All right. Hello everyone, and welcome to the Everspin Technologies fireside chat. Again, my name is Robert Blum, Managing Partner here at Lytham Partners. And up next here, I will be moderating a Q&A discussion with Sanjeev Aggarwal, the President and Chief Executive Officer of Everspin, and Bill Cooper, Everspin's Chief Financial Officer. Quick reminder, Everspin trades under the ticker symbol MRAM on the Nasdaq. All right, let's get started. Gentlemen, welcome.
Thank you, Robert.
Thank you, Robert.
Fantastic. For investors who may be less familiar with Everspin, can you start with an overview of the company and really both your backgrounds as well in the memory industry?
Yeah, sure. Let me start with the background. Been in the industry for close to three decades now in the semiconductor industry. I started off my career working on ferroelectric memories, which I actually helped commercialize when I was at Texas Instruments. Then I moved to Freescale at that time, to join the MRAM group, and since then, helped spin out Everspin from Freescale. At that time, I was running the manufacturing operations, transitioned to running the joint development agreement with GlobalFoundries, and then managed the R&D group for a while, followed by operations before I became the CEO in 2022. Again, Robert, thank you for this opportunity.
Fantastic. And Bill?
Yes. Thanks, Robert. Bill Cooper, and I actually joined Everspin January of last year, 2025. Before that, prior to that, I was with Advanced Micro Devices, or AMD, for the better part of 20 years in the semiconductor space.
In terms of Everspin, Robert, we are a spin-out from Freescale. We spin out in 2008, to form Everspin. We are basically a semiconductor manufacturing or a product company that makes memories based on magnetic materials. It is a non-volatile memory, with very fast access speeds. The access speeds are as good as SRAM or DRAM, or data logging or the data center memory with the non-volatility of flash. You basically combine three different kinds of memories to form this technology. We've been shipping since 2008, actually since 2006, so almost two decades. We have shipped about 200 million + units, with a handful of returns, so pretty high quality. Also, we have about 2,000 + customers worldwide. Companies like IBM, Schneider, Siemens, Mitsubishi, so a pretty wide range of diversified blue customers.
We have about a patent portfolio of 700 + patents to actually support all the innovation and development that we do to bring new products to the market.
Fantastic. Thank you both for the overview and the background there. MRAM has been discussed for many years as an important emerging sort of memory technology. In sort of practical terms, what does MRAM do better than conventional memory, and where does that difference matter the most?
Yeah. So basically, MRAM is a very fast read and write memory, so you can basically access the memory in 25 to 30 nanoseconds, right? Compare that to a memory called NOR flash, where the write times are on the order of hundreds of microseconds. So there is basically a three orders of magnitude advantage to bringing MRAM into the NOR flash application. So why would you do that? If you are doing an over-the-air update, for example, in 2025 when all the Teslas were parked on the side of the roads to get the operating systems updated, that took about half an hour, right? If there was a failure, another 30 minutes. If you are using MRAM, that update would have taken less than a minute. It is basically as long as it takes you to turn the key in the ignition again, so pretty rapid.
So that is why you would use MRAM instead of NOR flash. The other application is in the DRAM industry, where you basically have DRAM is volatile, so basically if you turn off the power, you are going to lose all the information. With MRAM, it is persistent. You can actually turn off the power and the information is not lost. Because it is very fast read and write, so it is low latency, so it can actually protect data in flight. So that is the reason why you use MRAM instead of DRAM or NOR flash, and that is what we bring to the table.
All right, very good. You touched on these a little bit, but how should investors think about the company's major product families and the types of applications each is designed to serve?
Yeah. We have successfully commercialized MRAM in all three categories, for SRAM or the data logging memory. For example, in casino gaming or in industrial automation, we have products out there with customers like Siemens, Schneider, Mitsubishi, et cetera. If you're looking at the data center market, we have commercialized a 1- Gb DDR4, like MRAM, that is being deployed in IBM's FlashCore module. We've been designed in for the last five generations of their offering over there. Then we recently also commercialized our MRAM to look like a NOR flash, going from 64 Mb all the way up to 256 Mb with a new part that is actually planned to tape out at the end of this year at TSMC, and that'll take us from 256 Mb all the way to 2 Gb.
