Everspin Technologies, Inc. (MRAM) Q2 2017 Earnings Report, Transcript and Summary

Good morning, and welcome to the Everspin Technologies' Second Quarter Financial Results Conference Call. [Operator Instructions] Please note this event is being recorded. I would now like to turn the conference over to Dave Allen. Please go ahead.

Thank you, operator, and thanks to all of you for joining Everspin's Second Quarter 2017 Financial Results Conference Call. Before we begin the call, I want to remind you that this conference call contains forward-looking statements regarding future events, including but not limited to our expectations for Everspin's future business, financial performance and goals, customer and industry adoption of MRAM technology, successfully bringing to market and manufacturing products in Everspin's design pipeline, the range of timing between design wins to design -- design ins to design wins to volume production, including the timing of Spin Torque product shipment and the execution of its business plan. These forward-looking statements are based on estimates, judgments, current trends and market conditions and involve risk and uncertainties that may cause actual results to differ materially from those contained in the forward-looking statements. I would encourage you to review our SEC filings including Everspin's second quarter 2017 financial results press release issued on August 10, 2017, the company's quarterly report on Form 10-Q filed with the SEC on May 15, 2017, Everspin's annual report on Form 10-K, filed with the SEC on March 29, 2017 and other SEC filings made from time to time in which we discuss risk factors associated with investing in Everspin. All forward-looking statements are made as of the date of this call, and except as required by law, we do not intend to update this information. In today's call, we will be referencing GAAP and adjusted EBITDA numbers. Adjusted EBITDA numbers are provided to enhance the investors' understanding of Everspin's operating performance as it primarily excludes certain non-cash charges for depreciation and amortization, stock-based compensation expense, compensation expense related to the divesting of common stock held by GLOBALFOUNDRIES resulting from our joint development agreement. The use of adjusted EBITDA is not meant to be a substitute for the results presented in accordance with GAAP but rather should be evaluated in conjunction with GAAP. This conference call will be available for audio replay in the Investor Relations section of Everspin's website, www.everspin.com. Joining me today are Everspin's CEO, Phil LoPresti, and CFO Jeff Winzeler. I'd now like to turn the call over to Phil. Phil?

Good morning, everyone. For those of you who might not have seen our financial results press release, you can find the press release and an updated Investor slide deck on our website. After our call is completed, we will post our prepared remarks and when available, we will post a transcript from this call on the Investors section of our website. Before I ask Jeff to discuss our Q2 results in detail and provide our Q3 guidance, I would like to quickly make a few comments about our second quarter results as well as other recent activities crucial to our future business. Revenue for Q2 2017 was $8.9 million, a record quarter for Everspin, and at the high end of our guidance of $8.6 million to $8.9 million. I am encouraged by the continued execution of our business plan and the customer reception we are experiencing with our MRAM technology. Our bookings for Toggle MRAM continue to be very strong, and our new design activity for our Toggle MRAM products remain robust with a 33% increase in the number of design wins compared to the prior quarter. With regards to our Spin Torque MRAM technology, we expect to move into production with our nvNITRO Accelerator cards, which use our 256- Spin Torque chips in the current quarter. At this week's Flash Memory Summit, Everspin is demonstrating our nvNITRO Accelerator card and U.2 solutions. Early sampling of our 1-gigabyte DDR4 Spin Torque chips to customers started in June. We expect to begin shipping preproduction customer samples by the end of 2017 and with production-qualified parts available by the end of Q1 2018. After Jeff's financial commentary, I will come back and share some additional thoughts on our Toggle and Spin Torque MRAM design activity, customer feedback and our visibility regarding ramping of our 256-megabyte and 1-gigabyte Spin Torque products, and our current revenue generators. I will also briefly touch upon the competitive environment as it relates to our embedded MRAM roadmap and our market opportunity. After that, we will then open the call to questions.

