Gsi Technology Inc Q4 FY2022 Earnings Call

Ladies and gentlemen, thank you for joining us for GSI Technology’s Fourth Quarter and Fiscal 2022 Results Conference Call. Before we start, the company has asked me to share some Safe Harbor statements. The discussions during this call may include forward-looking statements about future events and the expected performance of GSI Technology, which involve risks and uncertainties that could lead to actual results differing significantly from those projected. These risks and uncertainties are outlined in the company’s Form 10-K submitted to the Securities and Exchange Commission. Additionally, I would like to inform you that this call is being recorded today, May 5, 2022, at the request of GSI Technology. Leading the call today is Lee-Lean Shu, Chairman, President, and Chief Executive Officer of the company. Also present are Douglas Schirle, Chief Financial Officer, and Didier Lasserre, Vice President of Sales. I will now hand the floor over to Mr. Shu. Please proceed, sir.

Good afternoon and thank you for joining us to review our fiscal fourth quarter and the full year 2022 financial results. Our fourth quarter revenue grew year-over-year by 14%. And for the full fiscal year 2022, revenue increased by 20%. The growth in the quarter was due to new projects, expanded business in all categories, and the price increase effective in December 2021. Gross margin improved in the fourth quarter by 840 basis points year-over-year, reflecting a more attractive mix of higher-margin product sales and our ability to manage supply chain challenges and increased costs. As a result, we lowered our operating loss year-over-year by 19%. We ended the year with $44 million in cash, cash equivalents, and short-term investments. Our biggest use of cash remains R&D. The R&D budget and ongoing software development are two large funding projects, one is a software project, the upcoming launch of Version 2 of the Gemini-I compiler stack. The second is a major hardware project, representing the most significant portion of the R&D budget, the design of Gemini-II that we intend to complete by calendar year-end. We will incur a one-time loss charge when we pay that. Other than that, our R&D budget will remain at the current run rate for the foreseeable future due to the ongoing need for software development and our commitment to continuously advancing the hardware platform. We are on track to release Version 2 of the compiler stack in July which will be entirely Python-enabled for coding algorithms, applications, and libraries. This is exciting because our team will be armed with the tools to increase our Gemini-I customer engagement across the various markets we are pursuing. GSI has a highly talented software team that has achieved impressive outcomes in several competitions. This team is a valuable resource as the software is a crucial component to producing success in this market, and one of the reasons we commit a significant portion of our R&D budget to it. The GSI software team has been especially effective and efficient in their work on several high-profile contests, which resulted in the first place wins in two challenges. We took one of the top spots in the Billion-Scale Approximate Nearest Neighbor Search Challenge and improved our technology to perform on par with category leaders in AI. In addition to the great PR, the biggest benefit of this contest is how it raises GSI’s profile in key market sectors such as the U.S. military and defense agencies and other government organizations related to defense and security. Recently, we announced our first great win in the MAFAT Challenge for the human and object tracking categories. GSI loaded software and algorithms that could automatically detect, classify, and track humans and differentiate between males and females, plus identify objects of interest such as weapons that the person might be carrying. Our software was the fastest and most accurate in performing at various distances and under diverse lighting conditions using images from real-time HD video sourced from moving vehicles. This capability has many applications in the large global security market. The MAFAT Challenge gives GSI high-profile exposure to leading agencies in the Israeli and American military and defense organizations, including the U.S. Department of Defense and the new agency, the Irregular Warfare Technical Support Directorate. Using our contacts from the military and defense business, we are working aggressively to pursue additional opportunities as this recent win may offer. In fiscal year 2022, we also had wins in other challenges. GSI was among the leaders in the Billion-Scale Approximate Nearest Neighbor Search Challenge. We proved that APU technology and software could perform on par with the prominent industry leaders in EMEA. In addition to the industry exposure, participating in this contest also taught us a lot about how we need to work on and how we can improve our performance. Our first quest win in the MAFAT challenge resulted in several valuable benefits to GSI. First was our POC with Elta, which is funding a SAR image processing solution on APU technology. Second is the exclusive perpetual license of the FASTA algorithm that Elta will allow GSI to sell as a SAR system product