We have commercialized MRAM in all three types of memories, and they're being deployed in applications like lower orbital satellites, in FPGAs, in medical, in flight control systems. One thing that I forgot to mention, Robert, is that this technology is actually radiation immune or radiation tolerant. They are very apt to be used for LEO applications, or if the circuitry around the MRAM is radiation hardened, then you can use it in deep space. For example, Frontgrade did deploy our MRAM on the Mars rover, where we are actively collecting data today. Then also we are on our way to Jupiter on the Lucy mission because of the radiation hardness.
Very good. You've talked about some, we'll call them specialized applications, but where do you think MRAM adoption is today, and what needs to happen for the technology to move from these specialized applications into much larger markets?
Yeah. As far as becoming mainstream, I think the first opportunity that we have is in the NOR flash market. NOR flash stops scaling at 40 nm CMOS, and that's the reason why you saw all the foundries jump in at the 2x node, 22 nm or 28 nm node, offering embedded MRAM to replace embedded flash. You also need standalone flash. Because the technology stops scaling, the highest density NOR flash that you can buy today is on the order of 256 Mb, and in some cases 512 Mb. The densities that you need are anywhere from 256 Mb all the way to 8 Gb. So it's a natural opening for MRAM to actually serve that market. There isn't another technology that can actually replace NOR flash, to my knowledge.
This market is on the order of $3.5 billion to $4 billion over the next five years. So even if you capture 5% to 10% of the market, that's a huge inflection point for Everspin. The other one is the data center market. We talked about this in our prepared remarks for the Q2 earnings call. We are doing a demonstration of the CXL protocol with our MRAM technology. Once we do that in September, we're going to do that demonstration at the SNIA Developer Conference ( SDC), on September 28th through 30th. The idea is once that is done, you can then look at use cases, and there we are actually partnering with MaxLinear, which is a storage accelerator company, to see how we can deploy a CXL attached MRAM in their accelerators.
At least on paper, in theory, we believe that we improve the efficiency by almost 20% to 25%. So what does that mean? That means that the same server can take on 25% more workload, or you can use 25% less GPUs if you use the MRAM. So there's a cost saving or a power saving or a better workload that we can address with our technology. So I think we're getting to the point where we can become more and more mainstream, compared to a decade ago.
All right. Very good. What differentiates Everspin specifically from some of the other memory companies and, how difficult would it be for a competitor to replicate your technology, your manufacturing expertise and, sort of expand upon some of the customer relationships that you have?
Sure. I think, the way to think about it is that we are very conscious of the IP that we have developed. You can look that our patent portfolio is 700+ patents worldwide, and that is how we actually protect the technology that we have developed. It would be very difficult for somebody else to jump in and actually offer the same solutions that we offer. Also, a lot of the expertise or the know-how is actually in-house, and it is not actually filed in IP, so other people cannot actually go and develop that technology. A good example is our Toggle MRAM, the first- generation. We are the only ones in the world that can actually build that technology. Other people have tried unsuccessfully, so we are the only ones that can commercialize it.
STT-MRAM, or spin- transfer torque MRAM, has been commercialized by others like GlobalFoundries, took a license from us. TSMC took a license from TDK Headway. Samsung took a license or acquired Grandis, and they all have their own technology today that they offer as an embedded solution. But for standalone, we are the only company in the U.S. in the commercial market. Obviously, there is Avalanche that offers some solutions for the defense markets. Then in South Korea, there is a company called NETSOL that has one offering between 16 Mb to 64 Mb, for the commercial markets. So pretty much have the expertise and the broad product base that we have been offering for the last 15 to 20 years.
Okay. Taking a step back here, when Everspin Technologies is designed into a customer's system, how long can that opportunity generate revenue for and, what makes those positions difficult to replace? Bill?
Great question. I would say to the first part of the question, the normal design cycle, once we have developed a product, is going to be 12 to 18 months to get into a customer's systems and to start generating revenue at sizable volumes. One of the other really compelling attributes of the business is that once we are designed in, that business tends to be quite sticky. We actually have new orders sometimes that we will get for 10-year forward guarantee of supply. That is even from the originating product, the Toggle MRAM products, early Toggle MRAM products. So even today, 20 years in, we still get orders for 10 years out. So very sticky business.
