Ladies and gentlemen, thank you for standing by. Welcome to GSI Technology's Fourth Quarter Fiscal 2019 Results Conference Call. [Operator instructions]. Before we begin today's call, the company has requested that I read the following safe harbor statement.

The matters discussed in this conference call may include forward-looking statements regarding future events and the future performance of GSI Technology that involve risks and uncertainties that could cause actual results to differ materially from those anticipated.

These risks and uncertainties are described in the company's Form 10-K filed with the Securities and Exchange Commission. Additionally, I have also been asked to advise you that this conference call is being recorded today, May 2, 2019, at the request of the GSI Technology.

Hosting the call today is Lee-Lean Shu, the company's Chairman, President and Chief Executive Officer. With him are Douglas Schirle, Chief Financial Officer; and Didier Lasserre, Vice President of Sales. I would now like to turn the conference over to Mr. Shu. Please go ahead, sir..

Good afternoon, everyone, and thank you for joining us. GSI had a strong year in fiscal year 2019. We grew our revenue 21% and expanded gross margins by over 800 basis points. This gain help offset the increase in our research and development expenses on our APU as we continue evolving our business and bringing new product to market.

As a result of strong revenue growth and increased gross profit, we narrowed our operating loss significantly, ended the year with positive net income. Our capital allocation strategy is to fund the commercialization of our APU, our patented in-place associative computing technology for artificial intelligence applications.

Our APU utilizes a unique architecture that greatly improves computation, search and the response time, while reducing power. The configuration of the APU is designed to be optimum for similarity search applications, in drug discovery, e-commerce and facial recognition. These are the initial markets we are pursuing with our device.

We have identified bugs in first silicon [ph] device. However, we are able to work along the bugs for the device to perform in the target application, and we have successfully demonstrated the APU exceed the performance of existing processes for similarity search.

In the target categories of cheminformatics, which are computer measures used by pharmaceutical companies with aims of discovery, we successfully demonstrated our APU from the core for that -- our customer. The computation rates in our demo accelerated molecular search from minutes to seconds.

The Weizmann Institute team was very happy with APU's performance. We anticipate that with the final silicon of the APU, we can further accelerate this search time down to milliseconds. In e-commerce, search is advancing to perform video search online.

In fact, a number of large e-commerce platforms currently have the ability to search with image as well as packs. We believe that the APU improved the accuracy and speed in video search, resulting in better outcomes.

Similarly, for facial recognition, the APU can quickly, with high accuracy, identify faces in suboptimal conditions, such as low light, far distance and the moving of bird-view image. These are large markets that we are confident that APU can surpass existing solutions.

Our team is engaged with our target customers, who are looking to enhance performance and outcome. The next APU milestone is receiving our second silicon device by Frontier 7G [ph] in 2 to 3 weeks. At that time, we will test the device and then bring them onboard. This will allow us to determine if we have a device we can utilize for general demo.

We anticipate that the second silicon device will perform to our expectations. We had debut customers, training facility at our headquarters in Sunnyvale and have starting creating our marketing material. As a reminder, GSI currently holds 20 patents and the 23 pending patents for our APU in the U.S.

Our team of over 70 engineers based in Israel and rest is focused solely on our ALU map. These individuals have decades of product experience and AI expertise. Together, we have made solid progress in the goal of commercializing our APU.

At this time, I will like to hand the call over to Didier who will discuss our business performance in further detail. Please go ahead, Didier..

Thank you, Lee-Lean. Switching gears to our new RadHard production of our SigmaQuad product. I'm pleased to announce that we have successfully completed the series of critical testing for these parts. The internal qualification process has been initiated. We expect the qualification to be completed in the second half of this calendar year.

As we've stated before, we anticipate having the RadHard chip available for shipments later in calendar 2019. On our sales breakdown, sales to Nokia were 41% of revenue during the fourth quarter of fiscal 2019 compared to 37% a year ago and 45% in the third quarter.

We expect sales to Nokia to continue being strong for the remainder of the calendar year 2019. Military and defense sales were 22.1% of fourth quarter shipments as compared to 29.5% a year ago. The decrease was primarily due to higher sales from Nokia this year.

We saw an increase in military and defense sales, which increased from 16.7% in the fiscal third quarter due to shipments of RadHard engineering samples in this -- in the fourth quarter. SigmaQuad continues to be our best performing product category, sales of 67% of fourth quarter shipments compared to 45% last year and 70% in the preceding quarter.

