GSI Technology, Inc. (GSIT) Earnings Call Transcript & Summary

Greetings, and welcome to the GSI Technology, Inc. Third Quarter Fiscal Year 2023 Results. [Operator Instructions] As a reminder, this conference is being recorded. It is now my pleasure to introduce your host, Mr. Lee-Lean Shu, Chairman, President and Chief Executive Officer. Thank you, Mr. Shu, you may begin.

Good afternoon, and thank you for joining us to review our fiscal third quarter 2023 financial results. Third quarter revenue of $6.4 million was within guidance, but at the lower end. Revenue growth was impacted this quarter by the uncertain outlook for the global economy. Despite this, we continue to see demand for our SRAM products and interest in our radiation hardened and radiation tolerant products. While customer order patterns are variable right now, these fluctuations are related to economic and external factors, not changes in the market requirements for our products. Despite the lower revenue in the quarter, increased sales of higher-margin products, resulted in gross margin of 57.5%, exceeding the high end of our guidance range. While research and development costs declined sequentially, we saw an increase in selling, general and administrative expenses, primarily related to difference in the level of quarterly adjustments to contingent consideration and severance expense related to recent layoffs. To ensure success and align our resource within the company goals, we launched several cost reduction initiatives at the end of November 2022. The executive team took a comprehensive approach to identify and implement our expense reduction measures, which included a thorough review of all expenses and ways to streamline process and improve operational efficiency. We have 2 objectives with this strategy. One, to reduce our cash burn and; two, to align our resources around developing the APU. We are on track to achieve $7 million in savings on an annualized basis through targeted cost-reductions. These measures aim to right-size our operation and precisely manage spending to increase efficiency and focus our resources on advancing the proprietary APU technology. Let me update you on where we are today with APU hardware and software. The hardware development team is on track to take out Gemini-II in the first half of this calendar year, which puts us on the schedule to see the first win by late summer. In that case, we could test the Gemini-II chip by early fall. If everything goes accordingly, we could have a second table to [ fix back the early next ] calendar year. In parallel, once we have a chip that the software team can use, they will start developing the API and the library for Gemini-II. Keep in mind that it took [indiscernible] to fix the old box in the Gemini-I [ NFOs ] to increase the [indiscernible]. Gemini-II has 8X memory density over Gemini-1 and has 30X cost performance improvements. Gemini-II can greatly enhance our market push of APU technology and provide further substantial savings in power and mono server footprint while enabling large-scale real-time search and HPC workloads. Software for Gemini-1 is an area of intense focus currently. We have a full build up [ TV MAU ] library deployed and used by customers and one research institute has been able to rely on library based only for their projects. Our GRA library is developed for SAR applications, and we have completed a POC project with IAI/Elta based on this. Elta is also evaluating the GPU solution to benchmark against the APU. We could see some initial sales once the APU performance is proven favorable. In the meantime, we are marketing the SAR solution to other customers. We have recently improved our GSL library for similarity search application. We are engaging with a large corporation for the POC project for our own [ plan ] similarity search project that requires very high accuracy and low latency. The improved GSL library is perfectly suited for this application. On the competitor front, we have completed [ C competitor ] that customers are using to program APU with [ c core ]. We are in the process of completing [indiscernible] to allow customers to run API application and their library in-person. Currently [indiscernible] is still internal use and will be released for general use in July. Now I will hand the call over to Didier, who will discuss our business performance further. Please go ahead, Didier.