containing these algorithms. We plan to go after the entire SAR market with this product. The third benefit is Elta's credit to deploy decisive volume, assuming that the POC is successful. In fiscal year 2022, we increased the customer engagement for the APU and made contact into new market segments. We are in the early stages of working with numerous POCs and expect to further our engagement with certain versions of our competitors. In fiscal year 2023, we plan to launch our OpenSearch platform on AWS marketplaces with the software starting to select sector searches. Our OpenSearch software platform is particularly good at market model search that can search with multiple inputs like images, text, voice, etc. We are working with our partners to build a software platform that performs multimodal and multilingual factor search with hardware accelerations based on the APU in AWS OpenSearch. The software platform, named OpenSearch, will be presented at the conference in June being held in Berlin, Germany. We have the servers in place and are now building the website with the goal of having the OpenSearch platform available for client use by the third quarter of calendar 2022. Ideally, we are targeting a fully functioning platform, where clients can download the platform and work independently of our support through the process. Another important goal will be the new version of the compiler, utilizing the same product data center to allow customers to work with the APU more easily. This will enable us to build a funnel of costs and move towards building the future revenue stream for the APU. The U.S. team has been working on the design of Gemini-II and finalized the layout for taking out the chip. We are very excited about how Gemini-II could deliver. Gemini-II could expand our addressable market with its dramatically intense performance and much lower cost. We have a great team, and I appreciate their dedication and hard work. We are aligned on our goals for this year and have a lot of motivation to achieve them. I thank all the GSI shareholders for their support, and I value the contributions from the many of you who provided feedback and input as the CEO and the shareholders determined to make the APU successful. 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 start with an update on our radiation-hardened or Rad-Hard business. We have seen increased traction in this area, and in the fiscal year 2022, we shipped four prototypes to three separate Rad-Hard customers. All the customers are currently testing these systems that require chips to withstand harsh conditions. In the first quarter of fiscal year 2023, we have another program with a major prime contractor for a Rad-Hard prototype, which is anticipated to ship in the first half of this fiscal year. This prototype order could potentially be the same revenue magnitude as all of our radiation-hardened orders from last fiscal year. As a reminder, Rad-Hard is a high-growth profit product. And if a few of these prototypes move to production, it could add incrementally to fiscal years 2023 and 2024 revenue. In terms of building a pipeline for APU right now, we are creating awareness of our technology and getting the brand known in the right places. We are encouraged by the response and the incoming interest that we have received from these efforts. The company is evolving from the engineering to the production stage, meaning we are getting the tools and systems in place and modifying how we work to build a sales culture for the APU. I’m optimistic we can progress on these fronts in the upcoming fiscal year. Let me switch now to customer and product breakdowns for the fourth quarter. In the fourth quarter of fiscal 2022, sales to Nokia were $2.0 million or 23.1% of net revenues compared to $2.8 million or 36.5% of net revenues in the same period a year ago, and $1.9 million or 24.0% of net revenues in the prior quarter. Military defense sales were 22.3% of fourth quarter shipments compared to 22.5% of shipments in the comparable period a year ago and 27.1% of shipments in the prior quarter. SigmaQuad sales were 47.6% of fourth quarter shipments compared to 52.9% in the fourth quarter of fiscal 2021 and 40.5% in the prior quarter. Lastly, I want to comment on the current supply chain situation. We moved the majority of our chip substrate suppliers out of China last year to limit the potential damage from any COVID-related lockdowns in China. We are actively addressing supply chain variances that could potentially impact our order fulfillment, but the situation remains fluid. Overall, most of the cost increases from our suppliers were successfully passed on to our customers in December of 2021. We are closely watching costs and evaluating if we need to change prices in the future. GSI and the industry as a whole are constrained by the availability of FPGAs. Since we use FPGAs in our leaderboard, we could potentially be constrained if we receive a large order for an APU platform. For SRAM sales, we have implemented some changes to our supply chain management to ensure that we can fulfill orders for our SRAM customers in the upcoming quarters. We are optimistic that we have sufficient capacity to meet this demand. I’d now like to turn the call over to Doug. Please go ahead, Doug.

