bit shorter than normal to make up for my longer portion of this thing, and then we'll open up for questions.

I'll start with a brief overview of the quarter's key highlights and share updates on the key markets we're targeting for our semiconductor test and burn-in, including notable progress in those new markets. We're excited to share the significant progress we've made on the key objectives we outlined at the start of the fiscal year, particularly expanding our product reach into additional large and fast-growing markets. Our market diversification into sectors such as artificial intelligence processors, gallium nitride power semiconductors, data storage devices, silicon photonics integrated circuits, and flash memory is driving new opportunities to attract customers and drive revenue growth. This progress includes our wafer-level burn-in solutions and also the success we're achieving with the new semiconductor package part test and burn-in product lines we acquired through our acquisition of Incal Technology that we closed last August.

This acquisition has led to the acceleration of our market diversification, with particular success and leverage expanding our total available market in AI processors. We're pleased to report meaningful progress in our efforts to penetrate the high-power processors market, including artificial intelligence or AI processors, graphics processor units or GPUs, and application-specific integrated circuits or ASICs, AI processors, and accelerators. These advancements include wins in both wafer-level burn-in and package part burn-in applications. Let me start with our new success with the AI processors for wafer-level test and burn-in. Last month, we were thrilled to announce a major milestone for us, securing our first AI processor customer for wafer-level burn-in. After successfully demonstrating the performance and throughput of our systems at our headquarters here in California, this innovative AI customer selected our new high-power Fox XP solution for production wafer-level test and burn-in of their AI processors.

This includes initial volume production orders for multiple high-power Fox XP systems and our proprietary WaferPak contactors, which enable full wafer contact for testing and burning in of AI processors in wafer form before system integration. This achievement represents a significant technological and commercial breakthrough for Aehr, significantly expanding the market potential for our Fox XP wafer-level test and burn-in systems. AI processor burn-in is a significant opportunity for Aehr. AI processors have tens of billions to even greater than 200 billion transistors on a single semiconductor die. Transistors act as the fundamental building block, functioning as an electronic switch to control the flow of electrical current, essentially representing the on and off states necessary for processing information on digital circuits, allowing for the creation of complex logic operations within a single chip by combining numerous transistors together.

Large complex processors have always been subject to the challenge of early failure rates that are higher than the application requires. Companies like Intel, AMD, and NVIDIA have always had many of their products, and for some companies, all of their processors require a production burn-in for screening out early failures. This is also the case for AI processors and ASICs. Companies use test and burn-in systems to qualify a new process and AI processor design to determine the device's early failure rates, intrinsic and steady-state failure rate over time, and also how long the devices will last before they begin to fail due to the end-of-life wearouts. This is really where our acquisition of Incal plays a huge role in our strategy for AI processors, as the Sonoma ultra-high power package part test and burn-in system is extremely good for this qualification application.

I'll also talk about Sonoma in more detail later. During the semiconductor qualification process, customers characterize the failure rates over time for a given device and then implement a production burn-in to effectively accelerate and induce the early failures so that these devices can be screened out and not shipped to customers. In the case of wafer-level burn-in and screening, they can remove the devices before they're packaged with other processors and high-bandwidth DRAM memory or HBM into very expensive packages using substrates like CoWoS packaging from TSMC.

Once you understand that AI processors need production burn-in, and then you look at devices such as Hopper or Blackwell from NVIDIA or the AMD Instinct MI325X GPU accelerator, it becomes intuitively obvious why customers would like to move this production burn-in step to wafer level if they can, to avoid throwing away very expensive packages and the remaining good processors and HBM.

With Aehr providing the first-ever solution for wafer-level burn-in in an AI processor in partnership with this customer and their outsourced assembly and test or OSAT partner, we have shown that our high-power Fox XP multi-wafer systems and proprietary WaferPak contactors can be a viable solution for testing and burn-in of AI processors in wafer form to avoid having to burn in these devices in package or system form, where the cost of test and the cost of yield loss due to failing devices during burn-in is much more expensive and impactful to overall manufacturing yield. We've already shown the solution working at our facility in California and will begin shipping the production systems and WaferPaks already this month.

The systems will have a very visible footprint at the OSAT they will be installed and used at, and we're working together on marketing this wafer-level production burn-in capability to other AI processor suppliers in partnership with this OSAT as they speak. Now, let me turn to the package part test and burn-in level of AI processors using our new Sonoma system. During the second quarter, we secured our first production AI processor customer for package part burn-in, receiving initial volume production orders for multiple Sonoma ultra-high power systems. This customer is a large-scale data center hyperscaler, providing computing power and storage capacity to millions of individuals and organizations worldwide. System shipments have already commenced to their OSAT contract manufacturer, doing test and burn-in for them in Asia. Again, this is for production burn-in screening of all devices shipped to screen out early failures.

This customer was and continues to use the Sonoma system for process and device qualification, as I detailed a few minutes ago. The Sonoma system is a system that combines high parallelism of ultra-high power devices under test or DUTs, with very high current, high-performance power supplies that allow many voltage zones ranging from multiple core power zones to many other lower power zones on the same device, very flexible digital test resources per device, individually controlled liquid cooling of each device under test, and what we feel is the world's best software and user interface for device testing and debug for device qualification and reliability testing and bring-up.