Thank you, Lee-Lean. Let me switch now to customer and product breakdown for the third quarter. In the third quarter of fiscal 2023, sales to Nokia were $1.3 million or 20.0% of net revenues compared to $1.9 million or 24.0% of revenues in the same period a year ago and $1.2 million or 13.6% of net revenues in the prior quarter. Military defense sales were 26.2% of third quarter shipments compared to 27.1% of shipments in the comparable period a year ago and 22.4% of shipments in the prior quarter. SigmaQuad sales were 45.2% of third quarter shipments compared to 40.5% in the third quarter of fiscal 2022 and 58.1% in the prior quarter. Regarding increased production costs, we are evaluating where we can pass on the increased wafer prices that TSMC announced last year, which became effective starting January of this year. Gemini-I hardware is now market-ready. We have 2 board configurations, the Leda-E, which is in production and the Leda-S, which is an SSD form factor board and is being finalized today. In the third quarter, we shipped 1 Leda-E board to a potential SAR customer, and we shipped 1 Leda-E server to a research institute that will explore Gemini-I for encryption applications. I'd now like to hand the call over to Doug. Doug, go ahead, please.

Thank you, Didier. We reported a net loss of $4.8 million or $0.20 per diluted share and net revenues of $6.4 million for the third quarter fiscal 2023 compared to a net loss of $4.6 million or $0.19 per diluted share and net revenues of $8.1 million for the third quarter of fiscal 2022, and a net loss of $3.2 million or $0.13 per diluted share on net revenues of $9 million for the second quarter of fiscal 2023. Gross margin was 57.5% compared to 55.3% in the prior year period and 62.6% in the preceding second quarter. The changes in gross margin were primarily due to changes in product mix sold in the 3 periods. Total operating expenses in the third quarter of fiscal 2023 were $8.5 million compared to $9 million in the third quarter of fiscal 2022 and $8.8 million in the prior quarter. Research and development expenses were $5.5 million compared to $6.2 million in the prior year period and $6.4 million in the prior quarter. Selling, general and administrative expenses were $3 million in the quarter ended December 31, 2022, compared to $2.8 million in the prior year quarter and $2.4 million in the previous quarter. Third quarter fiscal 2023 operating loss was $4.8 million compared to $4.5 million in the prior year period and an operating loss of $3.2 million in the prior quarter. Third quarter fiscal 2023 net loss included net interest and other income of $61,000 and a tax provision of $84,000 compared to $15,000 in net interest and other income and a tax provision of $64,000 for the same period a year ago. In the preceding second quarter, net loss included net interest and other income of $14,000 and a tax provision of $37,000. Total third quarter pre-tax stock-based compensation expense was $654,000 compared to $740,000 in the comparable quarter a year ago and $661,000 in the prior quarter. At December 31, 2022, we had $35.2 million in cash, cash equivalents and short-term investments and $0 in long-term investments compared to $44 million in cash, cash equivalents and short-term investments and $3.3 million in long-term investments at March 31, 2022. Working capital was $39.2 million as of December 31, 2022, versus $45.8 million at March 31, 2022, with no debt. Stockholders' equity as of December 31, 2022, was $54.8 million compared to $64.5 million as of the fiscal year ended March 31, 2022. Regarding our outlook for the upcoming fiscal fourth quarter, we anticipate net revenues in the range of $5 million to $5.6 million, with gross margin of approximately 49% to 51%. Operator, at this point, we would like to open the call to Q&A.

[Operator Instructions] And our first question will come from Kurt Caramanidis with Carl Hennig, Inc.

What are you thinking cash burn looks like maybe out -- kind of looking out this year with the revenues now being -- looking quite a bit lower per quarter may be?

We were looking at somewhere around $12 million or $13 million a year. And the cost cutting will save us about $7 million a year. So we'll be better off than we were a year ago, I believe.

I mean is it going to be like $4 million or $5 million a quarter?

No, no, no. I think we'll probably be something less than $12 million or $13 million that we were previously seeing for the year.

Is the sale leaseback an option for the building? Or has that been looked into as you're doing other measures?

It's something that we looked at, and it's something that we can consider for the future.

Well, good luck. Hopefully something with APU comes through here in the next few quarters for number one.

Great. Thank you, Kurt.

And our next question is from [ Luke Bowen ], who's a private investor.