Thank you, Phil. And good morning, everyone. I'd like to start by reviewing the second quarter 2017 income statement. Revenue in the second quarter was $8.9 million with product sales representing 90% of total revenue or $8.1 million. While licensing, NRE and royalty contributed $863,000 in the quarter. The Q2 2017 revenue at $8.9 million was $1 million or 13% higher than the revenue in Q1 2017. Looking at product sales alone, revenue was $8.1 million compared to $7.2 million in product sales in Q1 2017, an increase of 12% quarter-over-quarter. Toggle MRAM sales represented 81% of product sales, and were up $935,000 or 17% from the first quarter of 2017. Legacy product sales were $1.6 million and were roughly flat with Q1 2017. We did not record any Spin Torque MRAM revenues in the second quarter of 2017. Gross profit for Q2 2017 was $5.8 million, an increase of $1.6 million or 37% over Q1 2017. The resulting gross margin for Q2 2017 was 65% versus 54% in the prior quarter. The gross margin expansion in Q2 2017, when compared to the prior quarter was largely the result of better yields for Toggle MRAM products. This yield improvement was a result of a fix in a parallel test program at final test that was falsely rejecting good units. Upon remediation of this issue, the base yields for our Toggle MRAM products improved. This is a benefit, expected to provide higher yields and therefore, lower costs in future periods. In addition to the ongoing benefit of the test program fix, we were able to process a number of previously rejected units, which provided a 1x increase of 2.3 margin points in Q2 2017. While our product gross margins in Q2 were aided by the benefit of increased yields and of onetime benefits, we are still targeting a gross margin for the company of 48% to 52% in our financial model. Licensing, NRE and royalty revenues added another 4 margin points in the second quarter of 2017, due to the achievement of a milestone in one of our NRE agreements. Licensing, NRE and royalty revenues are dynamic due to contract terms, milestones and other factors. While we opportunistically look for engagements to license our technology portfolio, this revenue is not the focus of our core business and will continue to be sporadic over time. Q2 2017 operating expenses were $10.6 million compared to $10.1 million in the prior quarter, an increase of $460,000. Breaking down operational spending, research and development expenses in Q2 were flat at $6.4 million versus $6.4 million in the previous quarter. Within R&D spending, joint development agreement expenses were $436,000 lower in Q2 than in Q1. This decrease was offset by $148,000 of increased product engineering expenses and $200,000 increase in noncash stock comp charges associated with divesting of Everspin shares owned by GLOBALFOUNDRIES. SG&A spending for Q2 2017 was $4.1 million compared to $3.7 million in the previous quarter. The increase was due to recruiting, travel and payroll expenses related to staffing changes in the sales and marketing groups for Everspin. Interest expenses for Q2 2017 were $176,000 compared to $230,000 in Q1. Other income was $24,000 and the loss on extinguishment of debt was $246,000, due to the unamortized balance of the debt discount and a prepayment penalty for the Aries loan we retired in the quarter. GAAP net loss for Q2 2017 was $5.2 million compared to a $6.1 million net loss in the previous quarter. The Q2 GAAP loss per share was $0.42 compared to $0.49 loss per share in the previous quarter. At this time, I'd like to discuss our year-over-year income statement results. Total revenue increased by $2.3 million, or 34% from $6.7 million in Q2 of 2016, to $8.9 million in Q2 of 2017. Product sales increased by $1.5 million, or 22% from $6.6 million during the second quarter of 2016 to $8.1 million during the second quarter of 2017. The increase was due to $1.5 million of increased sales volume of our Toggle MRAM products. Licensing, NRE and royalty revenue increased by $800,000 from $100,000 during Q2 2016, to $900,000 during Q2 2017. The increase was primarily due to the achievement of NRE milestones during the second quarter of 2017. Licensing, NRE and royalty revenue remain highly variable revenue item characterized by a small number of transactions annually, with revenues based on size and terms of each transaction. Gross profit from the second quarter of 2017 was $5.8 million, an increase of $2.3 million, or 65% from the same quarter last year. The resulting gross margin increased from 52.6% during the second quarter of 2016, to 64.8% during the second quarter of 2017. The increase was due to better product mix, the previously discussed higher yields on Toggle MRAM products and the $800,000 increase in licensing, NRE and royalty revenue. Focusing on operating expenses, research and development expenses increased by $300,000, or 6%, from $6.1 million during Q3 2016, to $6.4 million during Q3 2017. The increase was due to a $400,000 increase in expenses in our joint development agreement with GLOBALFOUNDRIES, a $200,000 increase in depreciation expense and rent expense, related to the move of our RD laboratory and $200,000 increased employee and contract labor-related expenses. These increases were partially offset by a $500,000 decrease in the amount attributable to divesting of shares of common stock issued to GLOBALFOUNDRIES. SG&A expenses increased by $1.7 million or 68% from $2.4 million during Q2 2016, to $4.1 million during Q2 2017. The increase was attributable to personnel-related and stock-based compensation expenses due to increased headcount as well as accounting, insurance, legal and investor relation expenses incurred as a result of becoming a publicly traded company. Interest expense decreased by $500,000, or 75% from $700,000 during the second quarter of 2016, to $200,000 during the second quarter of 2017. The decrease is related to a $400,000 decrease in interest payable to prior convertible promissory notes and to GLOBALFOUNDRIES. Additionally, interest expense decreased by $100,000 due to the consolidation of our prior facility with Aries Venture financial into our 2017 credit facility with Silicon Valley Bank. Other income was $24,000 during the second quarter of 2017 compared to $300,000 during the second quarter of 2016. The change was primarily related due to the fair value remeasurement of our warrant liabilities, which were reclassified to equity upon the IPO in 2016. The loss on extinguishment of debt for Q2 2017 was $246,000 compared to 0 in the previous year. The company's net loss for the second quarter of 2017 was $5.2 million compared to $5.4 million in the same quarter last year. On an adjusted EBITDA basis, the loss for the second quarter of 2017 was $3.8 million compared to $3.5 million in the same quarter last year. Now turning to the balance sheet. Cash and cash equivalents were $21.2 million at the end of second quarter 2017 compared to $24.5 million at the end of the first quarter. Total assets at the end of the second quarter were $36.8 million compared to $37.5 million at the end of Q1 2017. Total liabilities were $18.8 million at the end of the second quarter compared to $16 million at the end of the first quarter of 2017, an increase of $2.2 million. The refinancing of our term loan was the primary factor in the increase. Stockholders' equity was $18 million at the end of the second quarter 2017 compared to $21.5 million at the end of the first quarter of 2017. Capital spending for the second quarter was $1.7 million, the bulk of which was the relocation of our R&D laboratory in Chandler, Arizona. Looking ahead to the third quarter of 2017, we expect revenue to range between $9.7 million and $10.1 million. We expect the resulting GAAP loss per share will range between a loss of $0.48 per share to $0.44 per share based on an average weighted share count of 12.5 million shares. I'll now turn the call back to Phil for additional comments about our existing business and expectations regarding our Spin Torque product shipments. Phil?