Thank you, Didier. GSI reported a net loss of $4.3 million or $0.18 per diluted share on net revenues of $8.7 million for the fourth quarter of fiscal 2022 compared to a net loss of $5 million or $0.21 per diluted share on net revenues of $7.7 million for the fourth quarter of fiscal 2021 and a net loss of $4.6 million or $0.19 per diluted share on net revenues of $8.1 million for the third quarter of fiscal 2022. Gross margin was 58.6% in the fourth quarter of fiscal 2022 compared to 50.2% in the prior year period and 55.3% in the preceding third quarter. The improvement in gross margin was primarily due to changes in the mix of products sold and price increases effective in December 2021. Total operating expenses in the fourth quarter of fiscal 2022 were $9.4 million compared to $9.1 million in the fourth quarter of fiscal 2021 and $9 million in the prior quarter. Research and development expenses were $6.5 million compared to $6.1 million in the prior year period and $6.2 million in the prior quarter. Selling, general and administrative expenses were $2.9 million in the quarter ended March 31, 2022 compared to $3 million in the prior year quarter and $2.8 million in the previous quarter. Fourth quarter fiscal 2022 operating loss was $4.3 million compared to an operating loss of $5.3 million in the prior year period and an operating loss of $4.5 million in the prior quarter. Fourth quarter fiscal 2022 net loss included interest and other expenses of $47,000 and a tax provision of $21,000 compared to $21,000 in interest and other expenses and a tax benefit of $304,000 for the same period a year ago. In the preceding third quarter, net loss included interest and other income of $15,000 and a tax provision of $64,000. Total fourth quarter pretax stock-based compensation expense was $714,000 compared to $753,000 in the comparable quarter a year ago and $740,000 in the prior quarter. At March 31, 2022, the company had $44 million in cash, cash equivalents, and short-term investments and $3.3 million in long-term investments compared to $54 million in cash, cash equivalents, and short-term investments at March 31, 2021. Working capital was $45.8 million as of March 31, 2022, versus $56 million at March 31, 2021, with no debt. Stockholders’ equity as of March 31, 2022, was $63.1 million compared to $75.6 million as of the fiscal year ended March 31, 2021. For the fiscal year ended March 31, 2022, we reported a net loss of $17.7 million or $0.73 per diluted share on net revenues of $33.4 million compared to a net loss of $21.5 million or $0.91 per diluted share on net revenues of $27.7 million in the fiscal year ended March 31, 2021. Gross margin for fiscal 2022 was 55.5% compared to 47.7% in the prior year. Total operating expenses were $36.2 million in fiscal 2022, an increase of 5.1% from $34.5 million in fiscal 2021. Research and development expenses were $24.7 million compared to $23.3 million in the prior fiscal year. Selling, general and administrative expenses were $11.6 million compared to $11.1 million in fiscal 2021. The increase in research and development expense was primarily related to the development of Gemini-II. The operating loss for fiscal 2022 was $17.7 million compared to an operating loss of $21.3 million in the prior year. The improvement in the operating loss was primarily due to the increase in revenue and gross profit. The fiscal 2022 net loss included interest and other expenses of $60,000 and a tax benefit of $45,000 compared to $94,000 in interest and other income and a tax provision of $335,000 a year ago. Operator, at this point, we will open the call to Q&A.

Thank you. Our first question is from Jeff Bernstein with Cowen. Please go ahead.

Hi, guys. I just wanted to get an update. I think you were supposed to get a ride for Rad-Hard parts satellite here at some point so you could get that space legacy. What’s the schedule on that now?

Great question. This is Didier, Jeff. So that was shipped out last year or – yes, some time last year. And so it was, as you said, it was going to be the very end of calendar 2021 or beginning of calendar 2022 it was supposed to go up. We’ve been in constant contact with this prime contractor, and they are still waiting for some other components for that satellite. As you know – I’m sure you know, the semiconductor lead times are way out right now for everything. And so there are still some components they have not received. And so the satellite has not been launched yet.

Any visibility on when that might happen?

No. They are not telling us too much about that. They are not telling us what components are missing and when. But they certainly feel that it’s going to be some time in the summertime they should have all the components.

Got it. Okay. And then just curious, what’s the cost of the mask set for the Gemini-II? What process node is that on? Is that a 14-nanometer part? Or what is it?

Yes. Well, you’re talking about the process or the...

Yes, the process, 16. So Jeff, the process is 16-nanometer.

The process node.

Yes, the process node, 16-nanometer. And as far as what would that mask set cost?

About $2.5 million.

Got it. Okay. And I was curious about what kind of visibility you have with Nokia in fiscal ‘23. I think you had better visibility this last year partly because of all the supply constraints, etcetera. What kind of visibility do you have now?

