Great, good morning, everybody. My name is John Vinh. I cover semis here at KeyBanc Capital Markets, and we are pleased to have Jim Anderson, CEO, and Sherri Luther, CFO of Lattice Semi. Welcome, guys!
Yeah, thank you.
Thanks.
Thanks for having us.
Hey, Jim, last week you reported strong Q2 results. I was wondering if you could just highlight what the growth drivers in the quarter were.
Yeah, sure. We feel pretty good about certainly the growth in the first half. We saw about 20% growth in the first half this year versus prior year. You know, you can break that down a couple of different ways. You can look at it from a sort of market, customer perspective, or from a product perspective. If you, if you start from kind of a market perspective, one of the things we feel really good about is how we've repositioned the company in what we believe are the, the right long-term end markets for the company. Over 90% of our revenue now comes from communications, computing, industrial, and automotive. When we look at those four markets, we believe they all have long-term secular growth trends underneath them that will drive semiconductor consumption in general, but FPGA consumption in particular.
We feel really good about where we're positioned from a market standpoint, which wasn't necessarily the case when we started four or five years ago. Great markets to be positioned in. From a product standpoint, that's the other way you can look at it. Look, the company's going through the biggest product portfolio expansion we've ever done in the company's history, John, we have so many different new product cycles now, and that's really helping drive the growth as well. We've got just in our, in our small FPGA platform, we've got five new versions of Nexus that are ramping, that are in production and ramping. Another one that's, that'll start ramping next year. Our newest mid-range FPGA platform, that product ramp, that revenue ramp is still ahead of us.
We feel really well positioned, both from a market and a product perspective as well.
Great, thanks for the.
Sure
... the overview. Maybe talking, touching base on servers, you talked about seeing strength in the server market.
Yeah.
You know, many of your peers have talked about seeing kind of a crowding out effect in the server market from the expansion of AI servers, where there's obviously a much higher ASP. You know, a lot of the cloud providers right now are kind of confined in terms of their CapEx budgets this year. What are you guys seeing from a server's perspective between traditional AI and then maybe if you could also touch base on just how your content differs on AI server versus traditional?
Yeah. In general, for servers has been a great growth area for us. Actually, if you look at the last four years, we've grown double digits in communications and computing segment for us for four years in a row. A lot of that growth has been driven by servers. We believe we continue to be really well positioned, whether it's a general- purpose server or an AI-optimized server, we think we're really well positioned to continue to grow. If you look at what's driven the growth over the last three to four years, actually, for us, it's primarily been about content expansion. With each new generation of server, our dollars of content per server has gone up with each new generation, and that's been the biggest driver of our revenue.
Not so much the underlying server units, that would ship in totality for the industry, but the fact that our content with each generation continues to go up. What's been driving that is higher attach rates, so more Lattice chips used per server. Also, we've been continuing to bring devices that have higher capability, newer features, which has driven our ASP up over time. That combination of attach rate and ASP has helped drive our dollars of content up. On the newest generation of server that's starting to ramp this year, we see about a 50% increase in the dollars of content per server in that new generation versus prior generation. We, we feel really good about the continued kind of growth opportunity for us in servers in general. Then you asked about AI in particular.
Certainly, there's now a, you know, a subset of servers that are more AI-optimized, have high GPU content. When we look at those systems, depending on the configuration of that system, we have anywhere from about the same level of content as a general- purpose server or higher content. For us, that's a, it's a net positive, but it's either equal to or greater content in those GPU, more GPU-optimized servers.
Great.
Yep.
You mentioned, you know, the TAM expansion. If you look at the.
Yeah
... TAM assumptions, for the small FPGA market, it's roughly about $4.5 billion. You know, if you go off of your target model, it says, you know, potentially you've got a third of that market in a few year, years. What's your sense of, you know, who makes up the other 2/3 of that market today? Is this mostly Xilinx? Then, how do we think about your ability to monetize the rest of that TAM going forward?