In the preceding quarter, SigmaQuad sales were unusually strong due to a onetime sale in a supercomputing application, and also a very strong Q3 with Nokia, which we didn't expect to repeat in Q4. I'd now like to hand the call over to Doug.

Doug?.

Thank you, Didier. For the fiscal year ended March 31, 2019, we reported net income of $163,000 or $0.01 per diluted share and net revenues of $51.5 million compared to net loss of $4.5 million or $0.21 per diluted share on net revenues of $42.6 million in the fiscal year ended March 31, 2018.

Gross margin was 61.4% compared to 52.6% in the prior year. Gross margin improvement for the full year was primarily due to product mix. Total operating expenses were $31.8 million in fiscal 2019, an increase of 18.3% from $26.9 million in fiscal 2018.

Research and development expenses were $21.4 million compared to $17 million from the prior fiscal year. And selling, general and administrative expenses were $10.5 million compared to $9.9 million in fiscal 2018.

The increase in research and development expenses was primarily due to development of the APU, including an expense of approximately $1 million related to a nonproduction mask set for the initial APU process, which we incurred in the second quarter -- second fiscal quarter ended September 30, 2018.

For the fourth quarter fiscal 2019, we reported a net loss of $102,000 or $0.00 per diluted share, on net revenues of $12.7 million compared to net income of $265,000 or $0.01 per diluted share, on net revenues of $11.2 million in the fourth quarter of fiscal 2018, and net income of $2.3 million or $0.10 per diluted share, on net revenues of $14.7 million in the quarter for fiscal 2019 ended December 31, 2018.

Gross margin was 61.2% compared to 56.2% in the prior year period and 68.3% in the preceding third quarter. Total operating expenses in the fourth quarter of fiscal 2019 were $8.1 million compared to $6.4 million in the fourth quarter of fiscal 2018 and $7.8 million in the preceding third quarter.

Research and development expenses were $5.6 million in the fourth quarter of fiscal 2019 compared to $4.3 million in the prior year period and $5.2 million in the preceding third quarter.

Selling, general and administrative expenses were $2.6 million compared to $2.1 million in the quarter ended March 31, 2018, and flat sequentially from $2.6 million in the preceding quarter.

Fourth quarter fiscal 2019 operating loss was $365,000 compared to an operating loss of $111,000 a year ago, and operating income of $2.2 million in the prior quarter.

The fourth quarter fiscal 2019 net loss included interest and other income of $186,000 and the tax benefit of $77,000 compared to $109,000 in interest and other income and a tax benefit of $267,000 a year ago. In the preceding quarter, net income included interest and other income of $96,000 and a tax provision of $70,000.

Total fourth quarter fiscal 2019 pretax stock-based compensation expense was $580,000 compared to $549,000 in the comparable quarter a year ago and $592,000 in the prior quarter.

At March 31, 2019, we had $61.8 million in cash, cash equivalents and short-term investments and $9 million in long-term investments compared to $58.4 million in cash, cash equivalents and shot-term investments and $7.9 million in long-term investments at March 31, 2018.

Looking ahead to the first quarter of 2020, we anticipate net revenues in the range of $12.3 million to $13.3 million with a gross margin of approximately 62% to 64%. Operator, at this point, we'll open the call to Q&A.

[Operator instructions] We'll now take your first question from Jeff Bernstein of Cowen..

Just a couple of questions. With regard to the RadHard business, I guess, you guys are also working on high rail line as well.

Could you give us an update on that?.

So that's the what we call rad-tolerant. So it's a little less robust than the RadHard. So that -- the testing for the device is done. So the rad-tolerant is ready to go. We're working with customers now on some design wins. So there's no further testing issues or qualification issues there.

It's just strictly with the RadHard, we were just having to complete the series of tests, which we've done, in fact, in the past month and then that's when we started the qualification on..

Got you.

And those rad-tolerant parts, they would be more in line with the cost for some of these smaller, but high-unit satellite projects or kind of Internet to the world kind of things?.

Generally, they're a better fit for the LEO trajectory, so things that are more like imaging satellites and low-orbiting satellites. Generally if it's a communication satellite or something in the geo orbit, that would generally need a rad-hard..

Got you. Okay.

So even some of these communication networks that are going to have many units would potentially take the RadHard parts?.

It depends -- it really depends on the satellite and who is putting it up there. So some will have RadHard, some will have rad-tolerant, some will have no memory. So it really depends on the satellite and who is putting it up there..