Same. So we have a 12-month – sorry. We have a 12-month rolling forecast right now and so that’s remained the same. So the visibility is constant. So we still have pretty good visibility.

Got it. And so that’s indicating what? A flat year? A lower year? What kind of a year with Nokia?

Good question. It looks right now it will be flat to maybe a touchdown, but it’s pretty constant.

Got it. Okay. And I’m curious, we had talked about the potential and the mosaic when kind of suggests this, that there might be application in the very large driver monitoring and cockpit monitoring applications that are going to be legislated here in Europe and then potentially in the U.S. What have you guys figured out there?

It’s still an ongoing process. So there are some areas we think there could be a fit. Other areas are closer to the edge, and we think it’s going to have to be more of a Gemini-II play. As you know, with the Gemini-I, it goes on a board that has an FPGA. With the Gemini-II, we don’t have those constraints. So we’re still looking into it. We think there is certainly some opportunity with Gemini-I, but I think a lot of it, because it’s more of an edge application, will require the Gemini-II.

Got it. Okay. And then you mentioned that you think the brand is getting known in the right places. What are the right places?

Certainly, the markets we’re going after, right? So we’ve made some announcements recently with the SAR initiative that we have. I mean, certainly, that came out of the original MAFAT Challenge we won. And as Lee-Lean mentioned, we have the POC with Elta. And so what we’re doing now is with that relationship, we got access to a very fast back projection SAR algorithm, which we’re going to be able to use without any license or anything for any other customers. And so, we are putting together a marketing sales package to be able to address that to all the other SAR players. We also, with this latest win in Mosaic, it’s giving us inroads into some of the DoD programs. And so certainly, what we’re – the areas that we’re going after, certainly the markets where there is certainly a need for our type of fast search.

Got it. Okay. And then you talked about this new OpenSearch multiple input search, which sounds very interesting that it can take natural language and symbolic language and pictures for searching. Is that correct? And just talk about what’s your angle here. What do you think the objective is with that?

Yes. Just as I mentioned, I mean, we are working on the multimodal model, which means you can have a query which comes in text, image, and voice, everything coming together for the query. And especially good is we have a software platform to OpenSearch. I mean the OpenSearch offers a large search capability that has a lot of in-store customer bases, and we – the software parking allows a customer not to require to learn a lot of software to perform the search. Right now, you see some websites can do the search service, but that requires proprietary software to do it. And now we’re going with the OpenSearch or large search, and all this software is really behind the customers. Customers don’t even have visibility for it. So, that is how we are excited about, with all the installed customer base in the industry. And so that’s a big marketplace we can go after.

So Lee, just so I understand, so if a retailer has a search button on its mobile website, you could plug in the backend, put your software in between, and allow somebody to take a picture of a bicycle and say, I want this bicycle in a different color. And the search algorithm would be able to search and take those two inputs, the picture and the person’s voice, and come out with search. They would essentially plug that functionality into their search without doing a lot of heavy lifting?

Exactly. And they can talk in the price range too. You can have a picture of the bicycle, and then you say you want the color. And then you want the price range of $500 to $1,000. You will get exact merchandise you’re looking for.

Got it. Okay. And then so this could be won by you as a business where those queries would go to APU boards that you own and it would be a cloud-accessible kind of search? Or is it caught on, you would just sell the APU boards to the retailer or whoever their cloud provider was?

Yes, they can – yes, we can provide either the cloud service or we can provide hardware for the on-premise applications.

Thanks, Jeff.

Our next question is from James Poko, Private Investor. Please go ahead.

Hi, couple of questions regarding the development of Gemini and the I and II and the APU. What is the budgeted cash burn for fiscal year 2023? What’s the expectations?

Right now, going into the year, I would expect something similar to fiscal 2022.

And what’s that?

Probably something less.

Approximately?

I don’t know, $10 million, $12 million.

Yes. And also regarding development of the Gemini-I, according to past company statements, the compiler stack was supposed to have been available in 2021 and it faced repeated delays. Now we’re looking at July, if we even get it done then. Could you tell me why that’s taking so long?

