Okay, it's 11:00 A.M. U.S. Eastern. Coming up next in today's RedChip Future Tech Investor Conference, Everspin Technologies, ticker MRAM on the Nasdaq. Presenting today will be President and CEO, Dr. Sanjeev Aggarwal, and CFO, Bill Cooper. Everspin, are you there?
Yes, I am here. I think Bill may not be able to join us.
Okay
Just Sanjeev.
All right. Sanjeev, great to see you. Let me get some preliminaries out of the way, as I'm doing that, if you're going to be showing your presentation today, could we start putting that up on the screen now?
Sure.
Thank you very much. As is always the case, we welcome your questions, everyone. We're going to keep your lines muted, but we do want your questions. To submit a question on Zoom, go to the bottom of your screen. There's a Q&A button. Click it. A text box will appear. Type in your question and push Enter, and we will get it. We appreciate that. Before we begin, let me get the safe harbor statement out of the way. This segment may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. All statements pertaining to future financial and/or operating results, along with other statements about the future expectations, beliefs, goals, plans, or prospects expressed by management, constitute forward-looking statements. Any statements that are not historical facts should also be considered forward-looking statements. Of course, forward-looking statements involve risks and uncertainties.
Sanjeev, please go right ahead.
Thank you so much. I really appreciate the opportunity to present Everspin's story at this RedChip conference, or Future Tech conference. Same safe harbor statement. Thank you for going through that. In terms of who is Everspin or what is Everspin, we're the leading provider of MRAM, which is basically a magnetic tunnel junction-based memory. We've made products for mission-critical applications in data center, in industrial, in IoT, in automotive, and in radiation hardened applications. The reason why we've been successful in these sectors is basically because MRAM is naturally radiation immune. It's a magnetic spin-based memory, so therefore it does not get corrupted by radiation.
Also, it is a very fast read and write memory, because of which it's used for applications in automotive, like a black box, or data logging in industrial automation, or in data centers where you want to protect data in flight. I'll get through some of those examples here in the near future. A few highlights. We've been in production for almost 20+ years. Our first product was actually manufactured in 2006 when we were still part of Motorola, Freescale. We spun out in 2008 and got listed on Nasdaq in 2016, with the ticker MRAM or M-R-A-M. We have shipped about 200+ million units, and we have about 2,000+ customers, so a very diversified customer base across the world. I'll show a few examples coming up. We are very focused on maintaining our patent portfolio. We have about 700+ patents and applications worldwide.
Another couple of things I want to hit over here is we do have our own 8-inch or 200 mm MRAM manufacturing line in the U.S. It is actually a line that we lease space from NXP, and we do the MRAM layers over here in addition to the copper layers before finishing the wafers and shipping them out for packaging and final test. Recently, we announced a foundry services agreement with Microchip, where we will actually duplicate the capabilities that we have over here in NXP, therefore giving a second source to our customers that are concerned about supply chain concerns. We have our 300-millimeter line. We actually use GLOBALFOUNDRIES to manufacture our spin-transfer torque, or STT-MRAM. We entered into a joint development agreement with them in 2014, and we've been shipping product from that fab in Singapore and in Germany since 2017.
Now we are planning to bring on TSMC as a second foundry, and we plan to tape out our first product over there towards the end of the year. If you look at the maps, our headquarters are actually in Chandler, Arizona, right here. We have a sales team that's actually based out of Austin, Texas. We have a sales and marketing team worldwide, with our packaging and final test team based out of Taiwan. What are some of the highlights for Everspin? As I just mentioned, we have an 8-inch fab to manufacture MRAM in the U.S., and we are the single domestic provider of MRAM for mission-critical applications. That puts us in a very strong position with the U.S. government to support many of their programs and deep space programs for the commercial market as well.