I have personally heard from multiple customers that they often use the Sonoma system to debug the test patterns of their devices and use this data to feed back to their multi-million dollar ATE systems from the likes of Advantest and Teradyne. We have a large and growing installed base of Sonoma systems around the world at test labs, OSATs, and IDMs used for this qualification and reliability testing of new devices and semiconductor process nodes. What is new is this is the first customer, and it's a large one, that is purchasing Sonoma for their volume production needs for test and burn-in of their AI processors.

This is an area where the combination of Incal and Aehr was very well received by customers, as Incal simply did not have the support infrastructure worldwide as Aehr does to support the needs of production customers, nor did they have the manufacturing capacity that we do to build a large number of systems for production customers. We see significant potential to expand our package part test and burn-in business in the rapidly growing AI semiconductor market with the ultra-high power Sonoma product line, and already have a team working on multiple enhancements to address both qualification and production test and burn-in needs of a wide range of AI processor suppliers, test labs, and OSATs.

Aehr is now able to offer our customers both the ability to do production wafer-level burn-in of their AI processors and accelerators, as well as reliability qualification and production burn-in at the package part level with our Sonoma system. Aehr Test Systems is the only company in the world offering the flexibility to customers of both wafer-level and package-level test and burn-in solutions for AI processors. We estimate that the combined market for wafer-level and package part reliability test and production burn-in of AI processors will surpass $100 million annually in the future. The AI semiconductor processor market is growing very fast. According to the SNS Insider, the AI chip market size was valued at over $60 billion in 2023 and is expected to grow to over $600 billion by 2032, a 10x increase and a CAGR of almost 30% over that period.

The types of processors and applications for AI processors or processors with AI focus are also expanding rapidly from the historical general-purpose GPU-based processors used for LLM generation and inference to application-specific integrated circuits aimed at specific accelerator functions in data centers and hyperscalers and edge AI processors used in autonomous vehicles, robotics, and security. These devices and their applications demand extremely high quality, reliability, and security. With our comprehensive product portfolio for reliability test and burn-in of AI semiconductors, we feel Aehr is poised to capture a meaningful share of this market. Now, let me talk about the expansion into our gallium nitride market. Last week, we're excited to announce another exciting milestone with our first gallium nitride or GaN semiconductor production order. This customer is a leading automotive semiconductor supplier and a key player in the gallium nitride power semiconductor market.

We're thrilled to receive their initial production purchase order, marking their commitment to advancing volume production wafer-level burn-in of their GaN devices using our Fox XP platform. This achievement expands our production wafer-level burn-in market for power semiconductors beyond silicon carbide applications used in electric vehicles, data center power conversion, and solar to now include GaN, a high-performance compound semiconductor optimized for mid-power applications such as data centers, solar energy, automotive systems, and consumer electronics and PCs. Over the past 12 months, we've collaborated with this lead customer using our Fox NP system, leading to their purchase of multiple WaferPak reference designs on a diverse set of GaN applications. GaN offers a much broader application range than silicon carbide and is poised for significant growth in the coming decade.

While the largest market segment for silicon carbide, about 70%, is for the electric vehicle and EV charging infrastructure markets, GaN is very diversified and is not dominated by EVs or autos. With many more end-use applications, there are many more customers and broader market for GaN semiconductor suppliers than for SiC, where there are fewer EV and charging customers, but are at higher volume per customer than the average customer for GaN suppliers. With an expected compound annual growth rate CAGR exceeding 40%, the GaN market is projected to surpass $2 billion in annual device sales by 2029, according to Yole Group. Additionally, Frost & Sullivan estimates GaN semiconductors will account for over 10% of the worldwide power semiconductor as soon as 2028.

This transformative technology represents a significant growth opportunity for Aehr's wafer-level test systems and WaferPak full wafer contactor, positioning us to capitalize on the expansion of the GaN market. My next topic is silicon photonics, which continues to show signs of market adoption for chip-to-chip communication and for optical networking switching. We remain very enthusiastic about the silicon photonics market, particularly for the new application of silicon photonics integrated circuits for use in optical chip-to-chip communication that we see as a significant market opportunity for our products. Last year, we shipped a new high-power configuration of our Fox XP system to our lead customer for a new family of silicon photonics integrated circuits aimed at optical chip-to-chip communication.

This system, with our proprietary WaferPak full wafer contactors, allows for testing over 8,000 high-power optical devices in parallel on each of nine wafers before they're singulated and placed into a fiber optic transceiver for data center and telecommunications infrastructure or for placement in co-package optics for optical chip-to-chip communication. Multiple companies, including AMD, NVIDIA, Intel, TSMC, and GlobalFoundries, have announced product roadmaps for devices using optical chip-to-chip communication, with new announcements in just the last week coming out of Taiwan on new silicon photonics-based ICs by TSMC and NVIDIA. The new Fox XP system configuration with higher power WaferPaks enables production tests of up to 3,500 watts of power per wafer and up to nine full wafers in parallel.