Thank you, Jeff. Overall, I am pleased with our business performance, strong demand for our Toggle MRAM as measured by our book-to-bill metric, and the execution of our business plan, particularly as it relates to our 1-gigabyte development. Our customer engagements with our Spin Torque MRAM products continue to advance. I am pleased that our first 256-megabyte design win, which is the SMART Modular's NVM Express PCIe Card has entered production and will begin shipping in Q4. At the Flash Memory Summit, we also officially launched 4 of our first planned nvNITRO products, just 6 months after we began preproduction sampling of our nvNITRO product in early February. As I mentioned in my opening remarks, customers attending the Flash Memory Summit had -- have had the opportunity to see a demonstration of our nvNITRO and U.2 solutions. These solutions are designed to deliver extremely fast, read and write times with ultra-low latency, offering initial capacities of 1-gigabyte and 2-gigabytes. Produced in conjunction with our ecosystem partners SMART Modular and Xilinx, we expect that these cards will be available for initial customer delivery in the fourth quarter. We are now accepting orders for the 1-gigabyte product and will be scheduling initial shipments in Q4 with pricing at $2,200 dollars. Based on what we've learned to date from our nvNITRO and Spin Torque customer engagements, customer evaluations of our nvNITRO products could take up to 9 to 12 months. Some customers have begun the process using our Alpha Data platform, starting in the beginning of February this year. Other customers will begin this process starting now with our production cards. Additionally, we are seeing a similar 9- to 12-month period for customers planning to use our 256-megabyte and 1-gigabyte ST-MRAM products to move from the design-in stage to production ramp of their product. Our 1-gigabyte product is attracting a lot of customer interest, and provides a high endurance, persistent memory with a DDR4 compatible interface. These features will allow storage system designers to enhance the reliability and performance of storage devices and systems by providing protection against power loss without the use of super capacitors and batteries. We are now providing engineering samples of our 1-gigabyte DDR4 Spin Torque MRAM to select customers. The 1-gigabyte chips are being produced in partnership with GLOBALFOUNDRIES and their 28 nanometer CMOS process on 300 mm wafers, utilizing Everspin's patented perpendicular magnetic tunnel junction. The focus of Everspin's sales and marketing team remains squarely on converting our Spin Torque opportunities into design-ins and then into design wins, and then ultimately into revenue generators. While we are not announcing additional Spin Torque design wins on this call, I am very encouraged by the strong customer interest in our Spin Torque MRAM solutions is generating, including several significant customer engagements that may advance to design wins later this year. As I stated in past calls, the exact timing of when customer projects will enter volume shipments using our 256-megabyte and 1-gigabyte Spin Torque solutions is difficult to predict and dependent on a variety of factors, many of which are outside of our control. With the 1-gigabyte chip sampling now, we believe, some customers may be reevaluating the product plans and may consider using the 1-gigabyte device in their initial products instead of the 256-megabyte. This reevaluation of which density chip to use could result in our revenue ramp of our Spin Torque chips taking longer than we previously expected. We can now also report that 1 of our 4 Spin Torque customer design wins we discussed on prior calls, is with our 1-gigabyte chip. We have been working closely with this customer, as they chose to adopt the 1-gigabyte product based on our early engineering samples. We expect to ship 1-gigabyte preproduction material, for this major OEM's flash array product in the fourth quarter of this year, which will be used for their pilot production and inventory builds in Q1 of 2018. The customer has informed us they plan to begin volume shipments of their systems in Q2 2018. I would like to point out that the timing of this 1-gigabyte project is a little atypical, for moving to the design-in and design-win stage prior to production release. As mentioned earlier, most customer projects will transition to production approximately 9 to 12 months from the design-in phase. The 1-gigabyte Spin Torque MRAM density further expands our market opportunity, while we are initially targeting the enterprise storage market with our 1-gigabyte capacity. Longer term we expect to work with our ecosystem partners and customers to offer additional solutions with other high-speed interfaces. We believe such products will significantly expand