We have a very diversified blue chip customer base that I just talked about 2,000+ customers, and I'll give a few examples in a few minutes. We have a large TAM, a large market opportunity, about $4.3 billion by 2029. We're basically expanding our TAM from $1 billion currently to $4.3 billion by the time we get to 2029. A proven management team with extensive experience. We have people from Samsung, from Intel, with background from Intel, Samsung, AMD, and that actually brings a lot of strength and a lot of diversification to the executive team to manage this company. We have a strong financial position. We have zero debt, and we have a free cash flow positive for the last four or five years. Let's get into what is the value of MRAM. The bottom line is we are persistent.
What that means is you do not require a battery to maintain the information that you store in this memory. Also, you can write and read to it unlimited number of times. It basically supports large memory workloads, like a DRAM or an SRAM case. Also in terms of performance, we are as fast as SRAM and DRAM, and in terms of reliability, we are best in class. Basically, we go all the way down from -40 to 125 degrees Celsius. Automotive applications, industrial applications, and obviously commercial applications as well. If I was to have one line to say what this memory is all about, we are basically a CPU-attached memory like SRAM or DRAM, and we bring the non-volatility or persistence of flash. We basically combine all three capabilities of these memories into one memory.
As close to a universal memory that you can get to. How have we commercialized it so far? We actually categorize our products into three different categories: PERSYST, UNISYST, and AgILYST. AgILYST is still under development. I will not spend too much time on that. PERSYST has been in production for the last decade and a half, actually almost 20 years. What the idea over here is, it is a persistent data memory. You don't need any battery to retain the data. It is very, very fast read and write, and it operates under extreme temperatures, automotive, industrial, all the way through to 125 degrees Celsius.
This memory actually serves. Toggle MRAM is the first product that we commercialized in 2006. This serves the customers like industrial automation, data centers, networks, medical. I will get into some of that in the next couple of slides. The next product that we actually built was for the DRAM industry, what we call the STT-DDR interface. The idea is, we have been shipping 256 Mb and 1 Gb DDR interface memory into IBM since 2018. These memories are used in the FlashCore Module. The idea is, it is used in between the storage memory and the working memory. It basically captures all the data in flight. It is actually a buffer memory that protects any working data that you have in the DRAM.
More recently, we introduced the xSPI family, or the serial peripheral interface, or what we call the EMxxLX family, based on our spin-transfer torque. We have been shipping these parts since 2022, 2023 timeframe. They basically allow us to extend the Toggle MRAM roadmap as well as address some more applications because these densities actually go from 16 Mb, 256 Mb, where the Toggle MRAM basically stopped at 16 Mb. More recently, we have been planning on bringing out a UNISYST, or basically which combines code and data memory. Today, most companies are using NOR flash for this application. NOR flash stopped scaling at 40 nm CMOS, and that is the reason why most of the foundries jumped in to offer embedded MRAM instead of embedded flash.
There is also a need for standalone or discrete flash, which stopped scaling. You can only buy densities like 256 Mb-512 Mb. We are bringing on a first product at the end of the year at TSMC on their 60 nm node, which will be a 256 Mb NOR flash replacement. It is actually an X node because it is three orders of magnitude faster than NOR and also two orders of magnitude more endurance or more rewrite cycles compared to NOR. This product at 256 Mb should be introduced by the end of the year. Then we will scale it to higher densities going from 256 Mb all the way to gigabit. Like I said, AgILYST is still under development. This is what is going to target the data center as well as AI inferencing on the edge.
It's at least two to three years out. PERSYST is in production today. UNISYST will go into production early next year, sampling by the end of this year. What does that do to our TAM? Like I was talking about, today, our TAM is on the order of $1 billion, and you add this UNISYST category that I was just talking about, and we increase our TAM all the way from $1 billion - $4.5 billion. Basically, the standalone NOR flash market at high density, 256 Mb and above, is on the order of $3.5 billion today, and it's growing very rapidly given all this focus on AI inferencing and memory workloads. I want to give a few examples of where our memory is being used today. One is these black box, and like I was talking about earlier, it's all about data logging.
How fast can you record the data in case of a mishap, and then can the data be protected at extreme temperatures? That's where our MRAM is being used. It's being used in EVs. It's being used in black boxes with combustible engines as well. Medical. Same example over here. If you're trying to, for example, monitor the performance of the pacemaker in the heart, you want to collect as much data as possible, as quickly as possible. That's where we have some design wins in the medical industry as well.