It also includes Aehr Test's latest chamber configuration, which has a smaller overall footprint and is compatible with Aehr's new WaferPak Auto Aligner that provides Aehr customers with fully automated material handling, enhanced reoperation of six- to 12-inch wafers using industry-standard wafer cassettes and FOUPs, and can also support mobile robot and overhead transfers of wafers in those FOUPs. This is the system that we leverage for the production system for wafer-level burn-in of AI processors as well. We've designed a number of new WaferPaks for engineering and initial samples for this lead customer of our high-power silicon photonics solution this fiscal year. These are very impressive devices with significant improvements in size, power, and data rates that we feel are likely to ramp to production in time to address the chip-to-chip optical communications market opportunity.

The hard disk drive market is another opportunity for Aehr's wafer-level test and burn-in systems, and we're excited about our opportunity for production burn-in and stabilization of devices used in hard disk drives using our Fox CP systems and WaferPak contactors. Our lead customer for this application is ramping this year and has told us they will purchase multiple production systems from us over the next few quarters to support their planned production rollout and ramp. This customer, first announced back in 2019 prior to the COVID-19 pandemic, initially purchased our Fox CP single wafer test and burn-in solution to support the qualification and early test stages of this new product aimed at the enterprise and data center markets. We view the data storage market both for hard disk drives and flash-based semiconductor solid-state disk drives as significant growth opportunities for our systems.

These markets have applications with devices made up of multiple die and complex structures or multiple die stacked on top of each other before they are put into higher-level packages or systems. These devices require exceptionally high levels of quality and long-term reliability of the die before they are put into the packages or systems, which aligns perfectly with the capabilities of our wafer-level test and burn-in systems. In speaking of solid-state disk drives and NAND Flash memory, we're making steady progress on our ongoing benchmarking project with a major flash memory supplier to evaluate the benefits of using our Fox XP solution for wafer-level test and burn-in other flash memory devices. This application is for 100% test and burn-in of devices to be used in mission-critical applications such as enterprise storage.

As part of this evaluation, we're advancing the development of a low-cost, high-pin count, fine-pitch MEMS-based WaferPak for full wafer contact of all NAND devices on a 300 millimeter wafer, including support for high-density 3D NAND technology supporting up to greater than 200 layers. One of the key challenges with addressing the new devices on customers' NAND roadmap is that not only is there a significantly higher number of die per wafer, but the power per die and therefore power per wafer to test these wafers has increased significantly. Support for high-power wafer testing is something that Aehr is particularly good at with our Fox XP multi-wafer test and burn-in systems and WaferPaks. This new WaferPak design is also capable to support DRAM testing should customers choose to pursue DRAM burn-in in the future.

We aim to complete the proof of concept phase in a few months, enabling us to advance this benchmark to either a production solution evaluation or secure a commitment from the customer to develop a production test cell. This would position us to generate our first revenue from this NAND opportunity as early as our next fiscal year. We believe this represents the leading edge of a significant opportunity for our solutions for semiconductor memories, with the NAND flash market being the key initial focus. Looking ahead, we see long-term potential to expand the DRAM wafer-level test and burn-in, further broadening our market reach. So you'll notice that I led with the new market opportunities, but will now provide an update on silicon carbide wafer-level test and burn-in, a market that in fact was almost 90% of Aehr's revenue last year.

Aehr also continues to expand its presence in silicon carbide power semiconductor market, a critical sector for power conversion for electric vehicle traction inverters, charging infrastructure, and a range of industrial data center and infrastructure applications. Based on recent market forecasts and large suppliers of silicon carbide semiconductors, growth in silicon carbide sales outside of China should remain challenging before recovering in calendar 2026. We believe we're well-positioned in this market as we have a large customer base and are currently engaged in benchmarking efforts with multiple potential new silicon carbide customers around the globe, including in China. While we remain cautiously optimistic about the opportunities in China, we also recognize the geopolitical, trade, and intellectual property risks associated with this market. Recently, we filed a lawsuit in China against a local supplier for intellectual property infringement.

This action relates to features of products by that company targeted at wafer-level burn-in and silicon carbide devices that we believe infringe on Aehr's intellectual property and patents granted to Aehr by the Chinese Patent Office. Our current fiscal year forecast includes contemplated orders and revenue yet to be booked for silicon carbide wafer-level burn-in systems and WaferPaks destined for silicon carbide manufacturers in China. It is important to bring this to our shareholders' attention as recent trade-related developments in the U.S. and the emergence of competitive offerings in China that we believe infringe on our intellectual property have heightened the risk associated with bookings and revenue from Chinese customers.

As we look at the composition of our total revenue for this fiscal year, silicon carbide is expected to account for less than half of our total revenue as we've seen our expansion into additional other markets capture real market share gains. AI processors, including wafer-level and package parts, could comprise as much as 40% of our total revenue this fiscal year, up from effectively zero revenue last year. GaN, hard disk drives, silicon photonics integrated circuits, and other semiconductor package part revenues will comprise about another 20% of revenue. We're not pivoting away from silicon carbide, but rather are generating what we see are the growth in the other market opportunities while not seeing the growth in silicon carbide this year like we saw last year.

According to recent market research from companies such as Yole, the estimated revenue for silicon carbide semiconductors in 2024 was around $2.5 billion and expected to reach $10 billion by the end of the decade, a 4X increase. To put this into perspective, the semiconductor market is projected to grow from about $600 billion overall in 2024 to over $1 trillion by the end of this decade. So silicon carbide will be about 1% of the overall semiconductor market by 2030. Aehr's innovative solutions are poised to capitalize on this growth in the overall semiconductor market by addressing the critical reliability needs of next-generation applications and leveraging key megatrends shaping the semiconductor industry. Reliability has become a critical priority across a wide range of industries, including combustion and electric vehicles, data centers, electrification of the world's infrastructure, and a wide range of artificial intelligence applications.