our current addressable market that would include but not be limited to new industrial and automotive ADAS applications. In fact, we are already engaged with industry participants, current and potential customers to address these future Everspin markets. We will provide updates to our road map as we progress in these areas in subsequent earning calls. Let me quickly comment on our current revenue generators. Our Toggle products, which target persistent SRAM applications with memory densities ranging from 128kb up to 16-megabyte remain the bulk of our business. Our legacy business along with our Toggle sales, we believe will continue to provide a stable and growing revenue stream until we see an inflection point when our Spin Torque shipments start to become a meaningful part of our revenue mix. As we ramp shipments of our Spin Torque products, we expect to begin to benefit from the financial impact at enterprise storage applications using multiple, higher-density chips that significantly higher ASPs will provide. Also highlighted in my introductory remarks, we continue to see a steady stream of new Toggle design wins. Let me share with you 3 recent Toggle design wins that represent the diversity of the types of applications that our Toggle MRAM supports. The first example is with a repeat customer, a Japanese supplier of SSDs. This win involves our 256 kb Toggle MRAM that they plan to use for capturing data on power loss. A design win like this with existing Toggle storage customers we believe can lead to additional higher-density Spin Torque sales in the future. The second one I want to highlight is in the energy management sector. Our 4-megabyte Toggle MRAM is being used as a data logging memory and a smart meter recording the meter usage information. The third recent design win is with a worldwide leader in the development of electronics for industrial automation and robotics. Our 4-megabyte chip will be deployed in there PLC or programmable logic controller, for recording manufacturing data and program code. Before moving into the Q&A section of the call, let me briefly comment on potential MRAM competition and our plans for embedded MRAM. First, we are focused on application-specific markets where the benefit of our products and their feature sets are valued. In other words, we are avoiding comedy -- commodity markets. Second, we have shipped over 70 million MRAM chips, established a number of important ecosystem partners, put in place a scalable manufacturing partnership with GLOBALFOUNDRIES, developed and implemented proven MRAM intellectual property and have accumulated a lot of know-how and experience. While there have been announcements by other companies, both large and small, regarding their MRAM plans, to date we have not seen or heard of any MRAM chips being sampled or currently commercially available. We recognize that the competition will ultimately have MRAM products. To address those likely MRAM entrants, several years ago we partnered with GLOBALFOUNDRIES to produce our Spin Torque discrete chips. GLOBALFOUNDRIES is producing these chips today on 300 millimeter wafers at 40 nanometer and 28 nanometer nodes. GLOBALFOUNDRIES has already announced plans to introduce embedded MRAM using Everspin's technology, to produce SOCs for their foundry customers on more advanced smaller geometries in the future. We have established a licensing arrangement with GLOBALFOUNDRIES to address the large embedded MRAM market where competitors are more likely to first enter. While I cannot comment on GLOBALFOUNDRIES' activities in this area, we are pleased with their decision and the progress they have made to date, which we believe will provide an important time-to-market advantage for our 2 firms as they initially target solutions for the low-power embedded and IoT applications. We are excited that our partner GLOBALFOUNDRIES has now made PDK kits, or process development kits, available for customers to design embedded MRAM SOCs on their 22 nanometer FDX process. We expect this to lead to early adopter design wins. In closing, the revenue opportunity in front of us from enterprise storage applications, using our higher-density Spin Torque MRAM is considerably larger than that of our existing Toggle designs, and we remain focused on expanded design wins with both Toggle and Spin Torque MRAM products. Customer adoption of our Spin Torque MRAM, we believe provides us significant opportunities to expand existing addressable markets and achieve our business plan. Once again, I want to thank our employees, our ecosystem and manufacturing partners and suppliers, who have given us an important time-to-market first mover advantage over potential MRAM competitors. And finally, to our customers and investors, without whose sport we would not exist. Now we will open the call up for questions. Operator?

