Good afternoon, everyone. Welcome to Oppenheimer's Tech Conference today. In this session, we have the investor presentation from QuickLogic. We're honored. Of course, the CEO, and CFO, Elias Nader, of QuickLogic Corporation. Before we start with the presentation, there's a brief, safe harbor statement, that QuickLogic would like to share. Please go ahead.
Hi, yes. I just wanted to remind everyone that QuickLogic does report its second quarter results next Monday after the market close, and as a result, we're not gonna be able to take any questions today, and can't really speak to any financial results. Slide two does have a safe harbor statement, which I will not read to you. You can read it at your leisure, and in the meantime, I'll turn it over to Brian, to talk through the company's strategy.
Thanks, Allison, and thank you, Martin, for Oppenheimer hosting this event. We're happy to be here today. Elias Nader, our CFO, and myself are presenting on behalf of the company. As Elias, or sorry, as Allison mentioned, here's the safe harbor statement. We're not gonna go into details. You can read this. This is on our website now under the Oppenheimer conference presentation. Just to echo what A- Allison said, we won't be discussing any financials on the call today. We'll be talking about markets, and products, and drivers, and things like that, that we foresee to be very important for the company now and in the future. For newcomers to the QuickLogic story, we are a fabless programmable logic company.
We fall under the semiconductor category in the technology space, and within semiconductors, we focus on two fundamental areas that you can size the markets for. One is what we call FPGA technology or field-programmable gate array technology. The second is AI/IoT software. There are two sort of distinct product types. And then within the FPGA category, we actually have three types of products in there, and they have different uses for different customers. At the very bottom there, the foundation is embedded FPGA intellectual property. This is a semiconductor IP business model, where we take our IP, and we license it for use in customers' ASICs or their custom chip designs. The second product category, working our way up, is Discrete FPGA.
In the event that a customer does not want to or cannot design their own custom silicon, they can buy a standard product from us that encompasses that FPGA technology. The third in the stack is SoC. It's essentially an FPGA with processor cores, and we have some devices that fall into that category. They've primarily been used for lightweight computing in battery-powered applications like phones, and watches, and hearable devices. The last category on top is AI/IoT software. AI is all the buzz lately. This AI/IoT software enables our customers or partners to develop machine learning models that can be used primarily in battery-powered devices at the edge where the sensor resides.
You can imagine things like watches that need to have some sort of AI, ML software running in them, or biometric systems that people would wear for health monitoring, or predictive maintenance equipment that would get attached to big pieces of equipment. All those are examples where there's sensors, there's batteries, and people wanna do sort of local processing with machine learning technology, and our AI software is very good at helping create and deploy those types of solutions. That's the full stack. We talk about end-to-end solutions. The nice thing about having a different sort of spectrum of products is that if a customer does their own chip design or they don't, or they want some software help, we really have the ability to serve a very diverse set of needs of our customers.
Customers can traverse back and forth across those if they, if they so desire. We're headquartered in San Jose, California, and Elias and I are dialing in from Silicon Valley right now. Moving to the next slide, we didn't start with that full range of products when we started our company, a couple of decades ago, a few decades ago now. It wasn't until recently, in fact, the last four years, that we really started to lay the groundwork for where the company is today, from a product perspective.
In 2019, people were just now starting to think about AI and edge use cases, and we partnered with a company called SensiML to develop a capability for our customers to take our SoC product, the FPGA plus the processor core, and develop machine learning models that they could deploy in consumer or industrial devices with our devices. Came across SensiML. They were spin out from Intel, and they had grown a certain set of ecosystem customers and partners around AI. We decided that, you know, from a strategic perspective, it made a lot of sense to offer that as a package to customers, and we saw some of the traction they were getting. So, we came to an agreement, and we acquired SensiML in 2019 to sort of build out that software capability around AI.
I'll get more later to how people are using that today and progress from an ecosystem perspective. Moving to the last three years, a lot of the focus for us around the strategic side has been to build out and enhance our capability to automate a lot of what we do today on the design side, and change our go-to-market strategy so that we could effectively be a more agile and nimble company in addressing what we foresee to be a very large opportunity in FPGA technology, embedded FPGA, IP semiconductor devices. Part of that was about sort of redoing the way we do development internal to the company, and part of that was around how we redo, how we go to market as a company, to the different customer base.