Factors such as smaller semiconductor geometries, the increase in adoption of compound and optical semiconductors, and the complexities of ensuring semiconductor reliability on ever-increasing power and performance of semiconductors and advanced packaging are driving the demand for wafer-level and package part test and burn-in systems. Aehr solutions are instrumental in reducing early operational failures and ensuring long-term device performance in these rapidly advancing markets. With strong customer engagements, expanding market opportunities, and innovative products designed to meet evolving demands, we're optimistic as we move into the second half of our fiscal year and maintain our previously stated financial guidance for the fiscal year.

As we've stated before, though, given the nature of our business with our high average selling prices of a single production system and a set of WaferPaks, our quarterly revenue can experience significant variability if system orders anticipated by the quarter end are delayed by even a few days. This was the case in last quarter, and one of the key reasons we do not provide quarterly guidance. In the case of both our new GaN and wafer-level AI customers, both requested us to pre-build systems that we fully expected to ship to them within the quarter. However, the purchase orders were not finalized until after the quarter ended.

Looking ahead and above the quarter-to-quarter variations, we're excited about the current and emerging market opportunities for our products, which not only position us for a successful fiscal year, but also lay a solid foundation for long-term sustainable growth in years ahead. Lastly, and before I turn it over to Chris, it's with great sadness that we acknowledge the passing of Avi Ray-Chaudhuri, our EVP of research and development, who lost his battle with cancer last month. On behalf of everyone at Aehr Test, we extend our deepest sympathies to his family. Avi's friendship, leadership, and the tremendous contribution to Aehr Test will always be remembered and cherished. It was an honor to work alongside Avi, and he'll be deeply missed.

In the interim, Don Richmond, our CTO, who previously held this role before Avi joined us a year and a half ago, had stepped in and assumed Avi's responsibilities at Aehr and would continue in this capacity until further notice. With that, let me turn it over to Chris, and then we'll open up the lines for questions.

Chris Siu (CFO)

Thank you, Gayn. Good afternoon, everyone. The company recognized bookings of $9.2 million in the second quarter of fiscal 2025, compared to $16.7 million in the first quarter of fiscal 2025. At the end of the quarter, our backlog was $12.4 million. In the first six weeks of the third quarter of fiscal 2025, we received $14.2 million in additional bookings. This growth was driven primarily by the first AI processor customer, utilizing our high-power Fox XP solution for wafer-level production test and burn-in of AI processors, which we announced in December.

With this recent bookings, our effective backlog has now reached $26.6 million. Turning to our Q2 performance, which included a full quarter of the financial results from the Incal acquisition, we faced a challenging environment due to overall softness in the silicon carbide power semiconductor market. Second quarter revenue totaled $13.5 million, a 37% decline compared to $21.4 million in Q2 last year. A significant portion of this revenue was driven by demand for our WaferPaks and Sonoma ultra-high-power systems acquired from the Incal acquisition, which support high-volume production test and burn-in of AI processors. WaferPak revenues came in at $8.6 million, accounting for 64% of our total revenue in the second quarter, an increase from 43% in the same period last year. This highlights the important role of our WaferPaks as a key source of recurring revenue for our business.

Additionally, system sales from our Sonoma and Tahoe package part burn-in products made a substantial contribution to our second quarter revenue. We are excited to see the significant progress we made to incorporate the Incal products into our product portfolio to address the AI market opportunities. We believe our strategy to expand Aehr's product offerings to diversify into sectors beyond silicon carbide applications, such as AI, gallium nitride power conversion, and other sectors, will drive revenue growth in the future. Non-GAAP gross margin for the second quarter was 45.3% compared to 51.6% in the same period last year. The decline was primarily due to a lower overall revenue level compared to Q2 last year, partially offset by a favorable product mix of higher margin WaferPaks. Non-GAAP operating expenses in the second quarter were $5.9 million, a 19% increase from $5 million in Q2 last year.

The year-over-year increase is primarily due to incorporating a full quarter of Incal's operating expenses into our financial results, as well as high legal and professional services fees. We expect to incur higher legal fees in the next few quarters as we vigorously protect our intellectual property rights in China and defend the class action and derivative complaints in the United States. In Q2, we recorded income tax benefits totaling $217,000. Non-GAAP net income for the second quarter, which excludes the impact of stock-based compensation, acquisition-related cost, the acquisition-related fair value adjustment to inventory, and the amortization of intangible assets, was $0.7 million, or $0.02 per diluted share for the second quarter. This compares to non-GAAP net income of $6.7 million, or $0.23 per diluted share in the second quarter of fiscal 2024. Now turning to the balance sheet.

At the end of Q2, our cash, cash equivalents, and restricted cash totaled $35.2 million, down from $40.8 million at the end of Q1. During the quarter, we used $5.9 million in operating cash flows, primarily to pay our suppliers and service providers. With a strong balance sheet, we continue to invest in scaling our business and pursuing new market opportunities. We have no debt and continue to allocate excess cash to money market funds. In the second quarter, we earned $228,000 in interest income. On October 15, 2024, we filed a new S-3 registration statement with the Securities and Exchange Commission to support potential future financing needs. The new S-3 shelf filing for $100 million was approved by the SEC on October 25, 2024 and is good for three years.