Yeah, the most common competitors for us in our traditional market, and this would apply to mid-range as well, so either small FPGA, which, I think you were referring to small FPGA segment of FPGAs, but it would apply to mid-range as well. Most common competitors we would face are Altera and Xilinx, which are now obviously owned by Intel and AMD. If you look at in the small FPGA space relative to those two competitors, you know, and our customers would say the same thing if you ask them. If you look at our product portfolio today, our customers would tell you Lattice has the strongest product portfolio that it's ever had in its, in its entire history, and we've been around for about 40 years, right? Why is that?
It's because, you know, four or five years ago, we started really reinvesting in the competitiveness of our product line. If you look at our newest product, Lattice Nexus, we would have, relative to competition, anywhere from 2x-3x better power efficiency, for example. Higher performance and not just 10% or 20% better power efficiency, bu 2x-3x , and that's a big deal to our customers because usually the primary design constraint of our customers is around total system power. We're bringing out a product that's, you know, you know, way more competitive than our competition, and so that's why you're seeing share gain, that's why you're seeing growth of Lattice in that small FPGA segment.
Then the next step is what we announced back in December of last year, is we're now bringing that expertise, our expertise around power efficiency, small physical device size to now the mid-range FPGA market, which doubles our addressable market and provides a whole new revenue stream for the company.
Yep, just, you know, how should we think about your ability to monetize the remaining 2/3 of that small FPGA?
I think you should think about it as good.
Yeah.
Yeah.
Okay.
No, I mean, seriously, yeah, I think with that big of a competitive advantage and the amount of software content that we provide, I think that we have a long runway of good growth opportunity in front of us. Actually, the best way to think about it is relative to our revenue base today, if you look at size of market relative to revenue base today, we could grow very strong for years without hitting any structural limit in terms of market share or size of market. I think that's probably the right way to think about it.
Yeah, nice. When you think about your share gains, how much is this coming from taking sockets from your existing competitors versus maybe new use cases and applications?
Yeah, it... That's a good question, and it's a combination, actually, three. I'll, I'll throw in a third category, too. You know, there's kind of three sources of the revenue growth in that market. One is kind of what we were just talking about, good old-fashioned share gain, where we're displacing an Altera, Xilinx, FPGA, and that's probably the biggest contributor in terms of revenue growth. The other one is the one that you just mentioned, where, you know, we're working with customers on new, basically new capabilities, new features that they've never had in their system before, and they're using our FPGA with the software that we're providing to bring, to get, design that into their system to bring new capability that didn't exist before. Then, to us, we call that TAM expansion because we're effectively expanding the TAM, creating new TAM for ourselves.
Then the third category is what we call TAM conversion, which is you know, a customer maybe in the past, they used a different type of silicon, like a microcontroller, and we're showing them how if they switch to a Lattice FPGA, again, with our software content, they can see a big performance per watt benefit. We view that as TAM conversion, where we're converting other type of silicon TAM to our FPGA TAM. I would say our revenue growth has been driven by all three of those categories, probably most of it from share.
Great, that makes, that makes a lot of sense.
Yep.
Maybe we can talk about kind of FPGAs versus MCUs. I think in your Investor Day presentation, you kind of laid out kind of the performance and the metrics of FPGAs versus MCUs in industrial and auto applications. I think you said 35 x faster in terms of AI inferencing, 2 x in terms of motor control. What other hurdles are there for customers for fully adopting FPGAs over MCUs?
I would love to tell you that all MCUs will convert to FPGA. That's not the case. I want to be clear about that, right? You will see a subset of applications where, yeah, the, the, the customer can switch to an FPGA and see a much better performance per watt, performance per dollar, whatever metric that you want to use, a really significant advantage by switching to an FPGA, and those are some of the use cases that we're focused on.
I'll give you one example, which is common, which is a common example that we see in the industrial space, where, let's say, a customer has historically used within their system a microcontroller, and now let's say they're trying to add more decision-making capability, more intelligence to their decision- to their system, to allow that system to be more adaptable to its environment, et cetera. What they're talking about then is adding some level of artificial intelligence processing, this is inference processing, right? To allow that machine to make localized decisions without a direct connection back to the data center. When they, when they're trying to make that transition, what they're finding is, you know, hey, a microcontroller is a sequential processor. AI algorithms are inherently parallel algorithms. They don't map well onto a microcontroller.