Without getting into the specifics quarter by quarter of the last few years, that has culminated in today, and I should say today, meaning as of our Q2 earnings call, our Q1 earnings call in May. A lot of progress in that area for the pipeline of opportunities, so we have a healthy sales funnel, and we've had a lot of wins in the eFPGA IP specifically, that I could go a little bit more into details now. As far as the new deals go, a lot of those new deals that we have announced previously have come from the fact that we've automated our development process through the Australis IP Generator. That's led to this growing pipeline, the growing sales funnel.
What that generator does is it allows us to really shrink the development time it takes us to address customer needs. In shrinking the development time and the development resources, we can effectively address more customer requirements with the same team. Again, being much more agile and nimble as we go. That's important because what we've found is a lot of customers want to do an IP license model with us, where we do an IP design, provide it to them under license, and they go off and do their own chip design. Other customers are starting to, I'd say, appreciate the value of FPGA technology, but they don't have the chip design capability within the company, and so they often will come to us and say, "I like the benefits of FPGA technology.
Can you do a device for me that uses that?" That's sort of this top row here, where we talk about chiplets and devices that are either bespoke or custom implementations for customers. The nice thing is that we're able to leverage our 30 years of experience and market presence in FPGA technology. People see us as sort of a trusted supplier with that, then they start to see, you know, that we can tailor these solutions for them in a very expedient and cost-effective manner, and they like that. That's generated new opportunity and new, new revenue for us as a company. A lot of people will think about business model and try to apply, you know, what business models is QuickLogic bringing to market now.
I'd say that, again, this has also evolved over the last few years. Historically, FPGA companies, a lot of semiconductor companies, they don't monetize, engineering work done at the front end of any design. They do engineering work on their own dime, and then they license IP or sell devices to customers. What, you know, we were part of that group, many years ago. You know, over time, we evolved and we said, "You know, this doesn't make sense.
If we're doing work for customers, you know, it shouldn't have to wait till the end of the day, end of product delivery, to monetize that, that work and that know-how." We introduced the services revenue category to the mix, and so as we sign contracts and do work for different customers, we're monetizing that work in the very beginning through service revenue. Once the service is completed, and it could be that we've done a rev, a license for a customer, meaning we've designed an IP, it could be that we've designed a chip for a customer. License revenue is essentially when an IP is done, we deliver it to the customer, they have the right to use that in their own chip now, and that means that there's a royalty recognition event for the license.
Again, that's at the, typically, the completion of the service period. At that point, the customer can go off and do their own design. They can do their own chip design. Our IP is simply one part of that. Once that customer wraps up their chip design, starts manufacturing and using that in some production capacity, then we start generating royalties on that from the customer. That's typically, you know, 1-2 years after we start the license to the customer because they have their own design phase. The nice thing is that royalties can last for many years. We have one example where it's, I don't know, it's been, like, 15 years now that we've been generating royalty revenues from a customer licensing the technology.
As you get more designs that move into that royalty phase, you get a nice layering effect to get a longer-tail revenue in place. The last category there on the right is something that I think from a terminology perspective, we've introduced to our investors in the last, I would say, six months or so, but it's this notion of being a storefront for finished goods. The storefront is essentially, it's an alternative path for a customer from the, the license.
In the event that the customer either doesn't have the resources or the capability or the know-how or the desire to do their own chip design, and yet they still want programmable logic capability, they can contract QuickLogic to do the entire chip design for them, and then provide that to them, the finished goods to them on a, you know, per-unit basis, for production use. We're essentially the, not just the services company designing the chip, but we're also the company taking care of all the back-end supply chain, like procuring wafers and assembly and test, and basically providing that finished good to the customer. That one clearly happens at the end of the phase when the design is done, and they're ready to go to production.