Looking ahead to the remainder of fiscal 2025, which ends on May 30, 2025, we are reaffirming our previously provided guidance. As Gayn noted, we expect total revenue of at least $70 million with a non-GAAP net profit before taxes of no less than 10% of revenue. Lastly, looking at the investor relations calendar, Aehr Test will be participating virtually in the Needham Growth Conference on Thursday, January 16, 2025. We look forward to connecting with many of you during this event. This concludes our prepared remarks. We're now ready to take your questions. Operator, please go ahead.

Operator (participant)

Thank you. At this time, we will be conducting a question-and-answer session. If you would like to ask a question, please press star one on your telephone keypad. A confirmation tone will indicate your line is in the question queue.

You may press star two if you'd like to remove your question from the queue. For participants using speaker equipment, it may be necessary to pick up your handset before pressing the star keys. One moment, please, while we poll for questions. Once again, please press star one if you have a question or a comment. The first question comes from Christian Schwab with Craig-Hallum. Please proceed.

Christian Schwab (Analyst)

Hey, good afternoon, guys. So Gayn, thank you for all the detail on all the different market opportunities. I guess the question really only has one question. It's a question we get all the time. We've been at $65-$70 million here for three years. What year is kind of the inflection point of long-term sustainable growth? You kind of ended with a summary of all that. You felt confident in long-term sustainable growth.

I'm just wondering, do we start seeing strong top-line growth rate next fiscal year, fiscal year 2026 and beyond? And if we do, what does that top-line growth rate look like?

That's it. Thank you, Gayn.

Gayn Erickson (President and CEO)

All right, Christian. Well, as you know, and maybe not everyone else knows as well or understands, we have yet to give multi-year forecasts, which I know kind of feel a little less than what you were asking for. But let me at least try and put it in perspective as well.

We had been, if you go back and look at multiple quarters, we've been talking about some of these other markets and alluding or being very direct about some of the expansion, whether it be us talking about the investments in the silicon photonics, our initial engagement with customers in GaN a year and a half ago that said we were evaluating them for production use in hopes that it would prove out that they need production burn-in, which we've now determined have engagements with the flash memory, and we were a little more guarded about it, but about a year ago, if you look at the notes, I was referring to another exciting application for a high-power application for wafer-level burn-in we were engaged in, which was actually the leading edge of the AI stuff.

Those were markets that we saw driving our revenue, but candidly, most of the energy was, of course, about the silicon carbide and EV. We saw those markets easily driving for, as we talked about AI, these could be $100 million a year numbers. And flash memory is probably the largest among it, and DRAM larger than that. There are enormous opportunities in the space kind of as a whole in semiconductors. But of course, we were seeing this year-over-year growth of 40%-50% for silicon carbide, and that seemed to be where the focus was. So I've always tried to balance, particularly with our customers that still really need us for their silicon carbide ramps and their fabs they're putting in. We're not abandoning the silicon carbide. I think we have the most competitive solution, the best roadmap as well to continue to meet customers' needs.

We'll see that growth. We think that silicon carbide still has a lot of legs on it as we still anticipate 30% of electric vehicles by the end of the decade of total vehicles being electric vehicles. It's just, by contrast to all the craziness that was going on a year and a half ago, it seems quite more disappointing. I think silicon carbide has a chance to grow next year from this year. I think the other markets for sure have a chance to grow year over year. We're still getting our arms around it and doing some introductions as we're taking the Incal package part burn-in production system and marketing it around the world, and at the same time, having conversations with customers about the wafer level. We'll hopefully be able to give you more insight as to what that looks like.

And of course, we'll provide our annual guidance in our July release, which will summarize the fiscal year and as we head into fiscal year 2026.

Christian Schwab (Analyst)

Great. Thank you for that, Gayn. No other questions. Thank you.

Gayn Erickson (President and CEO)

Thanks, Christian.

Operator (participant)

Up next, we have Jed Dorsheimer with William Blair. Please proceed.

Jed Dorsheimer (Managing Director and Group Head of Energy and Power Technologies)

Hi. Yeah. Thanks for taking my questions. Gayn, despite the miss, backlog looks pretty strong. So I'm just wondering, were there any tools on the shipping dock? Is this more of a timing issue? I'm guessing that it is, but I'm just curious if you could add any more color, and then I have some follow-up questions.

Gayn Erickson (President and CEO)

Yeah, they were on the shipping dock, and very much so. Yeah, we were quite frustrated. I don't want to get into it.

This isn't the first nor the last time we had, I'll just say, one or more customers, because kind of narrowed down to two, asking for some concessions and things that just seemed quite unreasonable, recognizing it was the end of our fiscal quarter and we dug our feet in at the same time but it's very interesting that here, several weeks after the quarter or even right afterwards, and now it's like, "Hey, I need you to ship ASAP" so we're still working on that expectation management with our customers around that but yeah, they're both orders for immediate shipments because we had pre-built them into it and had we had either one of those orders ahead of time, our quarterly number would have been significantly higher but it just moved into this quarter.