Okay. No, that's good to understand here. We've talked about a couple of different end market applications, but I really want to understand what's, in your opinion, what's industrial automation, energy management. We talked about some of the lower orbital, the aerospace and defense and so on and so forth. What do you see as driving the demand across those markets and where do you see maybe the greatest opportunity going forward amongst them?
Yeah, I think, what I would say is we've been really strong in terms of the industrial automation, transportation, gaming, casino gaming in particular. We've also had success in the data center and aerospace markets with the PERSYST family of products that encompass Toggle and STT-MRAM. As we look forward, one of the very significant opportunities for us is in the satellite market, the LEO market, low Earth orbit satellite. Those products, obviously, the attributes of the speed with which we can upload or download information to satellite in that narrow window that they have as they circle the globe is compelling. Also, the radiation hardness or radiation-resistant attributes of the product also come into play there in that market. Even still, when you might have to use several NOR flash products in the older satellite model, now you can replace those with just one Everspin MRAM product.
We do see something like four to eight units per satellite on the MRAM products that go up in the satellites. Also, we did just announce a deal with Astro Digital on the GEO market, the geosynchronous orbit satellite. Astro Digital is essentially a provider of those ground stations. Again, what you have is really the compelling value proposition for MRAM products because of the speed with which it can transmit and upload and download information. I think that market, that aerospace market, is one that's growing and burgeoning very rapidly, as we've seen by some of the well-known players out there. We can talk about Astro Digital and Blue Origin, but again, we're a very strong participant in that market generally. Then as we look forward, we do have strong data center offerings, but we are also working on additional data center offerings.
We have a new product that we'll be taping out this year and that we'll be sampling and going into production with next year. That's in the PERSYST family of products. It's at a high density, so it starts off at 256 Mb, which is generally at the high end of, say, the NOR flash market. That 256 Mb will then scale that up to 1 Gb, 2 Gb and so on. Again, very compelling value proposition at those higher densities in both the aerospace and the data center market.
All right. I could probably touch on all these various programs, and one in specific that I do not think we have touched on maybe enough here is the recent $40 million agreement that the company entered into supporting military-owned aerospace applications here. Beyond just the near-term revenue contribution, what does that award say about maybe the strategic importance to the U.S. defense broader industrial base?
Yeah. I will answer that one as well. What I would say is, Everspin has been a long-term, critical supplier to the U.S. government and really the DIB customers, defense industrial base customers. That most recent award was actually an award as a subcontract agreement to a prime contractor to the U.S. government that provides technology and services. That award is specifically to set up the MRAM Toggle process at a location here in the United States. As we set that up, we will recognize that revenue over time. But really what it leads to is enhanced production capacity capability and R&D capability for future products, and a base for future product development. Even though this particular agreement runs for two and a half years, we will then see additional products that come out of that or as a result of that setup of that process.
But Robert, if I can add some comments to that. The strategic importance to us is we are the singular domestic provider of MRAM to the U.S. government. There is not another factory or another company in the U.S. that can actually build MRAM. That partnership we have been fostering for the last five years or more with the U.S. government, we have had several agreements and contracts with them to provide them radiation hard Toggle MRAM, as well as radiation hard STT-MRAM, or strategic radiation hard FPGA as well, through our partnership with QuickLogic as well. What we see over here is an opportunity to share our technology for mission-critical applications with the U.S. government. It gives us an opportunity to turn our technology into something that the U.S. government can use for its applications as well. So very critical partnership for us in my mind.
All right, very good. Let us transition just a little bit here. You have established additional U.S., as you mentioned, manufacturing capacity through Microchip. Talk about how that partnership expands your capacity, your supply chain resilience, and really the ability to pursue maybe larger customer programs.
Yes, excellent. What I would say is Everspin is really ideally situated in terms of its supply chain resilience. We have the original facility, the Chandler fab, that produces Toggle products. As you mentioned, we did this deal recently with Microchip as a foundry service agreement for them to supply Toggle products for us, and that was a 10-year agreement, and that factory should be up and running in the next couple of years. That was a critical agreement. In addition to that, we actually do have a long-standing relationship with GlobalFoundries, where they manufacture our current, today, our STT-MRAM products. In addition to that, this new product that I mentioned, the PERSYST parts, will be manufactured in partnership with TSMC. We have a very robust diversity and resilient supply chain.