[Operator Instructions] The first question comes from Kevin Cassidy of Stifel.

On the gross margin, the improved yields with the Toggle products lifted gross margins, but you're keeping long-term gross margins the same. Is that due to product mix or maybe if you can explain some of the moving parts on that?

Yes, Kevin, it's a great question. So the reason that we're guiding -- looking forward in our business model that our margin will still be targeted at 48% to 52% is because, and you said it just right, the product mix aspect of it is much more impactful than the ongoing yield improvement that we're going to get from Toggle. Our Toggle product portfolio is made up of a bunch of different products, probably 10 to 12 major products in that product family and all of those have different margins associated with them, a lot depending on density and then to the markets that we're selling. So whenever we have a mix of products that's different quarter-over-quarter, it has pretty big impact in terms of our gross margins. And we think that 52% to 48% is kind of the right ongoing forward-looking target for our financial model.

Okay. Great. I attended the Flash Memory Summit also, and there seemed to be some excitement around the nvNITRO. Can you explain maybe what some of the obstacles to adopting or for designing in with the nvNITRO?

Kevin, this is Phil. So with the product like this that's targeted for the enterprise-class environment and mission-critical opportunities, there is frequently the requirement to do a lot of thorough testing of the device. Of course, using MRAM, which is a first-to-market product by Everspin, in this case it's our 256-megabyte, but then again, putting it into an NVMe half-height half-length card or into our U.2 form factor, most of the customers want to plug it into systems, and do a lot of vigorous testing and evaluation of the performance of the device before they certify it to go out to their large customer base. So we expect that process is going to vary. Some of these customers started evaluating at the Alpha Data cards state. That was the prefab card from Alpha Data where we were able to put our MRAM DIMMs into it. That was not our production version. Now those parties have started evaluations, but other customers preferred to wait until the actual production card was available, and that's the one that we've worked and collaborated on together with SMART and we have now brought out and started sampling, and as we've indicated in this call, we're now accepting orders for production. So it's going to vary customer-to-customer and application-to-application, but we expect that to be somewhere in the 9- to 12-months range before they start ramping into production.

And the next question will come from Matt Ramsay of Canaccord Genuity.

A lot going on in the call there and congratulations very much on the 1-gigabyte progress that you're making. I think Phil, in your prepared script there you guys talked a little bit about the possibility, and it sounds like one of your design wins in fact, took this path, of folks maybe being impressed with the progress on 1-gigabyte maybe a little faster than the customer base has anticipated and choosing to use that higher density and more capable solution from you guys in initial products and maybe skipping over the 256-meg product. And I mean, that would be great in the long term and, obviously, there'll be a little bit of revenue impact from that in the near term. So maybe if you could talk through about those dynamics. I guess one, why customers might make that decision and what that means for product time lines and maybe, Jeff, if you could talk a little bit about what that means from a revenue standpoint here over the next few quarters.

Sure. Thanks, Matt. So as we've mentioned during our IPO roadshow, the primary targets for the 256-megabyte product were in SSD applications, whether they be PCIe cards or U.2 form factors and the 1-gigabyte chip we envisioned was also going to be able to address the SSD market, but we really had more visibility into that -- the use of that part for the rate-controller market. Now given the timing on the time horizon of our products with our 256-megabyte getting qualified in -- by the end of the quarter, Q1 this year, and some pretty good progress that we made in getting early samples out of our chip, we obviously, had customers that were capable of not only interfacing and evaluating the 256, but they wanted to evaluate the 1-gigabyte chip. And in some of those cases, we provided that silicon quite early, and of course, with some errata as we had anticipated we were going to have to do some revisions on the chip. And some of those customers got comfortable with the result in our ability to deliver that product on their time horizon, and have made decisions to move to the higher-density part, which offers them some performance advantages. The other situations that are kind of happening, which are pretty typical, in the industry whether you're selling controllers or memory, is customer projects tend to kind of sometimes stretch out or go into stasis for a little while and restart later. And in some of those cases, when the design goes on hold and then reemerges the timing might be quite different for when that product is planning to go to production. And they may choose at that point, considering whatever their cycle is, to transition to the 1-gigabyte product, and we are -- we do have some customers that are considering that, but we don't have any conclusion to that yet. So those are some of the high-level reasons why, obviously, they might transition based on the timing. And Jeff, do you want to address his other question?