What we're showing them is, those AI algorithms do map really well onto a Lattice FPGA, and we've given them a software stack that makes it really easy for them to take their unique AI algorithm and map that onto a Lattice FPGA, and basically turn that Lattice FPGA into a customized AI processor for their unique inference algorithm. They know that that algorithm is gonna evolve and change over time, that's not a problem with a Lattice FPGA. You just reprogram the FPGA as you change your algorithm, so it allows them to future-proof the system to some extent as well. So that's one of the common use cases where we're seeing customers switch over to a Lattice FPGA, is that kinda inference, more AI at the edge, and industrial application.
We're seeing that in other, other markets as well, but that would be a common place where we're doing TAM, as I talked about earlier, TAM conversion.
Nice. Speaking of inferencing, as you know, generative AI has been a huge topic of, of interest for investors. You've mentioned...
Yep
AI as a long-term growth driver there. Can you just expound upon that a little bit more, and how you can benefit from increasing interest in AI?
AI in general? Yeah, the two areas would be, number one, kind of the area we were just touching on, which call that Edge AI, AI in all sorts of Industrial IoT or IoT applications in general. That example that I just gave is a great example in the industrial, would apply in an automotive context, in other markets as well. Certainly, we see good opportunity for continued growth in those Edge AI, processing applications, and we've already seen great growth in that. We expect good growth moving forward. Then the second category is, kind of back to the server space. In generative AI, or just call it AI machine learning workloads in general that are being done in the data center, that pushes a tremendous amount of compute cycles into the data center, right?
Data center infrastructure will have to be built out and expanded to, to accommodate that, and that's certainly AI-optimized servers, you know, heavy GPU content, but also even general- purpose AI or general- purpose servers, and all of the networking equipment associated with that, right? Data center networking equipment has to feed all of that, those processing elements. I think AI in general, generative AI is one example, will continue to drive that data center growth. Now, the way we benefit from that is Lattice chips are used for control management and security of server platforms, right? Both general purpose as well as those AI-optimized server platforms. As I said before, in those AI-optimized server platforms, we have equal to or greater than greater content in those versus general purpose.
We think it's a net benefit, certainly a benefit for the industry and net benefit for Lattice as well.
Great.
Yep.
Any questions? Great. Hey, Sherri, maybe a question for you. Software attach rate right now is over 50%. Where can that go in a few years, and, you know, how critical is it for customers in terms of design engagements, and what types of premiums and ASPs and gross margins does it bring to your, to your business?
Yeah, so, so one of the elements of our gross margin expansion strategy that we talked about at our Investor Day earlier this year is, you know, pricing optimization, and part of that is, is higher ASPs on software attach. New products also add value there. From a software attach perspective, we have seen that, you know, ASPs generally are higher when they include our software solution stacks with it. You know, when we look at our, our new gross margin target that we put out at the low 70s, you know, that is certainly an element of it. Software attach, as well as, you know, pricing optimization, which is something that we've been looking at now for into our fifth year of gross margin expansion strategy.
Product mix is also a contributor, and certainly, product cost efficiencies, with respect to yields and, and cost efficiencies in that area. All of those items together really help drive our gross margin expansion strategy.
Yeah. Maybe a follow-up to that. You've laid out kind of a gross margin target in your low 70s. Pretty much already there, right? Software attach rate sounds like it's gonna continue to go up. You're gonna ramp higher ASP products. Can we expect that gross margins will kind of continue to trend higher from here? Then potentially, what are the other offsetting factors to some of these tailwinds that you're seeing?
Yeah, so, you know, the target that we just put out in May, just a few months back, the low 70s, we're certainly excited about that target because we believe our gross margin is durable. We've since over the past five years or into the fifth year of executing on that strategy, we've increased our gross margin by just under 1,400 basis points. From that, you can see that there are a lot of elements to our strategy, and, you know, some of them kick in at various points in time. Certainly, new products, as I mentioned, certainly software attach helps drive that. But we believe our gross margin is durable.