Jed Dorsheimer (Managing Director and Group Head of Energy and Power Technologies)

Got it, and thanks for the color.

Maybe to Christian's previous question, just but framed a little bit differently. If I kind of read the tea leaves in terms of all of the market details that you provided at the beginning, it sounds like the hard disk drive, I think, for the HAMR process product or platform, that's the biggest near-term driver of the business beyond silicon carbide, but you've got a lot of different irons in the fire around AI and even HBM. So am I misunderstanding? I'm just trying to understand sort of near-term drivers that then potentially lead to inflection on future growth.

Gayn Erickson (President and CEO)

So we've never given a lot of clarity on what the hard disk drive application is. I'll just leave it at that. Not confirming or denying, okay? But realistically, if you look at our so if you go through the numbers of our just by the numbers, we did what?

26 something. We have 26 something in backlog. We have 75% almost to the dime of revenue scored and/or backlog of the $70 million. So we have like 16, 17 million to go, I think, whatever the math is. We've identified what those customer opportunities are. There's upside and downside to kind of each one of those. Some of them could push out, etc. So it's always a challenge to try and put a number around it. You guys think this is easy. It's not with the granularity of our stuff. But near term, we've got some silicon carbide orders in our expectation for Q4 with some new customers. We've got more AI production, both in package part and wafer-level burn-in. I think both of those, all three of those, are probably bigger than the hard disk drive one in the near term.

Over the next several years, as they ramp out, the hard disk drive business looks to be really attractive. And there's some variation on some of the stuff we can't get into with respect to their forecast, but we're pretty happy with that. And they're very happy with us about that application. Flash, for sure, is kind of bigger than everything, although maybe the AI stuff will end up being bigger than flash. We'll see how it plays out. But that's a year out. And we're not alluding to a timing of when the DRAM stuff comes. But it's a pretty full plate. You guys, we're still. You hear it all the time. We're still doing benchmarks. I mean, Vernon and I will tell you, we've won customers that haven't placed orders yet for silicon carbide.

It's a very interesting market with all the dynamics that are going on in there, winners and losers and market share shifts between customers, between the customer's customers, different model years that we have a lot of insight into, and the implications of both China and also just cars going to modules versus discrete devices. There's just a lot of moving parts. But I try and use this 80/20 rule. Vernon and I talk about this all the time. Vernon is our VP of sales for people that don't know. We can't wake up every day and spend 80% of our time on silicon carbide. We've spent a ton of energy. We've got these great solutions. We've won these benchmarks. The customers need us to be there.

But most of the days we wake up, we're focusing on these other market opportunities where we can really have an impact. We'll let Adil, our VP of operations or COO, deal with manufacturing capacity. Just take the orders and ship them. Because we certainly bought the inventory against what we really believed a forecast was going to be there last year. Bought the inventory against those forecasts from verbal forecasts from customers that have pushed out, we still think are going to come.

Jed Dorsheimer (Managing Director and Group Head of Energy and Power Technologies)

Got it. One last question for me, and then I'll jump back in the queue. Maybe it's for Chris here. Just margins, gross margin took a steep decline. I'm assuming that that's mixed shift from the WaferPaks to the Incal systems, Sonoma. But I'm just curious, do you expect this to be at these levels?

You mentioned in backlog that you have some silicon carbide. I'm assuming some WaferPaks there. How should we think about the normal margin levels? Certainly, to achieve your $70 million would imply getting a pretty big bump in revenue. So how should we think about that for the rest of the year?

Gayn Erickson (President and CEO)

Yeah. So you're correct, Jed. So as we talked about it before, the Incal products, the gross margins is a little bit lower than what we have in legacy Aehr Test Systems and WaferPaks. So the order you can think about it is the WaferPaks always has the best gross margin. And then the Aehr Test Systems, and then would be the Incal system. So it really depends on the product mix even within Incal because

Chris Siu (CFO)

the Incal consumables have lower too.

Gayn Erickson (President and CEO)

Yeah, consumables have lower too. The other piece we haven't given a ton of, I think we may have said in early discussions around the Incal acquisition. So we're at a good point right now. We, of course, do all these live. And you don't hear the construction in the background, but we're doing our remodel here, adding a huge amount of capacity, capability out on the floor, clean room space, etc., before we move the Incal products over. They're only a few miles down the road. But that whole team and the manufacturing will be in here, we think completed by the end of our fiscal year. So it's only a few months out. That facility is also burdening our cost of sales too. So that comes off our books by a year and a half. That's about $500,000 a year, something like that too.

So we'll get some goodness out of this thing. We don't really focus our energy on the goodness of the Incal merger. Mostly, it's the revenue opportunities and the customers. But there will be some operational efficiencies as well.

Jed Dorsheimer (Managing Director and Group Head of Energy and Power Technologies)

Great. Thanks for the color. I'll jump back in the queue. Thanks, guys.

Gayn Erickson (President and CEO)

Thank you, Jed.

Operator (participant)

Once again, if you have a question or a comment, please indicate so by pressing star one on your touch-tone phone. The next question comes from Tom Diffely with DA Davidson. Please proceed.

Tom Diffely (Analyst)

Yeah. Good afternoon, and thanks for the question. Gayn, curious what your view is of the Chinese market for silicon carbide if you're not successful in your patent infringement case. Is there still a market there for you?