Yes. So I mean, clearly the timing of customer rollouts has the biggest impact in terms of revenue timing. If your question is around kind of ASP differential between 256 and 1 gig, it's an interesting market that we are selling into. We're selling into a replacement market where we're replacing existing DRAM batteries and super capacitors with our MRAM chips. The density of those drives and those buffers, even though they may go up, doesn't have a significant cost increase to the customer because DRAM from a scaling perspective, there is not a huge difference in cost in the different densities of DRAM. So because we have to sell into that replacement market and it really drives the ASPs that we can get for our products. Now moving from 256 meg to 1 gig, one of the ways that we take care of that is we do the 1 gig on a 28-nanometer process, which significantly lowers our cost of die, and so we're able to kind of keep up with that from a cost perspective but we're not going to get significant ASP upside moving from 256 to 1 gig. And so even though some of those designs may move out, it's going to delay the revenue timing a little bit. We're going to get a little bit of a premium by them choosing the 1-gig product, but it's not significant.

Got it. Got it. That was helpful commentary just to kind of level set where things are. I guess I wanted to ask one follow-up question about the embedded MRAM work that yourselves and GLOBALFOUNDRIES are doing together. And, Phil, maybe you could expand on that a little bit. It sounds like some compelling updates to that engagement and also just kind of remind folks that are listening to the call here, I think that this a market that goes after some of the embedded flash type markets and -- or my team continues to do more work about the, I guess with regards to the challenges of scaling embedded flash down to some of these smaller geometries and what that might open up for MRAM as an embedded product market over time, whether it's your solution or others. But it seems like that's a market that's really on to come and you guys are at the forefront of it. So any additional color there would be really helpful.

Yes. thanks, Matt. So, yes, we're extremely excited about the embedded MRAM market. This was something we envisioned quite early with Spin Torque. In fact we believe that the concept of being able to take an MRAM technology and target it for different kind of memory use cases such as replacing embedded flash at small geometries or perhaps using it as a more efficient SRAM cell or even, for that matter using it for embedded MRAM. As Everspin's already highlighted in our products, you can optimize the MRAM bit to work like an SRAM like, or Toggle Generation 1 products, or you can optimize it to look like DRAM like we do in our Gen 3 Spin Torque products, and you could take that same kind of versatility onto an SOC if you incorporate it into a foundry process like GLOBALFOUNDRIES has done. And so with the advent of other foundries entering the embedded MRAM space, we're seeing a lot of customer traction in various application cases with GLOBALFOUNDRIES deploying it first in their fully depleted SOI or 22-nanometer FDX. Clearly, their initial targets are low power applications, things like controllers for -- IoT type controllers, which is a very good growing market, but there's also other process nodes that GLOBALFOUNDRIES has, like they have a 14-nanometer FinFET. They've announced their 7-nanometer FinFET and each one those process nodes target different types of applications. Obviously, a 7-nanometer is more along the line of the processor process, whereas maybe 14-nanometer and even their 28-nanometer technology nodes might be more targeted towards application processors. So, what's exciting there is in each one of those process nodes, it's quite possible for a foundry like GLOBALFOUNDRIES to retarget the bit to perform something other than just embedded flash. It could perform as a SRAM cell or it could perform as a IP core memory block that could store data and code simultaneously, in other words, a persistent SRAM. So that allows you to get rid of the embedded flash and the SRAM process that you need for an SOC and just use one memory technology. So, I think this a market that's going to evolve a lot more over time and we'll see the adoption of embedded MRAM into further and smaller geometry process nodes in the future.

Got it. And I'll just sneak one last one in for Jeff. Maybe you could talk to us a little bit about how you see operating expenses over the next couple of quarters here with some of the changes to the -- maybe top line expectations here, anything changing on the OpEx line that we should be aware of?

Yes, sure. So the OpEx, I don't see a lot of major changes. The big variables in our OpEx are primarily on the R&D side and they have everything to do with where we are in the qualification of products. So we've seen a little bit of an uptick in R&D spending in Q1 and in Q2. That's primarily related to both the GLOBALFOUNDRIES joint development expenses to develop the process technology on 28-nanometer as well as our own internal costs to do spins of mass and radicals and test wafers to get our products qualified. We're very much in the middle of, as Phil talked about, qualifying our 1-gig part and we're working really hard to get revs of that ready for production, so that we can get this product out and have good yield and be able to generate good margins. So, that work continues, we're heads down in R&D really working on that process. And so I don't anticipate huge changes in terms of our OpEx spending levels because we're continuing to focus on those things. SG&A spending is -- has increased over last year simply because we've become public. We've spent a few dollars since then in terms of our own staffing and getting ready to really go deploy the Spin Torque family of products and so I anticipate that there shouldn't be major changes in that spending as well.