Got you. Okay. And then I just want to make sure I heard the numbers right.

You said you have 20 APU patent issued already, is that right? And then 23 pending?.

Yes, correct..

Got you. Okay, great. And then I think at a conference, you guys talked about the potential for a third silicon spin.

And I'm just kind of wondering is that sure if something wasn't right in the second spin in terms of correcting bugs? Yes, that obviously would be a possibility or kind of -- can you put any kind of percentage on that?.

So as we've discussed, we've seen first silicon, which has bugs. As we discussed with the demo, we are able to do some workarounds to do a customer demo. Second silicon which is out in the next couple of weeks, we're looking to identify all the major bugs that we've gone through.

Now with that said, we don't know if we'll need a third spin, we're just anticipating just from history of how we've done our memories that we will need a third spin just to clean up all the smaller bugs or any kind of manufacturing issues..

Got you. Okay. And so in terms of the whole process and how this has gone, it is a sort of more memory like.

And so you sort of feel like you know what you're dealing with from that standpoint?.

I won't say it's more memory like, it's just historically from some of the complex devices we've done in the past, it just gives us a feel of what we expect. I mean, after seeing first silicon, the fact that there were no bugs, that were a mystery to us that we have -- we identified what the fix was for all of them.

We're hoping that we will be able to get all the bugs behind us so that it will be just is a small cleanup for the third spin..

Our next question comes from Kurt Caramanidis of Carl M. Hennig..

First off, congratulations on a number of achievements throughout this fiscal year. I guess, some of us have been here a while, so it's great to see things are really falling into line here.

On the APU, is that something that you have, let's say, 15 to 20 customers with lots of product or 80 to 100 prospects with less products? How do we -- how can I understand what the marketplace looks like? Is it some big players and a smaller number that have bigger demands? Or is it -- are there more customers?.

So depends on the market. So if you look at the cheminformatics, that's generally going to be smaller customers, but a fair amount of them. If you look at the certainly on the e-commerce, that's where you're going to see some of the large players as well some of the smaller, but certainly that's where the larger players kick in.

And then in the facial recognition, you'll see some small players and that's where you're going to also find some of the kind of the government entities laying as well..

So overall there is a pretty large basket of opportunities as far as numbers of customers?.

That's our anticipation, yes..

Okay. On RadHard, you talked about a program that was funded. How is that? So now are you moving toward a program that you had targeted or maybe kind of how is that going now that you're approved -- now that....

Yes. So we've -- obviously, that's the first one we've been going after because it is the closest to revenue. So yes, we're still in constant communications with that prime contractor along with the end user. And they are -- we've been showing all our results with them, the test results and what have you.

And that would be a program, that assuming nothing else changes, would have some type of revenue by the end of calendar 2019. With that said, we've identified other opportunities, but most of them are not going to kick in until sometime in 2020..

Okay. So calendar 2020, I know we've talked different numbers throughout the last few years.

Is $6 million to $10 million of ballpark of 2020 reasonable opportunities? Is that too high?.

For calendar 2020, that's reasonable..

Okay.

And then what -- and this may be Doug, but what R&D -- what part of the R&D spend is going towards APU? You had the $20 million, was it $21 million?.

Pretty much all of it right now..

What's that? All of it?.

Pretty much all of it for the year..

If we....

We're always focusing on the APU right now in terms of R&D sources..

So had we not had APU, and let's say we were just doing SigmaQuads, regular chips and working on RadHard, you would have generated $20 million of extra cash flow?.

We probably would have had $8 million to $10 million less R&D spending, had we not been working on the APU..

$8 million to $10 million? Okay..

Yes..

Either way you're generating a tremendous amount of cash from the business. It seems like to be able to do that on the revenue that you had....

Yes. The preliminary number on operating cash flow is around $3 million for the year..

Right. With that heavy spend is what, I guess -- weren't coming from it....

That's correct..

If that went away, you would have had quite a bit of cash flow. I heard Lee-Lean talk about the lab at the home office.

Are there customers -- you're planning on having customers there to work on the new APU when it comes in or -- I didn't quite catch it?.

Yes, that's correct. So for some of you who have visited our facility, we had a part of the building that was empty and we built it out. And there is a large training center in there. And so what we anticipate is that with second silicon, the majority or hopefully all the major bugs have been fixed, so that we can do some very active demos.

And active demos include showing how the APU works, but also we're going to do some deep dive for people who want to be able to write their own libraries for our device. So there will be workshops and trainings depending on the people who want to come in.