You know, with the focus on, you know, greater capacity, greater capability that our customers want from our products, that's also a contributor to our gross margin expansion strategy. We believe it's durable. You know, we'll continue to execute on that strategy and, you know, love to, love to raise the bar at the appropriate time, but we feel really, really good about our low 70s target at this point.
Yeah.
Have you guys seen any pricing dynamics change with Xilinx and Altera now being under the big, ugly companies that you're competing with?
I like your characterization. Yeah, we believe, we've seen very good stability in our pricing, right? I would say that's not just a short-term comment, but I would say that's over a multiyear basis. As Sherri mentioned, in the beginning of 2019, we put in place a gross margin expansion strategy. Part of that strategy was focused on pricing, what we call pricing optimization, our overall pricing strategy. We, as a company, became much better about how we price our products. If I look back over the... Now we're in our fifth year, right?
Yeah, yeah.
If I look back over the last five years, I see a lot of stability in our pricing, right? We would expect that to continue.
Great. Maybe we can touch base on Avant. Sounds like you've had really great success in terms of ramping Avant. I think some of the metrics you've talked about is 100+ customer engagements. 90% of them are from existing Nexus customers. I'm wondering if you could just talk about what progress you're making in terms of engaging kind of non-Nexus customers. How are things going on, on that front?
Well, I think you're probably not looking at it the right way, John, to be totally honest, is if you look at the, the target customers for Avant, right? 90% of those target customers are already Lattice customers today. The, the other 10%, I, I'm not so worried about the other 10%, right? I want to sell, you know, those 90% of customers that are already customers of Lattice today. What I want to do is sell them now an Avant product, right? To them, you know, they, they know us, they trust us, they've known us for 40 years.
We've done a good job of supporting them with small FPGAs, and now what we're offering them is, "Hey, we now have a mid-range FPGA." Same great power efficiency, all the great, you know, performance, features, capabilities that you've, you've trusted Lattice to deliver for small FPGAs, now in higher capacity, higher capability devices. I think our focus is those 90%, the 90% of the SAM or TAM that's with existing customers, where we don't have to go penetrate a new customers. These, these are customers that know us already, have known us for many years. There, we're just convincing them to use a new product line. One of the things that helps us is not just our history with those customers, but, the software, right? When they look at, well, what software would I need to use with Avant?
They can leverage all the same software that they've been using on our Nexus devices, actually, even our older devices or pre-Nexus devices. That same software they're already using is leverageable onto Avant, so they don't have to go, you know, figure out how to use a whole bunch of new software. That's our biggest opportunity is, is selling now Avant to our existing customer base, right? That's most of the doubling of our addressable market right there.
Great. How does the software attach rates on these Avant wins compare to Nexus? Are they comparable? Are they-
Very comparable, yeah. Very- which is what we would expect. We would expect a similar level of attach rate, because the attach rate has more to do with the particular customer and application example, right? Yeah, so far, we've seen very comparable attach rates.
Okay. What are you seeing from a competitive perspective? You know, I would imagine in low-end, small FPGAs, there's not as many competitors there.
Nothing low-end, about 70% gross margin.
Obviously, the other two big incumbents have, you know, broader offerings in mid-range.
Yeah, I, I think first of all, we take all competitors very seriously and are two traditional competitors. We, we always build our roadmap and our product portfolio, assuming there will be robust competition. We've always done it from day one of when we started. We always assumed there would be robust competition. That way, if the competitors do show up in new devices, we're ready, and if they don't, then great, that's upside for us, right? With that in mind, you know, certainly, we, we believe we have a very competitive offering. I think our customers would tell you the exact same thing.