The next question comes from Robert Mertens of Needham.

I'm here on behalf of Raji. So, some good questions before me but I guess, I just wanted to take a step back and look at the overall market for your ST-MRAM products. We talked about how they might get pushed out a little bit due to customers maybe switching over to the 1-gigabyte product instead of the 256-megabyte. I guess I wanted to wrap my head around if this is more in your view just of a timing issue, if you're thinking that it's going to push out a quarter or 2 or perhaps even a little bit further.

Yes, I mean I think this is still kind of consistent with prior remarks that we made is that, that kind of timing is still an unpredictable factor here. It's really all dependent on customer schedules, customer program plans, their take-to-market plans for their products, when they started evaluating using our either early engineering samples or whether they started evaluating at the time that we make a production sample available. Every customer engagement is unique and has its own kind of life of its own, and so it's really hard to give you a kind of general commentary there. We highlighted in our prepared remarks that one of the design wins that we talked about previously is an atypical situation, exactly the point I'm trying to make. You would not normally predict that prior to having your production samples available that you're going to get a design win for a product, and of course, we've had to work extremely close with this customer, and -- because they know that we're also making revisions to the product and getting it ready for production. As we announced earlier the 1-gigabyte will be released for production at the end of Q1 2018. But they're obviously moving ahead very aggressively with their design at a quicker pace than that so they could take advantage of exactly when we are in production to start launching their products. So, yes, it's really hard to predict that number for you. We've given you some ranges out there, which I think are reasonable, but you could certainly have outliers, you can have something as aggressive as what you see with this 1-gigabyte customer and you can also have customers that for various reasons their development gets dragged out to longer than 12 months, and other than us providing support and product on time, there is not much we can do about it. We have to let the ebb and tide of their development flow and basically work with them.

Okay. Great. That clears it up some. And then I guess as a follow-up, I think your guide was a little bit lower than what I had you guys at for the next quarter, maybe chunk of that is with the timing on the ST-MRAM product. But also just wanted to get your view on the progress with your Toggle products, if you're seeing continued design win within this product and if you could -- if you do break down sort of where those design wins are? What markets are ramping there?

Yes. So we highlighted one of the strengths of the quarter has been the increase in design wins that we see for Toggle from the prior quarter. Also, this has been a trend so far this year, we've seen our bookings strength to continue not only during Q1, but obviously, moving where we are today. So I believe you see some pretty good consistent growth there with Toggle. Jeff kind of highlighted some of the numbers somewhere around the 12% to 13% improvement, and I think that trend is pretty solid right now. The thing about the Toggle business is that, that product, once those go into production they're fairly stable and a little bit more predictable. We have a very wide customer base and so there is some variability, like Jeff has highlighted, in mix because a big part of our cut business base is through distribution. Some are on and some are off in a given quarter, and some will buy a lot and some may not buy in that quarter. But on an average trend, I continue to see Toggle as being more and more interesting to customers as they transition their designs from perhaps using a NOR Flash product or an eSquare product or an SRAM with battery back-up. As they refresh their products they're now looking more seriously as MRAM has an alternative. And because we do provide a superior solution, we do get to see that many of those applications transition to MRAM.

The next question comes from Richard Shannon of Craig-Hallum Capital Group.

Let me just ask a question on the guidance for the third quarter. I think I have got an idea but would love to help you -- have you help us nail this down a little bit better. But in terms of the growth from second to third quarter here, sounds like most of the growth here is coming from Toggle. I would imagine some from maybe 256, but wonder if you could help us out there and if there's meaningful change in the licensing NRE bucket as well quarter-on-quarter?

Yes. So, we really don't break down the components of that upside in guidance for Q3. I think I would characterize it at as we've talked pretty at length relative to where we are of qualification of Spin Torque and where we are relative to nvNITRO from a revenue potential perspective. We also talked about the fact that Toggle was a 17% increase from Q1 to Q2. That business continues to remain strong, continues to kind of be the bedrock of the company and we're working hard to continue to grow it because it is a great margin business and continues to provide cash to the company. And the whole NRE licensing thing, again it's really hard to anticipate what exactly those revenues are going to be unless they come from existing engagements. So we opportunistically take those when we can, but it's very up and down depending on the contract terms and the availability of that type of work. We just -- I think in general, we continue to see our base business growing at a good clip and on the spending side, as we talked about a little bit earlier, we continue to spend especially in the R&D area to try and get our Spin Torque products, both the nvNITRO family and discrete units, ready for production.