The more opportunities that we can close and convert to either royalty or finished goods, you really, again, start to see that layering effect, for the revenue in the out years, and that's really what our, I would say, medium and long-term business model is about, is closing as many designs as we can now with services revenue, converting them either to license or to the storefront model, in the future, so that we can, in fact, get that nice layering effect. Now, the other thing I'll talk about on this slide is the gross margin percentages for these different categories is obviously different. Working right to left now, being a storefront for devices is typically gonna be, you know, for our type of device, you know, 60%+ gross margins. Royalties is almost 100% gross margin.
License is probably 85%-ish gross margin. Services revenues, you're never gonna get those kinds of gross margins out of services, those are gonna be, you know, below 50% gross margin. The blended mix, though, is where we're targeting our margins to be, you know, and I'm talking about, you know, a couple of years in the future now, to be, you know, the mid to high 60s for that. As from now till then, again, the more deals we close, the different knobs we turn on, and these different revenue pieces is gonna impact our gross margin. We have our eye on the prize there of getting into the mid to high 60s as a balanced gross margin or, a gross margin as an aggregate gross margin as a company based on these different revenue contributors.
That's the focus of the company. That's the sort of double-clicking on these different revenue categories, and we'll be using these terminologies as we, you know, go through the rest of the slides. One of the deals that we've announced last year, around this time, actually, was this deal to do a strategic radiation-hardened FPGA technology development for the DoD. This is the first time that QuickLogic has been a supplier to the DoD directly. In the past, it had always been through the defense industrial base. People, you know, the big defense contractors using FPGA technology from us directly, but not the DoD.
In this case, the DoD is funding the development of something that doesn't exist in the world today, which is a strategic radiation-hardened FPGA technology with other characteristics that I'm not gonna go into on this call. We're the lead, we're the prime contractor, if you will, and we're working with different subs to us to realize this technology. Across the bottom there, you can see SkyWater, which is a onshore wafer foundry or, onshore U.S.-owned and operated wafer foundry. Everspin Technologies, which is an MRAM company. MRAM is a non-volatile memory that's used a lot to store data in a non-volatile way. They have commercial offerings, and they have other offerings via partners for radiation-hardened applications. Trusted Semiconductor Solutions, which is an IP provider, really focused in on the needs of the defense industry.
All of us are working and collaborating to create this FPGA technology for the DoD. The total scope of the program is $72 million, and that would be across four years. The $6.9 million that you see referred to there is really for the first phase of the development, and that $6.9 million is, is a subset of that $72 million. We've announced the first one as an 8-K, because the details are obviously material in magnitude, and as we get further awards on this towards that full $72 million, you can just keep your eye out for 8-Ks, because 8-Ks are gonna be the way we typically communicate this, and not press releases, because press releases can take a while.
We're awaiting the next award and the next chunk that's gonna go towards the development of the strategic radiation-hardened technology. QuickLogic is clearly, I think, leveraging a lot of our expertise in this and our history in selling to the defense industry, and we're really excited to be part of this program of designing something that really doesn't exist today. The other thing I'll mention, and this comes from a lot of questions I received on this from different investors. The total scope of this $72 is only for the development of this technology and the device. It is not for any shipments of this to the defense industrial base for use. That would be separate storefront activity after this development has been done across these 4 years.
Our goal as a company is to be that storefront for that device, and we're fully intending to do whatever we can to be given that responsibility, because that indeed is really where the rubber hits the road in terms of revenue potential for this in the out years. By out years, I'm talking about after the development of this, so essentially three years from now. Moving on to the next slide. We gave a lot of airtime on the defense side there, that's why it's on the left circle. We are not a defense-only company. We do a lot of business in the industrial IoT space, industrial customers. We are starting to get interest for AI and ML use cases using FPGA technology.
AI and ML is part of the SensiML market strategy because they're an AI software platform, but from an FPGA technology perspective, there are a lot of people using FPGA technology in the data center, where you do a lot of the training for AI. We're looking now at cases and with customers on how that could be sort of pulled, that inferencing can be pulled out to the edge, and how FPGA technology might be really beneficial to implementing that with low power or low cost from a system perspective. Also, historically, we've been used a lot for security of applications, and then to some extent, consumer IoT, both for our SoCs and our embedded FPGA IP.