When we launched the Avant devices back in December, that platform, and we, we showed the first Avant device, we showed measured silicon capability versus competition, and we showed, you know, up to 2.5x better power efficiency, up to 2x as much performance, and then a physical size of the product that's up to 6x better. We showed significant competitive advantages. We believe we're really well-positioned. We certainly don't take anything for granted. We're always paranoid about the competition, but I think, you know, based on what our customers would tell us as well, we feel we have a really good competitive position in both small FPGA and mid-range, in the mid-range FPGA respects.
Great. I know there's been a lot of interest on Avant, but want to also just to touch base on Nexus.
Yeah.
How's, how's that progressing? Can you give us an update in terms of how, you know, design traction is progressing on that front?
Yeah, very good. You know, Nexus has continued to ramp really well. If you look at the growth last year or the growth in the 1st half of this year, Nexus has grown faster overall than our overall company growth, which is what you would expect from a new platform. We're really pleased with the continued growth of Nexus. The great thing about Nexus as a platform is we believe Nexus will continue to ramp for at least the next three years, right? We have five different device families that are already in production. The 6th has been launched, goes into production first half of next year. We've got more Nexus devices that we're gonna be launching in future quarters. There's a steady stream of new Nexus devices, each of which create a new revenue stream.
We would expect that platform to, as I said, to continue to ramp for at least another three years.
Great. You know, going back to servers, I think recently a BMC supplier in the server space has talked about plans to go after server security and kind of I/O expander applications, where I think you have incumbency there. I know you don't take competition lightly, can you talk about confidence that you have in terms of being able to hold share in these areas and, and why?
We feel really good about just our ability to continue to expand the dollars of content per server. We, you know, whether that's a traditional FPGA competitor or competitor from a different type of silicon, right? Part of it is what's happening in the server space, right? If you think about what's happened over the past years and what's happening moving forward is, you know, there's a lot more differentiation within the servers with heterogeneous computing, many different types of CPU architectures, now graphics being used for, you know, GPUs being used for machine learning, generative AI. There's a lot more complexity in the server offerings from our customers, and they're also all of our customers, whether they're hyperscale data center customers or traditional enterprise customers.
Look, they're trying to figure out: how do I make sure I continue to differentiate in this complex world, right? And that's where we can help them because we can use a Lattice, a standard product Lattice FPGA, combined with their particular algorithm or differentiation, and we can show them how, hey, you can customize-- you can program that Lattice FPGA to implement custom features that are differentiated and unique to you using a standard product from Lattice. And I think that's-- that value proposition of being able to help them differentiate across all of their platforms, while they leverage a standard product from Lattice, I think that, that's worked really well over the past years, and I think that's gonna continue to work well as the complexity of offerings from them continues to expand.
Just as, you know, the, the number of different SKUs they have to manage and their, you know, their sense of urgency around continuing to be able to differentiate themselves in the marketplace.
Nice. Last question for Sherri. I noticed at the end of Q2, you're debt-free.
Mm-hmm.
Has that changed the way you approach your capital structure going forward?
Well, we're certainly pleased that subsequent to the end of last quarter, we did pay off the remaining outstanding debt on our balance sheet. This is the strongest balance sheet that the company's had in at least a decade. That's as far back as I went, certainly a very strong balance sheet. When we look at it from a capital allocation perspective, you know, our top priority still remains the organic growth of the business and investing in our long-term product portfolio, investing in demand creation. You certainly see that in our R&D spend as we continue to invest in new products and launch new products. That's our number one priority. Health of the balance sheet, I mentioned already.
Definitely want to make sure that we focus-- continue to focus on working capital, strong cash, generation, which we had 35% in Q2, which I'm also very proud of. Returning of capital to shareholders through our share repurchase program is also something that we've been executing now on 11 consecutive quarters. Q2, we bought back $10 million worth of stock. To date, since we started that program, we've actually reduced dilution by about 3.8 million shares, so we're pleased with the progress there. We do have another 110 million outstanding on the board authorization, $110 million, that expires at the end of this year. We continue to look at all of those things on a quarterly basis to determine the best use of our cash.
Great. With that, looks like we're out of time. Thank you, guys.
Yeah. Thanks, John.