Gayn Erickson (President and CEO)

That's a good question. Yes, I guess. I mean, there's kind of two things.

One of them is the patent infringement case against these guys, and the other one is our competitiveness, and obviously, the trade-related things between the U.S. and China, particularly kind of recently, those are harder to get your finger on. The system, and I want to be careful too much, that system that we've named in our 10-Q, etc., it's not very good. I know that's kind of pretty directed. We've had feedback from customers that it's not working very well. It has repeatability issues. It can't really do what we do. We have parallel advantage. We have a number of wafers. We have automation. We have a lot of capability, but nevertheless, we're not just going to put up with people trying to encroach on us and step on patents that we have in China provided by the Chinese patent office.

We're not going to put up with it. There's two things. There's a legal aspect of this thing, and then there's just the flat-out competitiveness. We have a team of people in China. I think we've already talked about it. I think we had 14 customers in China over time. One of them is a WaferPak-based package part from Aehr Test. We've done business in China for a long time. It's not our plan to change. There's just some of the uncertainties related to U.S. trade things that make it a little harder to put your finger on. We're just trying to be open with that with people.

Tom Diffely (Analyst)

Okay.

So maybe to summarize, if worst-case scenario, they're able to get away with whatever they've copied on their tool, you feel like you still have a pretty strong competitive advantage for both reliability and productivity versus their system as it is?

Gayn Erickson (President and CEO)

We do. We do. You always have to be careful because people have roadmaps, and we have roadmaps too. I don't want to get too carried away with that. But yeah, that is the case.

And again, it's very focused on silicon carbide. It has issues even testing gallium nitride. It doesn't do power semiconductors. So it's kind of a specific niche target at something. But nevertheless, we're taking it seriously, at least in China.

Tom Diffely (Analyst)

Okay. And then your concern over the orders this year is the concern that you lose them competitively or that they get pushed out?

Gayn Erickson (President and CEO)

Well, there's always I worry about a lot of stuff, Tom. But in general, when we were trying to describe China, we have a production system in our plans, at least in those numbers. We also have other opportunities and things like that too. But an ASP of our wafer-level burn-in systems, which uses a WaferPak, is several million dollars. And so if something happens and there's some regulatory thing that slows us down or precludes us or something, I don't know really how to handicap that. That's a new one for me. There's always the, "They still buy us, but they don't take it by May." That's kind of the normal thing we've been dealing with. I mean, I think a lot of people understand. I mean, a lot of our shareholders are experts in semiconductor tech.

I mean, semiconductors are in a downturn right now. I mean, outside of AI, there's a lot of it's tough out there in automotive customers, etc., and so as those of us that have been here for a long time, this too shall pass, and if you look at it in the scheme of a couple-of-year window, it always gets better, so sometimes we wake up and things have been pushed out. You get a call from your customer and saying, "Oh, can you help us with pricing because we're struggling?" And you're like, "You are 50 times larger than I am. I'm not sure I can help you with your numbers, but we're here for you," kind of thing, so there's just some of that in the background.

Tom Diffely (Analyst)

Yeah. Okay, and maybe one last question on the flash market.

When you look at the opportunity for you specifically, is that going to be driven by a technology change either in the chip or the package? Or what do you think really starts that market for you?

Gayn Erickson (President and CEO)

Yeah. I mean, flash memory is actually, Tom, as you know from my history in HP/Agilent era running the memory business there for many, many years. People refer to it as a treadmill. You get on it, and you always have to keep going because the customers every two, three years are shipping more bits at the same price. So they have to always look at ways to be more and more cost-effective. That's true in their process equipment, their front-end equipment, and their back-end equipment and test equipment.

So there are things specifically on the NAND flash roadmap that, in addition to just cost disconnects, meaning I've got to ship way more flash memory in four years than I am today, and my revenue is not going to be up 4X, how do I do it more cost-effectively? So that's sort of a commercial requirement. There's also technical issues. And I was pretty specific in there to be clear. These NAND devices are 3D stacked. They're not actually stacked. They're just printed several. I mean, NAND guys are out there, but they're all talking about 200-layer NAND. What that ends up being is you end up having way more power per device to think of like a 200-story building. And the amount of light, when you light up the light on the first floor, that takes certain power.

But when you have light on 200 floors, you have 200 floors of lights on. When you go to test, there are some test methodologies whereby you can do things by testing multiple floors at a time. But when you do that, all the energy is used as if you're lighting up all those floors. Whereas in normal application, you're reading and writing data only on one floor at a time. So the actual device in its application is using one 200th of the power. But you could do test methodologies to test 20 floors at a time, but then you need 20 times the power. That's a thermal problem. That's a test or resource problem. Those are things we're really good at. So I'm trying to give you some hints there as to what's going on.

So there's some technical disconnects that we are looking to address as well as the commercial side on the NAND side. And by the way, DRAM, I'll go into that too. Just the whole HBM is super dynamic. And there's so many things going on with sector because DRAM also needs to be burned in. So how are you doing that again before you put it into an HBM stack onto a CoWoS package along with an AI processor? So I like where we're at right now. There's a lot of vectors pointing towards this right now.

Tom Diffely (Analyst)

Great. Appreciate the color. Thanks, Gayn.