Similarly, for automotive, if you're actually trying to monitor the health of the battery in EVs, like the Bugatti, the Lucid Air, which is where we are designed in, or the BMW race engine, you want to understand how the battery is performing as you're accelerating or hitting the brakes, so we can then optimize the performance of the battery for the performance on the car. Another good example is casino gaming. Here, the idea is we are actually designed into the slot machines across the world. The idea is, today, if you're actually using a slot machine, you hear a lot of music between any actions that you can do on the slot machine. That's because it takes very long to record the data from this action that has been taken on the slot machine.
We've been slowly replacing this memory with MRAM, and what that does is it reduces that music time, and you actually can now have several more actions. You could go from three actions a minute to almost 10- 12 actions a minute. Three orders of magnitude faster. Now, like I said earlier, our memory is actually radiation immune, so that makes it ideal for space and satellite applications. Any low earth orbital satellites that are being launched today, we can talk about Blue Origin and Astro Digital, but if there is a satellite that's going up today, it is using our spin MRAM. We are the only ones that can offer high density MRAM for these applications, all the way up to 128 Mb- 256 Mb. A traditional design win for us is in this PLC modules.
Think about a manufacturing floor where a central computer is managing the robots on the floor. In case there is a power loss, if you're not using MRAM, basically the robots don't remember what they were doing, and all the work in progress is actually scrapped. With our memory, if you're using MRAM, it's insert on, insert off. If there is a power interruption, the robots start exactly where they stopped. These are similar examples. This is with the power management, same idea, data logging, and also the battery managing units that I just talked about a couple of minutes ago. Let's talk about how we are actually impacting the aerospace and defense industry. Good examples over here is, like I just talked about, low earth orbital satellites.
The idea over here is, the satellites need to communicate with the ground control, and the time window available is usually very short. If you're using the standard NOR flash, it takes forever to upload and download the data, and sometimes there could be an interrupt because of failures. If you're using MRAM, it actually reduces the time by three orders of magnitude, lower power, much more reliable upload and download of data to the satellites. That's the reason why all the constellations today are starting to use MRAM instead of NOR flash. Another good example is FPGA programming in defense systems or navigation control. It's the same concept again, radiation immunity and the ability to actually quickly download and upload information. Another good example that I'm excited about is eVTOLs or drones.
You can see that we're designed into them or into the flight control systems. Again, the idea being radiation immunity and rapid data logging and updating the system of the eVTOL or the satellites up there. A good thing about rocket and mission control, I'll go to the next slide. You can see over here today, we are actually on the Mars rover. We actually went up there in 2020. We've been designed in, and it's actually actively collecting data. We are on the way to Jupiter on NASA's Lucy mission. Both of these is because of the radiation hardness or the radiation immunity of MRAM, combined with partners like Frontgrade or Honeywell, where they radiation-harden the periphery or the silicon, and then combine the devices, actually radiation-hardened or radiation-immune.
Like I talked about earlier, we are in EVs, in the powertrain systems because of the rapid data logging and response to the battery life in these cars. Give you an example of some of the customers that we're designed in. On the enterprise side, we are designed in with IBM. This is the FlashCore Module that we've shipping our spin-transfer torque or one gigabit part, since 2018, 2019 timeframe. Broadcom, Microchip, basically all the hyperscale applications use MRAM because of the rapid data logging and protection of data in the data servers. In industrial automation, Siemens, Schneider, Omron, Mitsubishi are marquee customers over there. Again, they're used for managing the manufacturing floor. In medical, casino gaming, oh sorry, returning to m edical, GE HealthCare, Nikkiso are some of our examples. In network, NXP, Supermicro, Cisco, Lutron, some of the examples.