It is worth noting on this slide that, in addition to this just being a large served market, I think we have good first mover ability here, because we're the first programmable logic company to really embrace open source tools and how they can be brought into what has historically been a proprietary-only tool flow. We are using open source components today in how we do our chip design, not exclusively, but some open source components. We are also embracing open source tools, that users would use to actually target our devices. I'm really proud of the company being the first company to do that and maintain this contribution to that open source ecosystem, so that the tools can continue to get better and better, for the users now and in the future.
As far as the expanding customer base and ecosystem go, logos on this slide represent people that are either a customer or an ecosystem partner. As far as ecosystem partner goes, they could be a partner of QuickLogic to further the FPGA capability that we've talked about, or they could be a partner of SensiML to help go to market with AI and ML software into the user base. So you can just go to this slide offline, but a lot of well-respected companies here.
Again, I'll just highlight for you that the fact that we do business with all of the top five and eight of the top 10 DoD prime contractors is a big plus, as we look to expand, A, the contract that I've already talked about on the DoD side, and then other, other folks in that area as well. This slide is similar to what we've had in the past. Again, given where we are in the quarter, the fact we, we announced on Monday, we haven't updated anything on here financially yet. That'll come after the call. As of the May call, $16 million in new contracts over the last couple of years.
All of that, all of those wins are directly resulting from this fact that we have more automation now, from our way of doing design with Australis and our embedded FPGA IP, that has led to the $125 million pipeline size, of which, you know, there's a healthy chunk of defense business, of industrial, of, of consumer, and it really spans the gamut in terms of products that we would be selling. Either the services in the beginning leading to license, or the services in the beginning leading to storefront, or in some cases, just using standard products that we already have in existence, like our, our SoC products or discrete FPGA.
Again, I think it's nice that it's a very diverse funnel, and the focus of the company is really to move things from pipeline to contracts, which should be showing up on that, that top, horizontal bar there. Again, we'll be updating more on that on Monday during the earnings call. We've talked a lot about the IP generator, and for those curious about how we do that and how it's different, we have this slide in our deck. For reference, the way that almost every FPGA company, including QuickLogic, has always done our chip or IP design, is what I'll call the old way. The old way is very manual. You know, you have architects, and you have lots of engineers, and you essentially hand draw everything to create the intellectual property in the chip.
Because it's manual, it can take well over a year to take our IP and port it to a different process or a different foundry. An example of that would be, you know, if we have our IP running at TSMC 40 nanometer, as an example, and we wanna move it over to GlobalFoundries 40 nanometer, or we wanna move it to Samsung 28 nanometer, or SkyWater 90 nanometer, any of those moves would generally take more than a year, in some cases, like a year and a half. Not only is that a long time, but it means that our entire team is allocated to that for that longer time. That means they can't address any other new opportunities that come in that may have a shorter needs window.
A lot of the old way of doing things just was not working for us from a business model point of view. You know, we, we understood that that was a limitation, and we said, "We need to fix this. If we wanna be a real IP company, we need to have more portability, more agility in the way we do design." Through our, our, our voyage of discovering this and how we could do things differently, we knew that automation had to be part of that. We came across some work that DARPA was funding at The University of Utah called OpenFPGA. That is a workflow capability that does bring a lot of automation to some of the manual aspects of the chip design and IP design.
In our view, it was not sufficient for commercial use. It was really good and effective for creating things that you could write research papers about, but not really good enough for commercial use. We started to assess, you know, if it has automation and it can really shrink time, then we have all this 30 years of expertise in FPGAs and IPs and, you know, our devices being reliable and robust and shipping all over the place in different critical use cases. If we could marry those two together from a capabilities perspective, maybe that's, maybe that's the key that we need.