Gayn Erickson (President and CEO)

Thanks, Tom.

Tom Diffely (Analyst)

Yep.

Operator (participant)

Once again, if there are any remaining questions, please press star one on your touch-tone phone. The next question comes from Larry Chlebina with Chlebina Capital. Please proceed.

Larry Chlebina (Founder)

Hi, Gayn.

I got a question on your recent $10 million win on the GPUs for the accelerator. I know you said that the potential market with the package part would be over $100 million a year down the road. But I'm looking at the wafer level portion of it. You said that that customer was not NVIDIA, not the big guy. It has over 90% market share. So assuming it's the second largest, which is, I think, around 5%, and the $10 million represented two machines, that would be like 30, possibly 40 XPs for just doing the GPUs. Is that the right way to look at that?

Gayn Erickson (President and CEO)

Okay. So I don't want to help you with who the customer might be. So anyone listening, I'm ignoring you on that one, Larry. Yeah, the numbers are really big.

You don't have to try very hard to fall out of bed and hit a $100 million number, and we're not trying to pooh-pooh that, but this gets into available, addressable, timing of this. It could be a lot bigger than that.

Larry Chlebina (Founder)

Right. Well, that's not really what I'm getting at. I'm actually getting at the next component on the accelerator, the HBM, the high-bandwidth memory. The current 12-stack, 8, 12 stacks, if you do the math on the silicon involved, it's about six times the silicon in the GPUs currently, and they're talking about going to 16-high stacks later this year with a roadmap to get to 24-high stacks next year, which would be 12 times the silicon.

My point is, with your fine-pitch WaferPak, it just seems like you ought to have a very interested bunch of potential customers wanting to see if you could help them on their yield issues that, as they stack higher and higher, you could address that and help them out.

Gayn Erickson (President and CEO)

Yeah. When we talked a couple of years ago and imagined DRAM and what was going on, I would have and certainly did say that we thought that was something probably more towards the end of the decade because companies just weren't that motivated to try and figure out how to do DRAM wafer-level burn-in compared to what they were doing. And maybe it would take something else. And now you look at AI, which most of us weren't even talking about two years ago, right? And you look at these, and it's like, "Wow.

Are you motivated now?" So I agree with you in terms of the motivation to put in the DFT and local test modes and to try and address this issue.

Larry Chlebina (Founder)

I think part of the reason why HBM is more expensive than the GPUs is probably their terrible yields that they're experiencing. And hopefully, you can help them out on that.

Gayn Erickson (President and CEO)

Well, that's on each stack, if you will, or each individual die. And then just I always have to remind people, the reason you do production burn-in is because things fail. It's not to make you feel good. As soon as they stop failing, you would stop doing it. So when I state, "Oh, people are doing a production burn-in of these devices," you have to recognize that means there is a material number of failures that they do not want to ship to their customer.

Every one of those takes out all those HBM stacks in the CoWoS substrate plus the package. So it's our whole premise about things moving to multi-chip modules that we've been touting for the last five, six, seven years with kind of the leading edge of what's going on with wafer level. At that time, we were struggling. People were like, "Okay, which devices is that going to look like?" And we kept saying, "They're coming." And because of Moore's Law is failing, or it's failed, or it's fallen down, you now need to put multiple devices together, and you need to stack them up and put them side by side in order to get the functionality that used to just be done with a shrink of a wafer. So yeah, more vectors heading our way, Larry.

Larry Chlebina (Founder)

Assuming this comes to pass, would you be able to handle all that demand through the capacity of potential capacity in Fremont? Is that even possible, or would you have to pull?

Gayn Erickson (President and CEO)

For those that haven't attended us, especially with our new facility upgrades that we're doing right here, we have. I mean, the manufacturing capacity out here is bigger than I had at Verigy. So people might be shocked at that. So there's always ways of doing it. Keep in mind, we use subcontract manufacturers. In most cases, we have multiple suppliers of every subsystem. In many cases, one of the hard things is your printed circuit boards and your assemblies. That could be increased a hundredfold if you need it without actually impacting our resources.

So even though we do assembly tests here, what we do is just a small piece to control the final assembly and quality before it ships. And we have some other tricks up our sleeve. So right now, I have more capacity than demand by a lot. And we could increase our capacity significantly if we needed to for different market opportunities. And when we bring customers through, by the way, that is very obvious when they come in and look at our facility. They kind of go, "Wow, okay. This is not your typical burn-in company in terms of the capacity we have."

Larry Chlebina (Founder)

Well, you got great opportunities. I'll be looking forward to seeing you accomplish them. Good luck.

Gayn Erickson (President and CEO)

Thank you, Larry.

Chris Siu (CFO)

Thank you.

Larry Chlebina (Founder)

Thank you.

Operator (participant)

Okay. I'm showing no further questions in the queue.

I'd like to turn it back to management for any closing remarks.

Gayn Erickson (President and CEO)

Okay. Thank you, Operator. Everyone, I appreciate your time here. I know we tried to cover a lot of detail. I hope to see you at one of the investor conferences. Or if you want, please get in contact with our folks so we can set up a follow-on meeting to discuss your questions further. Until then, we'll see you next time.

Operator (participant)

This concludes today's conference, and you may disconnect your lines at this time. Thank you for your.