In casino gaming, IGT, [Crypsen and NOVOMATIC. The idea over there is we basically enable a better experience for the customers, and in the process, more revenue for the casino owners. Finally, because of the reliability that we have in mission critical applications, we are designed into several aerospace and defense applications. Airbus and Bombardier, BAE Systems, Frontgrade, they're all using our technology because it's radiation immune and it's reliable from a performance perspective. What is our business model? We actually offer services end to end. We actually do our own design. We have a design team in Austin, and then we do our own manufacturing on our 8-inch line, and then develop the technology, and then transfer it to a foundry for 12-inch manufacturing. Let's go back to our design services. We have designed all our parts on our own.
The xSPI interface that I was talking about, the Toggle MRAM, the DDR interface. We have a complete design team in Austin that can support this. In addition to that, we have done some custom STT-MRAM for the U.S. government and some of our customers. We have a full chip enablement team that actually supports our customers in case they need it. We have an 8-inch manufacturing line here in Chandler, Arizona, that's been in operation for 20+ years. The idea over here is we are the only provider of MRAM, or manufacturer of MRAM in the U.S. We support several trusted U.S. government programs as well as commercial programs out of our line over here. Also, we do all our R&D over here on our 8-inch line.
The reason is it's less expensive, also we can manage our IP so that there is no leakage. For 12-inch, we have been transferring our technology to our GLOBALFOUNDRIES. We are actually shipping products today at 28 and 22, and we are targeting a 12 nm product with GLOBALFOUNDRIES. Like I talked about earlier, we have a 16 nm MRAM plan with TSMC as well. Gives you an example of the licensing that we've done so far. We do have magnetic sensor technology that we've licensed to Alps and Bosch, for Mil-Aero, Honeywell, and Frontgrade. For our STT-MRAM, GLOBALFOUNDRIES. Any parts that GLOBALFOUNDRIES ships today, we generate royalty from the shippings. For head sensors, our technology actually applies to Seagate, TDK as well, for all the head sensors that have been shipped in the industry today.
This shows our roadmap. Basically, the PERSYST family that has been shipping for the last 20 years. You can see we go all the way from 128 kilobit all the way up to 1 gigabit with different interfaces, parallel interface, serial interface, also our DDR interface. On the UNISYST, we're actually bringing out this part that goes from 256 Mb- 2 Gb that will address the NOR flash market, also in the future, the LPDDR interface to address the AI as well as the data center markets. Finally, with this last slide, I'll talk about, we have put out a marker over there. We intend to increase our revenue from $50 million all the way up to $100 million by 2029. The idea over there is we already have these persistent products in production today. We expect them to grow significantly.
We have about 10%-15% licensing revenue, the examples that I just gave in the previous slide. The new product that we bring to market, UNISYST, will help us hit that $100 million mark, serving the FPGA market, the low Earth orbital market, and the aerospace and defense market. I want to thank you for your time, and I'll take any questions.
Thank you very much, Sanjeev. As we can only take your written-in questions today, audience, we ask that you not use the raise hand button, but we would invite you to use the Q&A button at the bottom of your Zoom window. You can see the text box that will appear, and you can submit your question that way. We already have several questions, Sanjeev. Many of your end markets have very long qualification cycles. As programs move from design win to production, how does the revenue opportunity typically evolve over the life of a customer relationship?
Yeah, that's a good question. Typically, for example, the part that we're bringing to market at the end of this year. What you'll see is that we will start giving out engineering samples to our customers so that they can start understanding the technology and to understand how they can actually integrate that technology or that product into their system. We provide them with qualified parts following basically in the next three to six months. The customers would go and qualify them in their systems. That takes about another 12 months. I would say a total of 18 months from sampling to qualification of product. At the customer, these design wins would ramp to production. You'll start seeing revenue typically in 24-30 months is when we start seeing revenue from the customers.
Most of our customers are actually, they're in love with our technology and with our products. We typically get requests for at least seven-year production lifetime. These are repeating orders that we get. It's not one and done. We have developed these customers over the last 20 years, and we have a good relationship with them.
One thing that stands out, Sanjeev, is your presence in aerospace, defense, industrial, and transportation applications. What are the characteristics that make those customers particularly valuable from a long-term business perspective?