We did diligence on this for about a year, being very, I think, thoughtful in how we analyzed it, and we came to the conclusion, in fact, yes, we should do this because this, this will open up a capability that we, A, we need for the market, and B, we just didn't have. C, we would still be able to protect our knowledge and crown jewels about how you really commercialize FPGA technology, but yet benefiting, excuse me, from open source automation. The results of all that is the Australis Workflow, that has been created and what we've been using now for two years to, to do all these different designs. I think if you look at the characteristics of the IP deals we've closed and executed on for revenue, ...
for the most part, they've all been on different foundry process combinations. We went from being able to do one in a year or a year and a half, to doing, like, more than four in one year, and they're all different. You can just see the type of scale that this gives us as a company to go execute. I think it's fantastic that we did this and took the risk to be the first company to embrace open source components and do this. Kudos to our engineering team for really, you know, bringing this in-house and having an open mind about how this could really revolutionize the way we do design. Lastly, I think, you know, a lot of people will say, "Are you crazy? Are you- you're adopting open source.
Are you open sourcing all of your information? The answer to that is no. We're, we're benefiting from open source. We are contributing to the open source tool chains because, you know, it's really expected that if you're benefiting from it, you're contributing some things back to the community. We are in no way contributing, like, crown jewel information back to that. It's all, you know, carefully separated by our teams internally for what's open source and what's not, so that we have this nice competitive moat that we can maintain for the years to come as we, we execute on our vision. Enough about the IP generator, but I think it was worth double-clicking into the details on that. For SensiML, just moving away from the FPGAs for a second and going to the SensiML AI software.
The SensiML acquisition was 2019. Initially, the strategy was, let's have as many users as possible of SensiML, and let's measure success by how many users we can have. I think that, you know, that works for things like social media, but it's not really for business-to-business applications. We did a sort of a modification, if you will, of the go-to-market strategy about 9 months ago. That modification of the strategy was, we don't want every cat and dog and everybody in their garage using SensiML. What we wanna focus in on are people that design real systems and ship real volume. The fastest avenue to that audience is not through our own organic sales force necessarily, but through people that have microcontrollers designed already into, you know, tens of thousands, hundreds of thousands of systems.
The obvious partner that we would go target from an ecosystem perspective are the microcontroller companies, because they in fact are in that big install base already. The nice thing about an install base is that if you're selling software, you can sell software on top of existing hardware in the field. You can also sell it for new designs that people are designing those microcontrollers into. That was the focus of SensiML. We have talked about a semiconductor company, a microcontroller company, that is going to private label thSensiML toolkit as part of their toolkit.
Again, stay tuned for Monday when we give an update on this, that's, I think, the, the right way to go to the market with this type of capability, because that microcontroller company has the install base, they have the large sales force that we simply don't have as a, as a company of our size. I think once they launch that, they start getting that out into the market, that's gonna be a really nice way of generating incoming leads that are qualified, turning them into SaaS agreements, and then also repeating that same strategy with other microcontroller companies. What we're doing with this one is not, is not exclusive.
We're hoping to rinse and repeat that with other folks in that, and really, start to establish a name for SensiML within the realm of edge AI, ML software. Lastly, again, this is a sort of a reduced summary slide, because we report on Monday. Again, the sales funnel, as of May, $125 million had been growing from the previous earnings calls. We like that because, again, it's very diverse in terms of product category, market segmentation, the strategic radiation-hardened opportunity with the options totaling 72. By far the largest contract in QuickLogic's history.
The $6.9 million in the beginning for that first phase, awaiting now the second phase to be awarded for that, and then the leading semiconductor company with the private brand. I think as soon as that's out and people are aware of who that is, that's gonna also be some... a lot of people will know who that company is, and they will be encouraged by what that could be, again, from a qualified lead and SaaS and in the future, royalty revenue. I think very well positioned, like where we are, and I would encourage everybody on this call to join us on Monday for the earnings call, where we can give further financial results and projections and more qualitative commentary that we're not allowed to today.
So with that, I will say thank you, and turn it back to you, Martin.
Thank you, Brian. You spent quite a bit of time talking about SensiML, as well as how you utilize open source tools to automate the design process. When you look at high-level industry exposure for the company in the longer term, how should we think about your background in consumer electronics and now this much expanded exposure to Defense, aerospace, and industrials? What would be the long-term ideal mix for you by verticals?