Okay. That's fair enough. Maybe hitting also on the gross margin side here. I know you talked about kind of the longer-term model thing unchanged despite the yield improvement here, but you said those would benefit future quarters. So at least on the product side, when those have some meaningful benefit above what you've seen at least on a product basis, excluding licensing. So could we see something near-term in the mid-50s or potentially a little bit higher in the near term?

Yes. Well assuming the product mix stays the same, I would agree with that assessment. But again, product mix is a pretty dynamic thing for us. From a long-term perspective, we always target 48% to 52%. Our immediate product mix and the fact that we have higher margins as a result of this yield increase are certainly helpful in terms of having us operate above that. We saw that in Q2, assuming we don't have a major change in our product mix then we certainly have an opportunity to beat our target for Q3.

Okay. That's kind of what I thought. That's helpful. Phil, maybe jumping over to you and talking about this customer that's been very early -- taken a very early transition to a 1-gig product. I know you've described this as an atypical situation, but are there any underlying dynamics that are -- something that could be extrapolated to what you might see from other customers or maybe what you're already starting to see? And help us think about how fast and how much interest there is out there for the 1 gig part in these enterprise applications?

So, Richard, as we've highlighted frequently, and I said a little bit earlier, the 1-gigabyte does obviously bring into play the DDR4 compatible interface and it also opens up opportunities for us to address some of the lower-end RAID applications, and being that we do have compatible RAID controllers that are out there, like from our ecosystem partner Microsemi, we see certainly the opportunity and interest and traction now in these RAID customers now that the 1 gig is in this early sampling stage and becoming, of course to them more real. As we've always highlighted the RAID market is frequently and often timed with Intel platform launches. As they launch a new RAID system with the next platform from Intel, they have to start getting samples at a certain point in time, and then they have to start doing all their testing and validation and then they want to typically launch their product. Frequently, we see that happen in the spring, early summer timeframe of a year. And so I think what we're seeing there is certainly now we can have those engagements with those RAID opportunities and in fact, we're having one with a very large OEM right now for our future RAID product opportunity, which we're excited about. But we have -- there's a long way to go to get this project sealed up, but obviously, they see the benefit of allowing them to eliminate a very costly battery that they have in their server that they have to use to protect the RAID cash. As far as the 256-megabyte goes, we've always felt that, that was a solid solution for the SSD RAID caches. We've seen traction in that area. We have quite a few customers that have been evaluating the product for that space. The same thing with the RAID controller, you have timing of the availability of SSD controllers and what's happening now is some of our ecosystem partners are offering the -- their SSD controllers with DDR4 interface, which is allowing the customer now to have an option to either use 256 or use 1 gig and again, depending on their product cycle and scheduling, which sometimes they'll have a program and then they decide to delay it or put it on hold or even cancel it and then they resurrect it again and then when they resurrect it, they decide, hey, well we're really close to your 1 gig so let's switch to that. So those types of things are really unpredictable. We just have to -- our sales and marketing team has to stay very close with our customer. We continuously update them on our product availabilities and schedules, and we try get from them, what's their timing and limitations and we don't necessarily see anything that says, hey, the 1 gig performs better per se. It's just -- I think what we're seeing is it's more of a timing issue now of when their product is planning to be out in the market and when we can deliver production parts.

[Operator Instructions] Seeing no further questions, this concludes today's question-and-answer session. I would like to turn the conference back over to Dave Allen for any closing remarks.

Thank you, operator. Everspin will be participating in 2 investment conferences this quarter. Later today we will be presenting at the 37th Annual Canaccord Genuity Growth Conference here in Boston. Our presentation will be webcast live at 2:00 p.m. Eastern Standard -- Eastern Time today on the Investor Relations section of Everspin's website at, www.investor.everspin.com. A replay will be available for those unable to listen to the web -- live webcast from our website. On September 6, Everspin will also participate in the Roth Data Center and Intelligent Infrastructure Corporate Access Day in San Francisco. Investors interested in this event should contact their Roth sales rep to request a meeting with Everspin. Thank you for your interest in Everspin, and goodbye for now.

The conference has now concluded. Thank you for attending today's presentation. You may now disconnect your line. Have a great day.