Yeah, that's a good question. Like I mentioned a little bit during the presentation. Our technology is actually immune from radiation, and that's what makes us very attractive to the aerospace and defense industry. Our technology is based on magnetic spins, they're not corrupted by any radiation. That's the reason why we get designed in over there with the aerospace and defense. The other thing is how do we differentiate compared to standard technologies? Our data logging speeds, we are at 25 nanosecond- 35 nanosecond read and writes, compared to flash, for example, that takes a microsecond to read and write, three orders of magnitude faster. Differentiation, the features of the technology, and finally, our reliability. We've been shipping products for the last 20 years, and we basically have a handful of returns. The customers really value the reliability.
We have a culture of innovation and quality within Everspin, I think that shines through the products that we provide to our customers, that brings our customers to us again and again.
Sanjeev, as you know, AI infrastructure continues to evolve, is there some sort of a misunderstanding about MRAM? Is there some way, this person wants to know, that investors may be underestimating the role that MRAM can play?
Let me start with where we are today already with AI inferencing on the edge. Today, many customers use NOR flash for weights on the edge to make all the inferencing on the edge. The challenge is, if you want to update those weights where they're using NOR flash, it takes forever, and that actually makes the system very inefficient. Many of our customers are actually replacing that NOR flash with MRAM for weights, which basically enables them to update the configuration of that system within minutes instead of less than a minute, as opposed to 30 minutes. That allows them to do more efficient inferencing on the edge.
The other example is if you have memory that can actually read and write as fast as SRAM, you can actually use it for compute on the edge. We do have that technology available. Why would you use MRAM as opposed to SRAM on the edge? First of all, SRAM is very, very leaky. The transistors are extremely leaky, and then you're constantly taking the data back from the SRAM to the cloud to make a decision and bring it back. If you can partition the memory on the edge, use some of it for SRAM, use some of it for weights, you don't have to waste power going back and forth to the cloud. Also, you can actually have zero standby current with MRAM on the edge, unlike SRAM, which you have to always keep powering up.
All of those combined, I think MRAM has a huge part to play on the edge or with AI inferencing. Of course, we don't have a product out there today other than the MRAM weights that actually addresses this AI inferencing on the edge. That's what I said, it'll take about two years for us to bring that first product to market once you have a customer signed up. We do have the technology ready to go.
Okay, Sanjeev. We've been getting a lot of questions around this. I'm going to kind of conflate them all.
Sure.
You've been talking about product revenue and licensing opportunities. How do these two work together? What are the opportunities there, please?
That's a good question. We spend a lot of time and energy and money to maintain our patent portfolio. The idea over there is we want to maintain our leadership, and then we want our customers to be able to take advantage of that leadership and of all those patents and all the technology that we have developed. A good example is we are selling memory products today, but we also developed the magnetic sensor technology or the tunnel magnetoresistance, or TMR sensors. We have, for example, licensed that technology to Alps and to Bosch, and we actually get revenue from there, as well as royalties anytime these parts ship. The other example is in the read arm industry, where we licensed it to Seagate, as well as, I'm forgetting the other one, Toshiba.
The idea over there is we also get royalties, and we got some licensing revenue over there. The idea is, by licensing our technology and by maintaining a patent portfolio, we are communicating to our customers that we are always looking and developing a leading edge technology to support these guys. Similarly, we license our technology to GLOBALFOUNDRIES so that they can ship embedded MRAM to their customers while they serve as a foundry for us for discrete MRAM devices. Overall, our goal is to have about 10%-15% revenue from licensing, and we are majorly focused on being a product revenue company, and that's what we are focused on, and that's what we want to grow over the next four to five years to hit the $100 million mark.
Sanjeev, as you look out across the roadmap there, your future, what's exciting you the most? What do you think is going to be the most transformative over time?
I think, in the near future, this huge TAM that we've increased from $1 billion- $4.5 billion by addressing the NOR flash market. There's no memory technology that can address this market other than Everspin. I'm excited about that. In the long term, I think as the systems people understand the value of MRAM, where we are breaking all the walls that traditional memory faces, I think we will see MRAM take a larger role in compute, in AI inferencing, in data centers.
Sanjeev, thank you very-