Well, the compiler has – okay, the Gemini software, we have two portions of the software. One is the library, functional library which can drive our processing rate. That’s one portion of the software. The other portion is we have an embedded processor, or we call it coprocessor in the chip, which is a console, the insertion flow, and the data flow. So right now, we have completed the compiler on the instruction side, the microcode, the functional library. And we can use a sheet for the embedded part for the coprocessor. So we do have a compiler available if a customer wants to use a high-level language to program the APU. Like I use a sheet compiler for the coprocessor, and I can use Python for the functional level. So the version two we are talking about can do Python for everything, right, for the coprocessor and for the library. So that’s not really a delay, it’s just an extra step we need to take to have a full compiler stack for it.

How many people are testing the Gemini-I APU? How many customers?

You’re saying about customers? Right, so it’s at multilevel. So we have some people who have purchased some boards for prototyping. We have a couple of boards that are out as loaners. And then a fair amount of folks are doing their testing with us, not with physical hardware, but over the cloud. And so we have a data center set up in our Israeli facility, here in our Sunnyvale, California facility. And then we also have, as we’ve talked about in the past, the setup that we have at CoreSite for the OpenSearch Elastic search. And so we have folks doing their testing there. And so, how many? Right now, we’re talking about dozens.

Well, we also have, just like I mentioned in the earnings call, I just mentioned that we have a partner developing the multiyear marketing search. So that’s the development and that’s going to be – and also that’s going to be presented at the conference. So that’s the customer also behind us. So he can promote the algorithm with the APU hardware.

Right. And just to provide a little more color, we also have begun working with some integrators. The integrators would be a single entity itself, but they are working with multiple customers on their end. So it’s hard to determine how many more customers we touch via these integrators.

And the share price sits at a multiple year low. What is management’s position there to try to improve on that because that’s sitting at quite a low level. I’m not sure what it closed today. But if you look at a chart, that’s kind of rock bottom. Why do you think the company’s shares don’t gain any respect for what they are doing with the APU and then Gemini-I and II?

Well, I think the biggest thing is that there is not a lot of visibility in terms of what the revenue stream is going to be. I don’t think people understand that well yet. We put out some TAM and SAM data a couple of…

Yes, you did. $4.5 billion on SAM, and you got some pretty large numbers there. We are a small company. We have an enterprise value of about, what, $50 million right now? A couple of hundred million in sales would make a huge impact. So, how does that…?

As Didier said, there is a lot of activity going on. And we don’t have a lot of input from customers in terms of volumes and quantities and so on, so we haven’t put any revenue forecast. But we believe that there is a large opportunity. There is a lot of traction as Didier and Lee-Lean have talked about here today that we believe will lead to that revenue growth.

I hope so. And regarding AWS, wasn’t there another 2.0 coming up or what’s happening there?

Correct. That’s – as you say, it’s 2.0. And AWS has not released it yet.

When is the expectations for that?

It was supposed to be March. Well, it was supposed to be last year, then it was supposed to be March of this year and it hasn’t come out yet. I mean we are still working on the current version just to get everything up and running, but we don’t control the release of 2.0. And so I don’t know what the latest projection is on the release of that.

Anything else you can add to help shareholders understand like what the potential is for APU because the shares are sitting at $50 million enterprise value. It’s pretty disappointing.

Well, hopefully, we will have a lot more to say in terms of commitments from customers, design wins and so on. And I think that’s where we are going to see the benefit to the stock price.

And that’s when you get the Compiler Stack done in July. Is that correct?

The Compiler Stack will only make the process easier. I mean, the first release in February of the initial Compiler Stack that Lee-Lean mentioned had the combination of Python and C+ depending on the… so that certainly started the process. This is the next step to make it even simpler. So everything is in Python. So that helps build the ecosystem. But we are certainly working – even outside of that, we are working with customers. As I mentioned, the SAR applications that we are going after and some of these other issues – some of these other areas that aren’t strictly tied to the Compiler Stack. The Compiler Stack is really to get us to a much broader, larger level where we are not involved in the POCs or the algorithm writing.

Well, we need to get that done. And I am holding you accountable for that July date that you guys mentioned.

Yes.

Alright. Very good. Thanks for your time and I appreciate you taking all my calls and all my questions. Thank you.

Thank you.

Thank you all for joining us. We look forward to speaking with you again when we report our first quarter fiscal 2023 results. Thank you. And that does conclude the conference call for today. We thank you for your participation and kindly ask that you please disconnect your lines. Have a great day, everyone.