Wondering if you all have come across any new application ideas and just generally, which APU applications are you most excited about? And just kind of in context of…

So we're set up…

Yes, go ahead.

I'm sorry, can you clarify your question.

Yes, I was going to clarify that maybe in the context of, as you've all been exploring your technology and exploring marketing channels, going to conferences like the Buzzwords conference. I was just wondering if anything new has arisen or light bulbs getting brighter?

Right. So right now, we're focused on the SAR, as Lee-Lean mentioned. So that's we've done a POC, and we've obtained some very nice algorithm to go along with our hardware. So the benchmarking we've done against CPUs and GPUs are very promising for us, both on a performance level, a power level and a form factor level, which is important depending on where they deploy some of these systems. And so that's one area that we've started -- well, not started, but we've been contacting all the SAR players, both on a commercial level and on a government level. As far as -- I'm sorry, and the other market is the fast vector search is something that we have already put in a plug and we've talked about in the past. Since then, there are a few other applications that we've had customers come to us with. One of them, we, I mentioned in my script, which was one of the boards or systems, I should say, it was a server that we shipped last quarter was for encryption application. And there are a few others that have recently come up. A little early to talk about them just because we haven't gone through the process of seeing what our advantage is yet. But there are certainly no lack of different applications for the Gemini chip.

That's excellent. Yes, it's really good to hear, you can be able to move forward confidently checking off more and more boxes. Okay. Kind of, yes. I think discharge is one of my other questions because I know you're -- well, I guess, yes, I'll just throw it out there as a general prospect, not necessarily for the near term, but I saw that Amazon Prime Air launched their first test sites for their delivery drone program, and I know your advisor was the original leader of that program.

Yes.

So just curious if that could be looking at mobile data and autonomous vehicles, drones, vehicles, yes, like small flying vehicles for commercial transportation and personal transportation. I wonder if you're still seeing that as a potential field for coordinating and managing mobile data like that?

So we are, but more for the Gemini-II chip. And the reason I say that is, if you're familiar with our solution, our Gemini-I chip goes on a Leda board, as I mentioned. And the Leda board for the Gemini-I has an FPGA on there that has certain functionality that is critical for our solution. With Gemini-II, we take that functionality that's on the FPGA, and we put it within the Gemini-II chip. And so now we can rid ourselves of that large FPGA. And so some of the applications you're talking about, the power and form factor being smaller is more important. And so being able to rid ourselves of that FPGA will allow us to pursue those markets that were really a bit too challenging for our Gemini-I chip.

Yes. All right. Yes, because thinking about that and those like next step prospects is kind of thinking about the sensitivity of timing and business relationships, especially these advanced fields that are requiring a lot of R&D and my engender kind of a commitment from these large companies that are developing their programs and the component companies and how they're, say, making systems on chips that are highly customized and requiring a lot of investment. They hope to get something back on and just trying to think about how they might approach or how you might approach that relationship in terms of holding the place and for the future development and not having to directly compete with all of these very like inefficiently developed system on chips toe-to-toe, but rather kind of for them to anticipate being able to adopt your hardware and even your software and to kind of have that in mind as they develop these programs. Yes, kind of wondering if that's something that you are seeing in terms of a 1 to 2 year development plan as you talk to potential clients?

So I want to make sure I -- I'm not sure I fully understood your question. But as far as custom silicon and system on chips and everything out there, most everything has really been geared towards the training portion of the market. And as we've discussed in the past, that's not the application we're focused on. We're focused on similarity search. And there are obviously other applications or computation and intensive that our solution lends itself well. And because of the way we've architected our part where we actually do the processing and the search in place as opposed to having effective data and rewrite data. We -- that technology we have is PAM protected, and we haven't seen anybody try to do that at this point. And so we have carved out a niche in the similarity search. And so as far as other silicon coming in or other SoCs, it's really, like I said, most of the solutions we've seen have really been geared around trying to make the training faster.

Our next question is from [ George Gaspar ], who's a private investor.