All right. Very good. Maybe just take a step back here from a macro perspective, tech summit here, talking to a bunch of different companies. What are you seeing really across the semiconductor market in terms of customer inventories, the demand visibility, manufacturing costs, and this growing emphasis, if you will, on domestic production? Just what are you seeing out there?
Yeah. All very important points. What I would say is we do see very strong and healthy demand, right? If I were to compare to the pandemic times, which was great demand, but it was sort of hockey stick-ish, and it went up vertically, and then it kind of dropped off suddenly. Maybe not as healthy of a demand cycle. What I think we see this particular go around is very healthy demand. We don't see a lot of inventory build. I think everybody kind of learned their lesson again from the pandemic days when they overbuilt their inventories. What we do see is, and we can see out, right? We have a six-month lead time, so we get visibility to that and a little bit more from some customers as well.
I would just say, in particular, the evidence is, we grew 30% in revenues on the product side from Q1 of 2025 to Q1 of 2026, and then even higher than that, Q2 of 2025 to Q2 of 2026, we grew 38% in our product revenues. We guided even higher in Q3 as well. Again, really strong, healthy, robust demand, and I don't see an overbuild or accumulation of inventories at all out there.
All right. You just sort of touched on it there, but maybe in the final five minutes or so that we have, let's talk about the recent financials. Second quarter revenue reached a new company record, with $18.7 million. It was up 42%. Maybe talk about in some more details for those that maybe weren't on the call or didn't get a chance to listen to it. What drove that step up, and how much of the improvement reflects sort of sustainable underlying demand?
Yeah, exactly. What I would say is, again, really strong, 30% to 40% revenue growth year-on-year. Then we guided even higher in Q3, so we guided $19.5 million to $20.5 million revenue in Q3, $20 million midpoint. Most of that revenue growth will be driven by product from Q2 to Q3. Again, just really strong pull and demand out there, and I would say healthy demand.
Yeah, just to add to that, Robert, most of the demand that we're seeing is across both Toggle MRAM as well as STT-MRAM. We're seeing a strong pull from industrial automation, from energy management, as well as the low Earth orbit satellite markets. So those are our primary, or I shouldn't say primary, but the leaders in terms of revenue growth, in addition to the data center that we talked about at the last quarter. So I think we're continuing to see that demand. There's a strong backlog, so that's the reason why we guided up in Q3 and hope to see that going forward as well.
All right, very good. As we kind of close up here, looking out 12- 24 months here, what are the milestones investors should be watching for? Maybe what do you believe the market maybe underappreciates today about Everspin and sort of the broader investment opportunity?
In terms of what to look out from Everspin over the next 12 to 24 months is, we should see an announcement for the tapeout of the 256 Mb device at TSMC. That should happen sometime in Q3 or Q4 of this year. You should see a successful demo with our CXL protocol with the CXL attached MRAM. You already saw the MOU that we signed with MaxLinear. Hopefully, we will see some prototype or a use case that is identified, and we can talk about that sometime in 2027. You should see the 256 Mb part go into production mid-2027, with some early revenue towards the end of 2027, early 2028. In terms of if the industry is not recognizing Everspin or not valuing Everspin, I would say that the tide is slowly changing. People don't have an option. NOR flash stopped scaling.
If you need a higher density boot memory, that's Everspin MRAM. There isn't an alternate technology. There isn't an alternate provider today. When we have the successful demo with the CXL protocol or CXL attached MRAM, and we build the next higher density DDR part, I think we will start getting recognition and traction in the data center market as well. I think those are the growing sectors for us, aerospace and data centers, and we are focused on actually getting that going.
All right. Fantastic. Well, Sanjeev, Bill, thank you so much for your insights, your participation here on the summit today. Greatly appreciate it. Thank you, of course, to everybody for watching. We do have additional presentations, fireside chats coming up, so please stick around for more. Sanjeev, Bill, thank you again so much. Really appreciate it.
Super.
Again, appreciate the participation over here. Talk to you again soon.
All right. Thank you, guys.
Thanks, Robert.