Good question. So I, I'll answer it in this way. The, if you go back, like, 5 years, we were heavily concentrated in consumer.... to the point where I think it was not healthy. Just too, too concentrated in customers that can change on a whim. We're very purposeful, I think, in, in bringing that back from that concentration point, again, and focusing a lot on markets that have historically used a lot of programmable logic, so defense, industrial. I think our ideal mix as a company, and I'm talking about, you know, a few years down the road from now, would be 30/70, where 30% is defense, 70% is non-defense. We're not there today. Today, obviously, we have a lot of things going on in defense, we're tilted more in that direction.
I would still rather be tilted more in that direction than too much consumer, by the way. We're, you know, more than 50% defense. Gradually, I think over time, as we get more of these designs that close into license, and royalty, and storefront for non-defense applications, we'll start to see that balance get back to where we'd like it to be. Again, 30/70, 30% defense, 70% non-defense.
Thanks, Brian. My next question is about edge intelligence, especially, you know, implementing machine learning models on edge devices. SensiML really provided a very easy way for maybe people without the expertise to implement machine learning models. You know, given the recent, I would say, hype over AI and large language models, do you see a expanded role for SensiML, or is there any other different angles for you to play the role, implementing AI on the edge?
Yeah, that's a, a good question, Martin, because, you know, I think with all the press that people are reading about large language models and ChatGPT-
Yes
... when they see that we have AI, immediately they start asking, "Do you have something like that?" We're not focusing on that area specifically. I think that there are areas where SensiML can play relative to generative AI, which is essentially what ChatGPT and those others, when they're getting all the press, are about. There's areas that we can explore in that, leveraging SensiML capabilities today. I think where the sort of rubber hits the road with engineers doing AI development for the edge, specifically, you know, the, the workflow process, without getting too technical, is that, A, you have to have lots of sensor data, right? AI is only as good as the training data that you give it. You need lots of data. You need clean data, so you're not injecting bad data into a decision-making process.
Then you need to build and test your AI models to infer that result, right? I saw a chart recently that most of the time spent generating an AI model to deploy in real life is actually not on the AI model creation, it's on the data acquisition and cleansing of that data, labeling of that data, so that you have a robust set to work with. And I think that's interesting because one of the sort of elements of the SensiML toolkit that doesn't get a lot of airtime is the Data Capture Lab. Which, in fact, is all about acquisition of data and cleansing of the data and labeling of the data.
I think as more people get in the realm of AI, and they want to try it because they see all the stuff that it could be done with, large language models, SensiML's gonna have an opportunity there, because of that component of their toolkit that actually helps, I would say, expedite and make less error-prone, one of the biggest, energy consumers in that whole process, which is data acquisition and, and cleansing and labeling. Again, just back to the earlier point, I think there are things around generative AI that SensiML can probably get into. We won't go over it right now, 'cause I think it's fairly early, but, you know, certain things in there I think could be done in talking with the team about that.
Thanks, Brian. I want to talk about the financial model for the longer term. You laid out the different business models and how the revenues will layer on, given their different margin level, margin profiles. Can you maybe talk maybe one step further to the operating profit line, how each business model flow to the operating margin line, and how that affects your longer-term, free cash flow generation?
Elias, do you want to take that question?
Yes, if I understand your question, Martin, it's how the cash flow line gets affected by these four businesses?
That's right.
All right, services, I'll, I'll just touch on the highlights, okay? Services revenue usually is, is a lower margin business, right? Licensing revenue is a very high margin business. Royalty fees will be superb margins because there's no cost associated with it. Storefront finished goods will be basically in bulks, meaning that the more quantity you sell, the better it is. I suspect, most of these, other than a few, I believe licensing revenue we're having, and, and royalty will kick in, probably in two years or so, in terms of high volume. Most of them really are very de minimis right now in terms of, overall revenue.
Okay, they're not there-- we're not there yet, let's put it this way, in terms of when they start hitting our, you know, bottom line. Okay?
Got it. With that, we're about time. QuickLogic, will be hosting meetings at our OpCo conference, and they will be reporting their June quarter results on Monday next week. Stay tuned. Thanks again, Brian. Thank you, for participating at our conference. Hope you have a good rest of the day.
Thanks, Martin. Appreciate it.
Thank you.
Thank you.