It will be either some type of hardware training or it will be a deep dive software training, so how they can write their own libraries..

Okay, great. And that's starting like now already or not until they are coming back and, I guess, we're ready to work....

No. Now without a true working demo, when I say true without any workarounds like we're doing now, it doesn't make sense. So we're waiting for those boards. And so it will be sometime earliest July, but probably more into August or September before we start getting all those trainings..

Okay. Let's say a year from now, and I don't know if you start selling -- if it's reasonable a year from now, you're starting -- if you have to do a third spin, then you're starting to sell maybe some APUs and you've got RadHard, rad-tolerant going.

Is $15 million to $20 million run rate per quarter in the ballpark as we're kind of ramping up? And then with that, I would assume margin enhancement based on the metrics of new products?.

So the answer is the revenues in the margins are correct statements. It's just hard to pinpoint when that will kick in. It's a little harder to define what time in the future that will kick in, but certainly, $15 million to $20 million and increased margins is our expectations..

Okay. I apologize just one maybe another one. On George Williams' blog, he's got something where you were testing on FAISS testing the APU. I believe that's Facebook similarity search.

For the layperson, how did you do that or what does that mean? Is Facebook a prospect or are you just using their system to see how this thing works?.

It's a benchmark. So we're -- there are certain industry benchmarks out there and they are known. And so we're just going to be testing ours against those benchmarks..

Okay. And are you doing some outreach in the next couple of months? I thought that was maybe coming up in a couple of conferences..

We're actually at one right now. We'll be presenting or actually hosting a panel, I believe, it's ODSC East. Is that what -- I believe -- I may be incorrect. I want to say it's ODSC East, but I could be mistaken there. But we're hosting a panel and we'll have -- we'll actually have a GSI employee on the panel as well as hosting.

And it's basically the convergence of space and AI coming together and there will be, I want to say including us, 6 companies on the panel. And then we have other conferences planned as well, but that's certainly the one that I want to say it's tomorrow sometime is when the panel happens and it's in -- the conference is in Boston..

Didier and I will also be in Boston and New York last week of May..

But that's where investment, people will be talking specifically on conferences for APU....

Yes. I am looking for investor conferences.

So you do have a couple of those coming up as well?.

Yes..

Your next question will come from the line of Ari Shusterman of Needham and Co..

So I am asking for Raji Gill. So I wanted to start off by talking about your gross margin outlook.

What would be, let's say, a good range for your long-term gross margin? How are you -- how should I think about that?.

Yes. We revised our model about a year ago to 59% to 63% gross margin and obviously one quarter, we blew that away. So we're obviously at the point where we're close to achieving that every quarter at this point. If you look at the last fiscal year, we were probably in there.

We had 68% quarter a year ago primarily because of the mix of product that quarter but I would certainly hope that we never have a quarter below 60%. As we start shipping the RadHard product, we will get some fluctuation in gross margin on a quarterly basis, but I would hope we're at the low point where we are today and that we grow from there..

Okay. That makes sense.

And the RadHard product -- not sure if you mentioned this, but the revenue opportunity from it, was -- would just like to quantify that if you are able to?.

So this calendar year, it won't be a lot. It would be a few million dollars this calendar year. But starting 2020 calendar year, as Kurt mentioned, it should be in the $6 million to $10 million kind of range into calendar 2020..

Yes. And just -- when it comes to Nokia, I guess the share of Nokia compared to, I think, it was military defense sales.

Would you expect the share of Nokia to -- how do you expect the share to change of -- for SRAM or I guess in terms of your end customers or FN [ph] markets rather?.

So I think Nokia is actually very healthy right now. And as we mentioned in the script, the sales looks steady for the rest of calendar 2019. And so as far as revenues with Nokia, I think that will remain steady.

So as far as their percentage of business with us, it's all gong to be dependent on how quickly we can ramp up RadHard in any other customers, and obviously that would bring down the Nokia percentage. But as far as a revenue basis, they'll remain constant for the next few quarters..

[Operator instructions] And we will now take our next question from Jeff Bernstein of Cowen..

Actually, my question was just asked and answered..

Ladies and gentlemen, there are no further questions at this time. I'd like to hand it back over to Mr. Lee-Lean Shu for closing remarks..

Thank you all for joining us. We look forward to speaking with you, again, when we report our first quarter fiscal 2020 results. Thank you..

This concludes today's call. Thank you for your participation. You may now disconnect your lines, and have a wonderful day, everyone. Take care..